1
00:00:00,000 --> 00:00:07,600
No, your usual line is... Hello guys, how are you doing?

2
00:00:07,600 --> 00:00:09,100
Hello guys, how are you doing?

3
00:00:09,100 --> 00:00:10,100
Yeah, there you go.

4
00:00:10,100 --> 00:00:11,100
Oh, what?

5
00:00:11,100 --> 00:00:12,100
Oh, good.

6
00:00:12,100 --> 00:00:16,800
Alright, I'll have to put it like this, because my phone connector is broken.

7
00:00:16,800 --> 00:00:19,740
I think it's my phone, actually, it's broken.

8
00:00:19,740 --> 00:00:23,920
And if it wiggles, then it disconnects, and I have to get into the things to recover

9
00:00:23,920 --> 00:00:24,920
the audio.

10
00:00:24,920 --> 00:00:29,760
And I don't want to do that, because it's the full point of the setup is for it to

11
00:00:29,760 --> 00:00:32,360
be fully automated, in a sense.

12
00:00:32,360 --> 00:00:34,960
So, we should be fine.

13
00:00:34,960 --> 00:00:36,360
Oh, wrong.

14
00:00:36,360 --> 00:00:37,360
Here.

15
00:00:37,360 --> 00:00:38,360
Look, first mistake.

16
00:00:38,360 --> 00:00:39,360
Okay.

17
00:00:39,360 --> 00:00:40,360
Okay.

18
00:00:40,360 --> 00:00:42,360
How have you been doing, Jesus?

19
00:00:42,360 --> 00:00:43,360
Pretty good.

20
00:00:43,360 --> 00:00:45,360
Here in the island, chilling.

21
00:00:45,360 --> 00:00:46,360
Yeah.

22
00:00:46,360 --> 00:00:50,360
The cohort ended and you told me, I'm going to be here for a couple of months.

23
00:00:50,360 --> 00:00:51,360
Yeah.

24
00:00:51,360 --> 00:00:53,360
And I sit here.

25
00:00:53,360 --> 00:00:58,160
So, I was like, okay, I'm going to grab you one day, and we'll go in a nice walk,

26
00:00:58,160 --> 00:01:04,120
and you're going to tell me everything about trusted execution environments and all

27
00:01:04,120 --> 00:01:05,120
those kind of things.

28
00:01:05,120 --> 00:01:06,120
So, here we are.

29
00:01:06,120 --> 00:01:07,120
Yeah.

30
00:01:07,120 --> 00:01:08,120
So, what have you been working on?

31
00:01:08,120 --> 00:01:14,000
You gave an awesome presentation during the SEC-08 about the evolution of confidential

32
00:01:14,000 --> 00:01:15,000
compute.

33
00:01:15,000 --> 00:01:16,000
Yeah.

34
00:01:16,000 --> 00:01:18,000
Orcrics to Ardias.

35
00:01:18,000 --> 00:01:23,440
He was the one that created that presentation, that is slides.

36
00:01:23,440 --> 00:01:27,680
But yeah, I'm working with him on in Enclava.

37
00:01:27,680 --> 00:01:32,240
Enclava is a product around the TEEs.

38
00:01:32,240 --> 00:01:38,360
It's also an open platform, because it is a part of it that is open source, that if you

39
00:01:38,360 --> 00:01:45,680
have some compatible hardware, because TEEs at the end is all attached to the hardware,

40
00:01:45,680 --> 00:01:54,920
they are not such things as T on consumer hardware that you need specific chips.

41
00:01:54,920 --> 00:01:59,520
Maybe explain why that is, like, why do you need special chips and special manufacturers

42
00:01:59,520 --> 00:02:00,520
to make this work?

43
00:02:00,520 --> 00:02:09,840
It's basically because the vendor basically engraves some keys in the silicon.

44
00:02:09,840 --> 00:02:15,440
So, the private keys of the vendor are put directly in the silicon on the chip?

45
00:02:15,440 --> 00:02:16,440
Yeah.

46
00:02:16,440 --> 00:02:18,160
And that's how you do the verification, right?

47
00:02:18,160 --> 00:02:19,160
Exactly.

48
00:02:19,160 --> 00:02:26,080
Like, you have some keys that are in the CPU, and those keys are used basically to encrypt

49
00:02:26,080 --> 00:02:29,400
the memory on the process on RAM.

50
00:02:29,400 --> 00:02:30,400
Yeah.

51
00:02:30,400 --> 00:02:36,560
And those keys are basically, I mean, the private keys are obviously not known, but the vendor

52
00:02:36,560 --> 00:02:37,560
can attest.

53
00:02:37,560 --> 00:02:40,560
It's just public private key cryptography, right?

54
00:02:40,560 --> 00:02:41,560
Yeah.

55
00:02:41,560 --> 00:02:44,760
Like, you shift the keys are known, exactly, exactly.

56
00:02:44,760 --> 00:02:49,200
So, every chip they produce, they just say, like, those are the keys, those are our chips

57
00:02:49,200 --> 00:02:50,200
and so on.

58
00:02:50,200 --> 00:02:51,200
Yeah.

59
00:02:51,200 --> 00:02:59,880
And then you challenge the vendor with the challenge, and give you back the results of

60
00:02:59,880 --> 00:03:06,400
the attestation or a report, and then you can verify, anyone can verify that the hardware

61
00:03:06,400 --> 00:03:07,400
is genuine.

62
00:03:07,400 --> 00:03:08,400
Yeah.

63
00:03:08,400 --> 00:03:11,080
So, that's the point of TEEs.

64
00:03:11,080 --> 00:03:14,840
You mentioned in your presentation kind of the evolution of the various chips and the

65
00:03:14,840 --> 00:03:17,520
various, yeah, and there's, I guess, sense on.

66
00:03:17,520 --> 00:03:19,000
So, talk me through that a little bit.

67
00:03:19,000 --> 00:03:22,840
We're in generation, I don't know, four or five now or something like that, like big.

68
00:03:22,840 --> 00:03:23,840
Yeah.

69
00:03:23,840 --> 00:03:26,320
If you take the big categories that you showed.

70
00:03:26,320 --> 00:03:27,320
Yeah.

71
00:03:27,320 --> 00:03:34,800
So, I think that, I mean, there's some prehistory of, of confidential compute, but so far

72
00:03:34,800 --> 00:03:45,000
the biggest mainstream one was in 2015, and it was Intel with SGX, if I can record

73
00:03:45,000 --> 00:03:46,000
properly.

74
00:03:46,000 --> 00:03:54,720
And, basically, that was the first iteration of TEEs, that went mainstream, and basically

75
00:03:54,720 --> 00:04:07,320
the idea there is that you can run software in confidential VMs, but they had SDK, and

76
00:04:07,320 --> 00:04:13,200
you have to basically make your software around the SDK.

77
00:04:13,200 --> 00:04:26,400
And then we have another iteration, that was, 2016, I'm not sure about this, probably

78
00:04:26,400 --> 00:04:28,400
with the time.

79
00:04:28,400 --> 00:04:37,880
But basically, we started to have the hyperb store as an adversarial.

80
00:04:37,880 --> 00:04:51,800
It was AMD with SEV, a line, and basically, you could run any container inside of it,

81
00:04:51,800 --> 00:04:57,480
instead of using an SDK to make your software around it.

82
00:04:57,480 --> 00:05:04,840
And that is where we are now, I mean, now we have a confidential compute also in graphic

83
00:05:04,840 --> 00:05:05,840
cards.

84
00:05:05,840 --> 00:05:06,840
Yeah.

85
00:05:06,840 --> 00:05:07,840
Yeah.

86
00:05:07,840 --> 00:05:13,760
That's kind of the main thing, right, because, I mean, we are living in the agentic age,

87
00:05:13,760 --> 00:05:20,240
I guess, and people are handing over all their data and their whole life, and their stool

88
00:05:20,240 --> 00:05:26,760
samples and their DNA to agents, and so they basically want to have certain assurances

89
00:05:26,760 --> 00:05:36,480
that, if you use a private provider and so on, if the data that is processed, stays confidential

90
00:05:36,480 --> 00:05:40,280
and no funny business is going on, the stuff that you execute is actually being executed,

91
00:05:40,280 --> 00:05:41,280
that's all.

92
00:05:41,280 --> 00:05:42,280
Exactly.

93
00:05:42,280 --> 00:05:52,680
We had now NVIDIA with some graphic cards that have this technology, so you can run confidential

94
00:05:52,680 --> 00:05:55,280
workflows inside of them.

95
00:05:55,280 --> 00:06:05,680
Like, before that, they were all used in the CPUs, so you can run a model in a CPU, right?

96
00:06:05,680 --> 00:06:06,680
Yeah.

97
00:06:06,680 --> 00:06:16,680
You can run anything, you know, that's the beauty about tiering machines, but, you know,

98
00:06:16,680 --> 00:06:25,880
like, if you want this thing to answer, while you're still alive, then you'll have to.

99
00:06:25,880 --> 00:06:32,880
But that's basically where we are right now, and I think that was a reason of T's to, you

100
00:06:32,880 --> 00:06:34,880
know, going mainstream.

101
00:06:34,880 --> 00:06:35,880
Yeah.

102
00:06:35,880 --> 00:06:42,880
Alias was, you mentioned him before, you know, like, he was talking about all this stuff

103
00:06:42,880 --> 00:06:46,080
as early as the C01 in the first cohort, remember?

104
00:06:46,080 --> 00:06:53,280
He was already kind of, yeah, aware of where the puck is going, so to speak, and he's been

105
00:06:53,280 --> 00:06:54,760
working on this ever since, right?

106
00:06:54,760 --> 00:07:03,240
Like, he basically, yeah, got enamored by the idea of confidential compute and also the

107
00:07:03,240 --> 00:07:11,840
things you could do, if you would have something like we are having now at a large scale,

108
00:07:11,840 --> 00:07:17,520
and what kind of things you can run, and I remember in the past, like, I mean, as a C01

109
00:07:17,520 --> 00:07:25,960
was basically where the idea of DBMs was born, and basically communicating over a roster,

110
00:07:25,960 --> 00:07:29,960
but having the compute obviously done and executed somewhere else, and how would it look

111
00:07:29,960 --> 00:07:35,440
if you would do all these execution in a trusted execution environment, all those kind

112
00:07:35,440 --> 00:07:36,440
of things.

113
00:07:36,440 --> 00:07:44,040
And here we are, you know, like, it's not much later, eight cohorts later, we can measure

114
00:07:44,040 --> 00:07:45,040
the time on cohorts.

115
00:07:45,040 --> 00:07:49,440
Yeah, no, it's only like, it's only been like two years or something, like two and a half

116
00:07:49,440 --> 00:07:55,760
years or something like that, it's not been a very long time, and now it's a very normal

117
00:07:55,760 --> 00:07:56,760
thing.

118
00:07:56,760 --> 00:08:02,680
We have actual private providers, you know, we have maple, of course, with others that

119
00:08:02,680 --> 00:08:10,240
offer this to a consumer market, basically, and talk to me more about the product that

120
00:08:10,240 --> 00:08:15,920
is in Enclava, what the idea is, you mentioned also the parts of it are open source, and

121
00:08:15,920 --> 00:08:19,760
that, you know, if you have the right hardware, you can self-host, like, talk to me about

122
00:08:19,760 --> 00:08:21,440
how this works and what the idea is.

123
00:08:22,080 --> 00:08:26,160
In Enclava is the product or the platform.

124
00:08:26,160 --> 00:08:30,360
Platform is composed by these two parts.

125
00:08:30,360 --> 00:08:38,600
One is called the cap, and the other is called the pass, or this name by Argus.

126
00:08:38,600 --> 00:08:45,640
The cap is like the contraplane, it's the open source parts, you can run it in proper

127
00:08:45,640 --> 00:08:52,840
hardware, it basically allows you to do all the management and the contraplane.

128
00:08:52,840 --> 00:08:57,920
So just that I understand it correctly, cap, like a taxi cap, and bus like a proper bus,

129
00:08:57,920 --> 00:09:06,080
the big one that you get in, bus with P, like a P, A, A, S, B, A, S, what does it stand

130
00:09:06,080 --> 00:09:07,080
for?

131
00:09:07,080 --> 00:09:15,600
The tradition of nerds being terrible at naming things.

132
00:09:16,600 --> 00:09:19,600
I'm working with him for already a couple of months.

133
00:09:19,600 --> 00:09:21,600
I still don't know what it means.

134
00:09:21,600 --> 00:09:29,600
I used to assume that it's, you know, the prototype is part of Enclava, because basically

135
00:09:29,600 --> 00:09:36,000
the cap is what allows you to do all the management of the hardware and so on.

136
00:09:36,000 --> 00:09:41,320
The pass is like a hostage service.

137
00:09:41,320 --> 00:09:46,760
So you have users and different stuff to basically blue tax fits.

138
00:09:46,760 --> 00:09:47,760
Okay, I see.

139
00:09:47,760 --> 00:09:48,760
I see.

140
00:09:48,760 --> 00:09:57,760
Maybe Alias himself can resolve the mystery behind the names and have to reach out to him.

141
00:09:57,760 --> 00:10:08,800
Okay, so you have the hostage service, and that means that if I would want to use confidential

142
00:10:08,800 --> 00:10:16,400
compute, I could sign up there and other people provide a compute.

143
00:10:16,400 --> 00:10:20,600
Is there a marketplace idea on that as well, similar to how the routes the people have

144
00:10:20,600 --> 00:10:21,600
it?

145
00:10:21,600 --> 00:10:22,600
I mean, maybe in the future.

146
00:10:22,600 --> 00:10:29,360
I don't know what are the plans from Alias, but so far, we are running this product.

147
00:10:29,360 --> 00:10:31,040
It's not out yet.

148
00:10:31,040 --> 00:10:33,480
We are very close to release it.

149
00:10:33,480 --> 00:10:44,200
It's basically the idea is that you don't need to be a T engineer to run confidential workloads.

150
00:10:44,200 --> 00:10:55,280
Like you can be a dev and you deploy a container in a confidential environment and you don't

151
00:10:55,280 --> 00:10:57,880
need to basically know everything and be an expert.

152
00:10:57,880 --> 00:10:58,880
Okay.

153
00:10:58,880 --> 00:11:01,880
So similar idea to what Docker did in terms of like pre-docker.

154
00:11:01,880 --> 00:11:04,480
All this stuff was very hard and not communicated very clearly.

155
00:11:04,480 --> 00:11:05,480
Exactly.

156
00:11:05,480 --> 00:11:06,480
Exactly.

157
00:11:06,480 --> 00:11:13,120
Like Heloku, like this kind of providers that if you are a nice service around a technical

158
00:11:13,120 --> 00:11:14,120
thing.

159
00:11:14,120 --> 00:11:15,120
Yeah.

160
00:11:15,120 --> 00:11:16,120
So that's it.

161
00:11:16,120 --> 00:11:17,120
That's basically it.

162
00:11:17,120 --> 00:11:18,120
Okay.

163
00:11:18,120 --> 00:11:19,120
And the CLI.

164
00:11:19,120 --> 00:11:23,520
And what are you working on specifically if I'm allowed to ask?

165
00:11:23,520 --> 00:11:24,520
Yeah, yeah.

166
00:11:24,520 --> 00:11:25,520
Of course.

167
00:11:25,520 --> 00:11:31,360
Basically, you know, like in this world, everything is confidential, so you can't be.

168
00:11:31,360 --> 00:11:32,360
You can't be too careful.

169
00:11:32,360 --> 00:11:35,840
Yeah, but we are not in a confidential environment right now.

170
00:11:35,840 --> 00:11:37,160
That's okay.

171
00:11:37,160 --> 00:11:41,600
So what are you working on on the cloud side?

172
00:11:41,600 --> 00:11:42,600
Like what's the?

173
00:11:42,600 --> 00:11:53,400
So basically I docked through in the platform, catching all the bags, fixing them.

174
00:11:53,400 --> 00:11:55,400
That's my role.

175
00:11:56,240 --> 00:12:04,480
Yeah, basically the difficulty of all this is that these are kind of very aromatic environments.

176
00:12:04,480 --> 00:12:07,160
So it's kind of like a black box, right?

177
00:12:07,160 --> 00:12:09,240
Like your hand over the workload.

178
00:12:09,240 --> 00:12:10,240
You don't?

179
00:12:10,240 --> 00:12:11,240
Yeah.

180
00:12:11,240 --> 00:12:20,040
I mean, you as an operator, you can see inside the host, sorry, the provider of the infrastructure,

181
00:12:20,040 --> 00:12:22,080
kind of look inside.

182
00:12:22,080 --> 00:12:23,080
Which is the whole point?

183
00:12:23,080 --> 00:12:24,080
Which is the whole point.

184
00:12:24,080 --> 00:12:25,080
Yeah.

185
00:12:25,440 --> 00:12:31,640
If you are the platform, kind of, it makes debugging those kind of things harder, right?

186
00:12:31,640 --> 00:12:32,640
Oh, yeah.

187
00:12:32,640 --> 00:12:40,560
And the time, the iteration time is also slower, because every time that's, for example,

188
00:12:40,560 --> 00:12:50,320
you fix a back, and you have to do a layout, a new version of the password example or something.

189
00:12:50,320 --> 00:12:55,440
You need to wait for all the layers to be attest and to have proper attestation.

190
00:12:55,440 --> 00:12:56,440
So yeah, I see.

191
00:12:56,440 --> 00:13:02,160
So you say it's slower, more difficult to the back, because there is no, sometimes there

192
00:13:02,160 --> 00:13:04,360
is no clear locks of what's going on.

193
00:13:04,360 --> 00:13:05,360
Yeah.

194
00:13:05,360 --> 00:13:06,360
Yeah.

195
00:13:06,360 --> 00:13:07,360
Yeah, sure.

196
00:13:07,360 --> 00:13:08,360
Yeah, sure.

197
00:13:08,360 --> 00:13:14,360
I mean, that's basically, you know, like the, the trade of you enter into, like you want

198
00:13:14,360 --> 00:13:15,360
confidentiality.

199
00:13:15,360 --> 00:13:16,360
Yeah.

200
00:13:16,360 --> 00:13:25,080
And you want all those attributes that T provides, and, yeah, well, you know, like there

201
00:13:25,080 --> 00:13:30,360
is, there is always a cost to pay, it's similar to, I don't know, going through an SSH

202
00:13:30,360 --> 00:13:35,960
tunnel or using a VPN or whatever, like you won't have the same bandwidth, and just

203
00:13:35,960 --> 00:13:41,840
on like, you know, unencrypted, unsecured communication channel in its own, or similar

204
00:13:41,840 --> 00:13:45,600
with chat, you know, like we talked about this in the past as well, a lot with your

205
00:13:45,600 --> 00:13:46,600
confidential.

206
00:13:46,600 --> 00:13:47,600
Yeah.

207
00:13:47,600 --> 00:13:48,600
We are very close.

208
00:13:48,600 --> 00:13:49,600
So we honest.

209
00:13:49,600 --> 00:13:53,600
Um, I have been used, uh, two weeks, it's going to be really, yeah, indeed.

210
00:13:53,600 --> 00:13:54,600
Yeah.

211
00:13:54,600 --> 00:14:00,120
Yeah, because this, uh, last week, I have been used, uh, polishing the, you know, the unhappy

212
00:14:00,120 --> 00:14:04,880
path and things to, uh, one, one, uh, it was wrong.

213
00:14:04,880 --> 00:14:10,880
Um, that's already almost, almost, almost there, uh, and so I suppose that's two weeks

214
00:14:10,880 --> 00:14:12,880
ago.

215
00:14:12,880 --> 00:14:13,880
Yeah.

216
00:14:13,880 --> 00:14:14,880
Yeah.

217
00:14:14,880 --> 00:14:22,240
For me and to be honest, what decides me more for, from confidential computers, what the

218
00:14:22,240 --> 00:14:24,320
features that it brings to the table, right?

219
00:14:24,320 --> 00:14:30,120
Because one is the confidentiality, that is, uh, the obvious part, right, like at the,

220
00:14:30,120 --> 00:14:33,800
you can grab a very sensible, sensible workloads inside.

221
00:14:33,800 --> 00:14:34,800
Yeah.

222
00:14:34,920 --> 00:14:41,360
Even store, like data, like a keys and some, the provider of the infrastructure is going

223
00:14:41,360 --> 00:14:47,320
to be able to look inside, but the other part that it's also a site for me is, uh, the

224
00:14:47,320 --> 00:14:48,320
attestation part.

225
00:14:48,320 --> 00:14:49,320
Yeah.

226
00:14:49,320 --> 00:14:57,840
Because if you're running a, uh, container inside the, the T, NABM, you can attestate

227
00:14:57,840 --> 00:14:58,840
what is running inside.

228
00:14:58,840 --> 00:15:03,560
So you, you get approved that the container that is running inside is a supposed, the

229
00:15:03,560 --> 00:15:05,560
container that is supposed to be running.

230
00:15:05,560 --> 00:15:06,560
Yeah.

231
00:15:06,560 --> 00:15:11,240
So it's for, for example, you're running a, I don't know, a container that have a, uh,

232
00:15:11,240 --> 00:15:18,920
random number generator, a function, you can attest that that container is, uh, genuine

233
00:15:18,920 --> 00:15:25,320
and it's not biased or, or something that it's actually executing the code that you want

234
00:15:25,320 --> 00:15:26,320
to, to execute.

235
00:15:26,320 --> 00:15:27,320
Exactly.

236
00:15:27,320 --> 00:15:28,320
And anyone can verify.

237
00:15:28,320 --> 00:15:29,320
Yeah.

238
00:15:29,320 --> 00:15:35,240
It's very exciting because, uh, it brings a lot of, uh, interesting properties, uh, you

239
00:15:35,240 --> 00:15:43,280
know, like, uh, oracles, for example, um, things that, applications that before, you

240
00:15:43,280 --> 00:15:47,680
will not, you will think that, uh, you can do, you can deploy it in cloud because, uh,

241
00:15:47,680 --> 00:15:48,680
they were, yeah, yeah.

242
00:15:48,680 --> 00:15:53,160
You can cryptographically prove that it actually executed the thing that it was supposed to

243
00:15:53,160 --> 00:15:54,160
execute it.

244
00:15:54,160 --> 00:15:58,800
In some ways, like, uh, blockchain without a blockchain, you know, yeah, it's, it, uh,

245
00:15:58,800 --> 00:16:00,720
it uses a form of serenology proof, right?

246
00:16:00,720 --> 00:16:03,400
Like it's a similar idea in some sense.

247
00:16:03,400 --> 00:16:04,400
What?

248
00:16:04,400 --> 00:16:05,400
Serenology proof.

249
00:16:05,400 --> 00:16:10,880
It's a similar, it's a, it's a similar idea to, to how, how serenology proofs, uh, work,

250
00:16:10,880 --> 00:16:11,880
right?

251
00:16:11,880 --> 00:16:16,080
Where, where it's, you, you can't, you have cryptographic proof and cryptographic idea

252
00:16:16,080 --> 00:16:23,080
without, uh, cryptographic guarantee, without revealing what is actually going on under

253
00:16:23,080 --> 00:16:24,080
the hood.

254
00:16:24,080 --> 00:16:31,960
Uh, more or less, I mean, uh, not totally right, because, uh, here, basically, you have

255
00:16:31,960 --> 00:16:36,560
a, uh, the image that you, that is supposed to be running, yeah, and you can basically

256
00:16:36,560 --> 00:16:45,400
challenge the, uh, the, the, the, the host with, uh, with a challenge that you provide.

257
00:16:45,400 --> 00:16:50,640
So you create the challenge and you get an answer, you get a report, yeah, and then you

258
00:16:50,640 --> 00:16:53,120
can verify that that report is genuine.

259
00:16:53,120 --> 00:16:56,080
So you know exactly what it's going on.

260
00:16:56,080 --> 00:16:57,080
Okay.

261
00:16:57,080 --> 00:17:01,280
So better comparison might be, a better comparison might be reproducible bills.

262
00:17:01,280 --> 00:17:02,280
Yeah.

263
00:17:02,280 --> 00:17:03,280
Okay.

264
00:17:03,280 --> 00:17:04,280
Yeah, exactly.

265
00:17:04,280 --> 00:17:09,480
Like, uh, you get, uh, uh, not the station of the, from the hardware, because you said

266
00:17:09,480 --> 00:17:11,480
before anyone can verify it.

267
00:17:11,480 --> 00:17:12,480
Yeah.

268
00:17:12,480 --> 00:17:16,720
So, so if you do the same steps, then everyone gets the same result and it's cryptographically

269
00:17:16,720 --> 00:17:20,240
exactly, exactly, that's the key.

270
00:17:20,240 --> 00:17:21,240
Yeah.

271
00:17:21,240 --> 00:17:25,160
No, I'm just trying to make comparisons to, to kind of so that people can get an idea

272
00:17:25,160 --> 00:17:29,560
of, of, of how to think about it in the, in the big picture of what's happening under

273
00:17:29,560 --> 00:17:30,560
the hood.

274
00:17:30,560 --> 00:17:31,560
Yeah.

275
00:17:31,560 --> 00:17:37,040
As important as it is, uh, there are some kind of grown assumptions about these and how

276
00:17:37,040 --> 00:17:44,800
they work, um, because some people I heard that, they think that, uh, when they deploy

277
00:17:44,800 --> 00:17:53,120
a workload, you know, a container, whatever, they won't be able to, to look inside.

278
00:17:53,120 --> 00:17:54,120
But that's not true.

279
00:17:54,120 --> 00:17:58,560
Like, as you as a operator of the container, you can, you can look inside.

280
00:17:58,560 --> 00:17:59,560
Mm-hmm.

281
00:17:59,560 --> 00:18:00,560
It should be M.

282
00:18:00,560 --> 00:18:01,560
Yeah.

283
00:18:01,560 --> 00:18:05,800
Who, who cannot look inside is a, the vendor or, sorry, the provider of the, of the

284
00:18:05,800 --> 00:18:07,400
infrastructure.

285
00:18:07,400 --> 00:18:09,600
So everything is encrypted.

286
00:18:09,600 --> 00:18:18,160
So basically, the, the space is, we had, uh, encryption for this, uh, storage, right?

287
00:18:18,160 --> 00:18:19,880
Like, like, that's, uh, the best.

288
00:18:19,880 --> 00:18:20,880
Yeah.

289
00:18:20,880 --> 00:18:24,280
We have encryption for the transport, that's TLS.

290
00:18:24,280 --> 00:18:28,520
And the missing piece was this encryption, the working memory.

291
00:18:28,520 --> 00:18:29,520
Yeah.

292
00:18:29,520 --> 00:18:31,800
And that's what, what T is sort.

293
00:18:31,800 --> 00:18:37,560
A, you as an operator or as a host, you can look inside, of course.

294
00:18:37,560 --> 00:18:38,560
Yeah.

295
00:18:39,560 --> 00:18:45,040
Um, I think that's, that's interesting to, to reflect on because it's not like, uh,

296
00:18:45,040 --> 00:18:48,920
you deploy something and you don't know what is, what is going on inside.

297
00:18:48,920 --> 00:18:49,920
You can look inside.

298
00:18:49,920 --> 00:18:50,920
Yeah.

299
00:18:50,920 --> 00:18:56,840
Different, different is that you deploy a container or something that, uh, the software

300
00:18:56,840 --> 00:18:59,200
doesn't provide any way to look inside.

301
00:18:59,200 --> 00:19:00,200
And you can prove that.

302
00:19:00,200 --> 00:19:01,200
Yeah.

303
00:19:01,200 --> 00:19:02,200
Exactly.

304
00:19:02,200 --> 00:19:03,200
Okay.

305
00:19:03,200 --> 00:19:04,200
Yeah.

306
00:19:04,200 --> 00:19:05,200
Yeah.

307
00:19:05,200 --> 00:19:10,600
Um, I'm just, I'm just trying to, to come up with other examples, like we had a similar

308
00:19:10,600 --> 00:19:15,720
issue on the VPN side, right, where, uh, your VPN could spy on you, for example.

309
00:19:15,720 --> 00:19:16,720
Yeah.

310
00:19:16,720 --> 00:19:18,400
And, uh, no one else can.

311
00:19:18,400 --> 00:19:24,080
And then, uh, the evolution of VPN's is basically, uh, obscura would be one example,

312
00:19:24,080 --> 00:19:33,080
I think, where, um, the way that the VPN is set up, the VPN itself can prove that it

313
00:19:33,080 --> 00:19:34,080
can't spy on you.

314
00:19:34,080 --> 00:19:35,080
Yeah.

315
00:19:35,080 --> 00:19:39,160
I mean, uh, so that's a similar idea, right, where it's like, yeah, kind of, in a trusted

316
00:19:39,160 --> 00:19:49,840
execution environment, you could execute code that, um, the, the operator itself, um,

317
00:19:49,840 --> 00:19:55,640
where, where, where you, where you can prove, for example, that there is no spying going

318
00:19:55,640 --> 00:19:59,880
on because that's the way the code is structured and that's the way everything is run.

319
00:19:59,880 --> 00:20:00,880
Exactly.

320
00:20:00,880 --> 00:20:08,080
So, um, you hand over that, that code, basically, and you run it, and with the, at the stations,

321
00:20:08,080 --> 00:20:12,960
you have a cryptographic, graphic proof that this code is actually being executed.

322
00:20:12,960 --> 00:20:13,960
Exactly.

323
00:20:13,960 --> 00:20:21,080
And that's the operator actually can't spy on you, and, and, um, and, again, like, there

324
00:20:21,080 --> 00:20:24,920
is no funny business going on that the code is not swapped out, that there is no kind

325
00:20:24,920 --> 00:20:30,160
of, you know, like, the, the binaries are changed and, and you, like, there's no spying

326
00:20:30,160 --> 00:20:31,160
going on.

327
00:20:31,160 --> 00:20:32,160
Yeah.

328
00:20:32,160 --> 00:20:36,560
These are also very important in the, in web pages because you can deploy a, a web server

329
00:20:36,560 --> 00:20:44,040
in a, and an interesting thing is that you can on, of the, the host can swap the JavaScript

330
00:20:44,040 --> 00:20:45,040
for example.

331
00:20:45,040 --> 00:20:46,040
Okay.

332
00:20:46,040 --> 00:20:52,240
Because for, you know, you have some apps that are wallets, for example, crypto wallets

333
00:20:52,240 --> 00:20:59,000
or whatever, and they can swap the JavaScript and maybe screw you up, uh, in a random

334
00:20:59,000 --> 00:21:00,000
update.

335
00:21:00,000 --> 00:21:01,000
Yeah.

336
00:21:01,000 --> 00:21:05,960
But in this, in this case, you cannot do that because that will basically, uh, arm, you

337
00:21:05,960 --> 00:21:11,160
know, uh, like, give you the, the proof that it's not the same, uh, was supposed to be

338
00:21:11,160 --> 00:21:12,160
running.

339
00:21:12,160 --> 00:21:13,160
Yeah.

340
00:21:13,160 --> 00:21:14,160
Yeah.

341
00:21:14,160 --> 00:21:15,160
Yeah.

342
00:21:15,160 --> 00:21:16,160
It's very interesting.

343
00:21:16,160 --> 00:21:17,160
Yeah.

344
00:21:17,160 --> 00:21:18,160
I see.

345
00:21:18,160 --> 00:21:19,160
Yeah.

346
00:21:19,160 --> 00:21:20,160
No.

347
00:21:20,160 --> 00:21:21,160
All you have talked about is, again, already, like, like years back in the first cohort, like,

348
00:21:21,160 --> 00:21:26,840
what kind of, um, novel applications this would make possible because you have this

349
00:21:26,840 --> 00:21:32,760
cryptographic guarantees and because you have these, um, actual assurances, basically,

350
00:21:32,760 --> 00:21:34,440
and these, uh, stations.

351
00:21:34,440 --> 00:21:38,640
So anything that requires trust can be run now.

352
00:21:38,640 --> 00:21:39,640
Yeah.

353
00:21:39,640 --> 00:21:45,680
So that is the cloud we don't need to trust the, the clouds, um, because basically you

354
00:21:45,680 --> 00:21:48,080
have this confidentiality and the, uh, testability.

355
00:21:48,080 --> 00:21:49,080
Yeah.

356
00:21:49,080 --> 00:21:53,640
I think I talked with Kelly about this also about, uh, when we talked about mince and

357
00:21:53,680 --> 00:22:00,960
running mince in TEEs and kind of how this would be different and that kind of, um, I

358
00:22:00,960 --> 00:22:04,880
don't remember precisely what we talked about and the point he made.

359
00:22:04,880 --> 00:22:11,200
I'll have to really listen to that again, but there is some interesting new characteristics

360
00:22:11,200 --> 00:22:20,600
if you run Cashu mints inside the TEE where the mint operator, um, something changes.

361
00:22:20,600 --> 00:22:22,200
I don't remember which, which one it is.

362
00:22:22,200 --> 00:22:23,200
Yeah.

363
00:22:23,200 --> 00:22:24,200
I don't know.

364
00:22:24,200 --> 00:22:27,720
Maybe that you kind of do, uh, fractionary surf.

365
00:22:27,720 --> 00:22:28,720
Yeah.

366
00:22:28,720 --> 00:22:29,720
Something like that.

367
00:22:29,720 --> 00:22:30,720
It was something like that.

368
00:22:30,720 --> 00:22:33,800
There's still the rock pull risk of like the mint just going fully offline, but, um,

369
00:22:33,800 --> 00:22:37,600
you can prove that there is no additional minting going on that there's no fractionary

370
00:22:37,600 --> 00:22:38,600
surf going on.

371
00:22:38,600 --> 00:22:39,600
I think that was it.

372
00:22:39,600 --> 00:22:40,600
I think that, yeah.

373
00:22:40,600 --> 00:22:41,600
That makes sense.

374
00:22:41,600 --> 00:22:42,600
That makes sense as well.

375
00:22:42,600 --> 00:22:43,600
Yeah.

376
00:22:43,600 --> 00:22:46,760
I mean, there is, if you're running something inside of TEE, you kind of, you change it

377
00:22:46,760 --> 00:22:53,160
and, uh, supposedly not do it, you know, is that the good, basically, change it.

378
00:22:53,160 --> 00:22:55,080
That makes everything all the proofs and so on.

379
00:22:55,080 --> 00:22:57,760
So two weeks from now, this thing is going to launch.

380
00:22:57,760 --> 00:23:02,480
Uh, where, where should people go to sign up or provide compute or whatever?

381
00:23:02,480 --> 00:23:04,320
You're saying enclava.dev.

382
00:23:04,320 --> 00:23:05,320
And enclava.dev.

383
00:23:05,320 --> 00:23:06,320
Yeah.

384
00:23:06,320 --> 00:23:07,320
All right.

385
00:23:07,320 --> 00:23:08,320
Good.

386
00:23:08,320 --> 00:23:09,320
That's a good name, by the way.

387
00:23:09,320 --> 00:23:10,320
And Enclava is a great name.

388
00:23:10,320 --> 00:23:11,320
I like it too.

389
00:23:11,320 --> 00:23:14,480
I'm not sure about all the other names, the, the, the, the bus and the, the cab and so on.

390
00:23:14,480 --> 00:23:15,480
We'll figure that out.

391
00:23:15,480 --> 00:23:16,480
Okay.

392
00:23:16,480 --> 00:23:18,040
We'll figure out what that means.

393
00:23:18,040 --> 00:23:22,560
But then enclava.dev, people can go sign up, see what's up.

394
00:23:22,560 --> 00:23:23,560
Who cares anyway?

395
00:23:23,560 --> 00:23:27,240
Like, you, you will care if you want to deploy this kind of stuff and at that point, you

396
00:23:27,240 --> 00:23:32,320
don't care about the name, you know, yeah, I had, I had many conversations like these with

397
00:23:32,320 --> 00:23:33,320
other developers.

398
00:23:33,320 --> 00:23:34,320
Who cares about the name?

399
00:23:34,320 --> 00:23:35,320
Just name it anything.

400
00:23:35,320 --> 00:23:36,320
No.

401
00:23:36,320 --> 00:23:37,320
Like, I really do care about names.

402
00:23:37,320 --> 00:23:38,320
Yeah.

403
00:23:38,320 --> 00:23:39,320
Yeah.

404
00:23:39,320 --> 00:23:40,320
I think they're very important.

405
00:23:40,320 --> 00:23:44,080
Names are important, but, you know, when, when you have a, a product behind them, but

406
00:23:44,080 --> 00:23:48,560
if you're not, basically, it's an open source component and the other is a, I think names

407
00:23:48,560 --> 00:23:49,560
are always important.

408
00:23:49,560 --> 00:23:50,560
Yeah.

409
00:23:50,560 --> 00:23:51,560
So our brains work.

410
00:23:51,560 --> 00:23:52,560
And top of it.

411
00:23:52,560 --> 00:23:53,560
Right.

412
00:23:53,560 --> 00:24:01,320
I think names are terribly important, but that's a conversation for another day.

413
00:24:01,320 --> 00:24:11,520
And if you, if you would compare it to, like I, I brought up before the router as kind of

414
00:24:11,520 --> 00:24:21,440
yeah, some other product slash protocol that is dealing in the, at least like private

415
00:24:21,440 --> 00:24:29,520
provider space, like how to access confidential compute in that way, talk to me about that

416
00:24:29,520 --> 00:24:30,520
a little bit.

417
00:24:30,520 --> 00:24:37,600
Like, how, how is Enclava different from something like Routstr and something like Maple?

418
00:24:37,600 --> 00:24:39,320
Like first of all, it's way more general, right?

419
00:24:39,320 --> 00:24:40,800
Like it deals in T's.

420
00:24:40,800 --> 00:24:41,800
Yeah.

421
00:24:41,800 --> 00:24:42,800
Period.

422
00:24:42,800 --> 00:24:43,800
Yeah.

423
00:24:43,800 --> 00:24:46,680
Like, whatever you run, it doesn't have to necessarily to do anything with AI.

424
00:24:46,680 --> 00:24:47,680
Exactly.

425
00:24:47,680 --> 00:24:48,680
Yeah.

426
00:24:49,680 --> 00:24:55,120
I mean, Routstr, for example, they are very interested in Enclava and then in conversations

427
00:24:55,120 --> 00:24:56,120
with red.

428
00:24:56,120 --> 00:25:03,440
They want to go to access the platform already, but we see pushing, holding it because they

429
00:25:03,440 --> 00:25:09,680
would be interested because they could, they could basically, they could run models inside

430
00:25:09,680 --> 00:25:10,680
T's.

431
00:25:10,680 --> 00:25:17,680
No, not just models, because basically the hardware that we have available is AMD, except

432
00:25:17,680 --> 00:25:25,400
like it says CPU, we don't have GPUs, but they are interesting on, it's basically running

433
00:25:25,400 --> 00:25:35,440
raster nodes inside the T's, because then you can prove that the provider, the node didn't

434
00:25:35,440 --> 00:25:40,560
any kind of prom injection attack or temper proof, okay.

435
00:25:40,560 --> 00:25:46,720
So the routes are known, it's like, yeah, I see, I see, I see, the portal into this world.

436
00:25:46,720 --> 00:25:47,720
Yeah.

437
00:25:47,720 --> 00:25:49,440
Also, it's clean and it's not doing any fun things.

438
00:25:49,440 --> 00:25:50,440
Exactly.

439
00:25:50,440 --> 00:25:54,440
They could also prove that they, they cannot read any logs or that they're not any logs

440
00:25:54,440 --> 00:25:56,640
being recorded.

441
00:25:56,640 --> 00:25:57,640
And so on.

442
00:25:57,640 --> 00:25:58,640
Yeah, okay.

443
00:25:58,640 --> 00:26:04,640
Similar ideas we mentioned before with VPN like Obscura, where you can, can prove that no,

444
00:26:04,640 --> 00:26:06,520
no logging or recording is going on.

445
00:26:06,520 --> 00:26:07,520
Yeah.

446
00:26:07,520 --> 00:26:13,840
And MapleSofar is basically offering the, the inference on, on graphic cards that are, you

447
00:26:13,840 --> 00:26:15,640
know, have this T.

448
00:26:15,640 --> 00:26:16,640
Yeah.

449
00:26:16,640 --> 00:26:23,280
Yeah, they are providing open source models that are running inside T's, right.

450
00:26:23,280 --> 00:26:27,400
In detail, yeah, Safar, very interesting web page, that is a confidential inference

451
00:26:27,400 --> 00:26:28,400
dot net.

452
00:26:28,400 --> 00:26:29,400
Fight the core properly.

453
00:26:29,400 --> 00:26:31,360
Confidentially inference dot net.

454
00:26:31,360 --> 00:26:32,360
Yeah.

455
00:26:32,360 --> 00:26:35,800
Where he compares all these providers and so on.

456
00:26:35,800 --> 00:26:36,800
Exactly.

457
00:26:36,800 --> 00:26:37,800
Okay.

458
00:26:37,800 --> 00:26:45,080
Basically, the reference today is teamfoil, teamfoil dot sh, very interesting in this.

459
00:26:45,080 --> 00:26:50,640
They also have a very good pricing and you can, it's T terminated, TLS terminated.

460
00:26:50,640 --> 00:26:52,960
So you can prove all, you can prove everything.

461
00:26:52,960 --> 00:26:53,960
Okay.

462
00:26:53,960 --> 00:26:54,960
Okay.

463
00:26:54,960 --> 00:26:55,960
I haven't used tinfoil yet.

464
00:26:55,960 --> 00:27:00,080
I'm aware of it, but I, I mean, I've been using voucher, I've been using Maple, I haven't

465
00:27:00,080 --> 00:27:01,080
used tinfoil yet.

466
00:27:01,080 --> 00:27:02,080
Yeah.

467
00:27:02,080 --> 00:27:03,080
Are you using it yourself?

468
00:27:03,080 --> 00:27:11,600
I use it because also I just is developing a library for infer, confidential inference.

469
00:27:11,600 --> 00:27:17,760
Basically, for you to not need to, to trust anyone, like I do, if you use this library

470
00:27:17,760 --> 00:27:23,800
to do inference, to connect with a model on tinfoil or whatever, you do the, the attestation

471
00:27:23,800 --> 00:27:28,480
on the proofs, when you, when you do the, the inference.

472
00:27:28,480 --> 00:27:34,800
So every time that you do it, that you do try to call the model, you prove everything

473
00:27:34,800 --> 00:27:35,800
before hand.

474
00:27:35,800 --> 00:27:36,800
Okay.

475
00:27:36,800 --> 00:27:45,160
So it's not like a something shady going on, you know, some day change, you can, you

476
00:27:45,160 --> 00:27:48,600
can, you can basically prove everything, that's an interesting part.

477
00:27:48,600 --> 00:27:53,840
What are you using right now in your day to day kind of, I assume, you know, like you

478
00:27:53,840 --> 00:27:59,880
are as, you know, I'm not sure what your level of AI psychosis is, you're probably not,

479
00:27:59,880 --> 00:28:01,480
I tried to keep it very low.

480
00:28:01,480 --> 00:28:06,560
It's probably not where, you know, like probably used to be or just in or, you know,

481
00:28:06,560 --> 00:28:07,560
mark him all me.

482
00:28:07,560 --> 00:28:13,240
I have some periods of that, but I decide that, like, you're not running around with your

483
00:28:13,240 --> 00:28:16,600
laptop open right now, you know, that's, that's how you can tell when things are really

484
00:28:16,600 --> 00:28:17,600
bad.

485
00:28:17,600 --> 00:28:21,480
I had, I had a couple of weeks like that, and then I decide that it wasn't good for my

486
00:28:21,480 --> 00:28:22,480
health.

487
00:28:22,480 --> 00:28:23,480
Smart.

488
00:28:23,480 --> 00:28:24,480
Smart.

489
00:28:24,480 --> 00:28:30,640
I basically tried to keep it, you know, low, the AI as, I mean, I use it every day.

490
00:28:30,640 --> 00:28:31,760
So what do you use?

491
00:28:31,760 --> 00:28:33,240
What does your setup look like?

492
00:28:33,240 --> 00:28:34,840
Like what kind of providers do you use?

493
00:28:34,840 --> 00:28:36,160
Do you run something yourself?

494
00:28:36,160 --> 00:28:37,160
Like what?

495
00:28:37,160 --> 00:28:40,200
So right now, I'm basically using Chinese models.

496
00:28:40,200 --> 00:28:45,840
I don't touch nothing from the US, especially because I don't want to do KYC with them.

497
00:28:45,840 --> 00:28:46,840
Interesting, right?

498
00:28:46,840 --> 00:28:48,360
Like what kind of world we're living in?

499
00:28:48,360 --> 00:28:49,360
Oh, yeah.

500
00:28:49,360 --> 00:28:52,840
Who would have guessed 10 years ago that that's where we would be?

501
00:28:52,840 --> 00:28:59,800
Like the United States of America is basically just, you know, fully cocked and fully surveilling

502
00:28:59,800 --> 00:29:00,800
you.

503
00:29:00,800 --> 00:29:01,800
And everything.

504
00:29:01,800 --> 00:29:04,560
And so the thing to do is use Chinese models.

505
00:29:04,560 --> 00:29:05,560
Oh, yeah.

506
00:29:05,560 --> 00:29:07,040
Who would remind you?

507
00:29:07,040 --> 00:29:08,040
I think it's really funny.

508
00:29:08,040 --> 00:29:09,040
Who would have thought?

509
00:29:09,040 --> 00:29:10,040
Who would have thought?

510
00:29:10,040 --> 00:29:15,840
I mean, you can kind of get a reflex of that in the current world, right?

511
00:29:15,840 --> 00:29:17,520
Like everything comes from China.

512
00:29:17,520 --> 00:29:18,520
Oh, yeah.

513
00:29:18,520 --> 00:29:23,840
Also, you're on the island half the cars are Chinese, everyone's driving the BYD electric

514
00:29:23,840 --> 00:29:24,840
cars.

515
00:29:24,840 --> 00:29:25,840
It's insane.

516
00:29:25,840 --> 00:29:26,840
It's insane.

517
00:29:26,840 --> 00:29:27,840
Yeah.

518
00:29:27,840 --> 00:29:31,760
Anyways, I use GLM mostly.

519
00:29:31,760 --> 00:29:35,840
Also I'm testing lately the Dishyami model, the Mimo.

520
00:29:35,840 --> 00:29:36,840
Okay.

521
00:29:36,840 --> 00:29:37,840
It's very good.

522
00:29:37,840 --> 00:29:40,360
And what do you use to run this?

523
00:29:40,360 --> 00:29:41,360
Spy.

524
00:29:41,360 --> 00:29:42,360
Spy?

525
00:29:42,360 --> 00:29:43,360
Spy, bye.

526
00:29:43,360 --> 00:29:44,360
Oh, pie.

527
00:29:44,360 --> 00:29:45,360
Okay.

528
00:29:45,360 --> 00:29:46,360
Yeah.

529
00:29:46,360 --> 00:29:47,360
Okay.

530
00:29:47,360 --> 00:29:48,360
This pie.

531
00:29:48,360 --> 00:29:50,160
And vanilla pie with some plug and some, I love it to be honest.

532
00:29:50,160 --> 00:29:51,160
Yeah.

533
00:29:51,160 --> 00:29:52,160
It's the best.

534
00:29:52,160 --> 00:29:55,480
The best harness I have been the hard test.

535
00:29:55,480 --> 00:30:01,400
Because open code and some, okay, they are, I don't know, very opinionated.

536
00:30:01,400 --> 00:30:02,400
Yeah.

537
00:30:02,400 --> 00:30:03,400
And it's okay.

538
00:30:03,400 --> 00:30:05,280
But they want to opinionate it myself.

539
00:30:05,280 --> 00:30:07,640
Have you tried herder as well?

540
00:30:07,640 --> 00:30:08,640
Oh, yeah.

541
00:30:08,640 --> 00:30:09,640
Yeah.

542
00:30:09,640 --> 00:30:10,640
Yeah.

543
00:30:10,640 --> 00:30:13,840
I run it for a couple of weeks and then I move to Orca.

544
00:30:13,840 --> 00:30:16,760
That is same thing.

545
00:30:16,760 --> 00:30:22,560
But with the graphical interface instead of a toy.

546
00:30:22,560 --> 00:30:29,680
And all my, my agents and workloads in Orca inside, so P inside Orca.

547
00:30:29,680 --> 00:30:30,680
Okay.

548
00:30:30,680 --> 00:30:36,680
It's basically you can keep track of agents and you get sounds when they finish and sometimes

549
00:30:36,680 --> 00:30:37,680
of all.

550
00:30:37,680 --> 00:30:38,680
I see.

551
00:30:38,680 --> 00:30:39,680
Is your laptop your main interface?

552
00:30:39,680 --> 00:30:40,680
Yeah.

553
00:30:40,680 --> 00:30:41,680
So, yeah.

554
00:30:41,680 --> 00:30:46,120
That's the key for not, not going through the AIC causes the track, you know.

555
00:30:46,120 --> 00:30:47,120
Okay.

556
00:30:47,200 --> 00:30:54,000
If you laptop in your workplace is move away and you don't have the pressure of you need

557
00:30:54,000 --> 00:30:56,680
to look at the phone or whatever.

558
00:30:56,680 --> 00:31:03,680
I mean, I get the point of, you know, going in a work like this and talk to your phone

559
00:31:03,680 --> 00:31:06,520
and keep working and so on.

560
00:31:06,520 --> 00:31:11,120
But I don't want to go crazy, man, that's very bad for your health.

561
00:31:11,120 --> 00:31:12,760
I appreciate my health.

562
00:31:12,760 --> 00:31:15,560
You're a very smart man.

563
00:31:15,560 --> 00:31:20,360
I'm on the other side of the spectrum.

564
00:31:20,360 --> 00:31:24,920
I just basically, I think it flipped for me a couple of weeks ago.

565
00:31:24,920 --> 00:31:27,520
I just realized there's a couple of days ago.

566
00:31:27,520 --> 00:31:29,160
Or maybe I even realized it yesterday.

567
00:31:29,160 --> 00:31:36,840
I probably, so my setup, I mean, you know, like I've built a vibe line a while ago, like

568
00:31:36,840 --> 00:31:39,440
a one and a half years ago or longer than that.

569
00:31:39,440 --> 00:31:44,280
And so I've been a fan of just using voice for a long time.

570
00:31:44,280 --> 00:31:49,520
And they've got more extreme over time and now I have now multiple laptops.

571
00:31:49,520 --> 00:31:55,120
And because you know, some stuff runs locally and so on and you know, sometimes something

572
00:31:55,120 --> 00:31:56,120
weird is going on.

573
00:31:56,120 --> 00:31:58,040
And so I need to kind of multiple hosts.

574
00:31:58,040 --> 00:32:03,320
Laptops are great because they have, you know, search protection and everything integrated.

575
00:32:03,320 --> 00:32:07,640
You know, like if the power goes out, it doesn't matter, which is fantastic.

576
00:32:07,640 --> 00:32:11,800
Like that's because I have other machines as well and every time, like, you know, there's

577
00:32:11,800 --> 00:32:16,440
a search or the power goes out or something weird is going on and like I have to physically

578
00:32:16,440 --> 00:32:19,200
go to the machine and make it work again and so on.

579
00:32:19,200 --> 00:32:23,920
And so laptops are fantastic for that reason that they have the battery and everything integrated.

580
00:32:23,920 --> 00:32:26,400
And so I've three laptops right now.

581
00:32:26,400 --> 00:32:33,680
And they are all connected to my phone and the vibe line and I also, like I use NIP17

582
00:32:33,680 --> 00:32:34,680
a lot.

583
00:32:34,680 --> 00:32:39,360
So all my agents, every project has an end pump and I just talk to the project and like

584
00:32:39,360 --> 00:32:41,360
mostly with voice and but sometimes also.

585
00:32:41,360 --> 00:32:47,960
So the question is, how do you vibe in there now and yes and no, it's like, I mean, you

586
00:32:47,960 --> 00:32:50,000
know, this thing is plugging into the vibe line.

587
00:32:50,000 --> 00:32:55,680
So as soon as I hit save, it just sends it to the right agent and stuff start happening.

588
00:32:55,680 --> 00:33:02,360
And so that's basically, that's basically has been my, like the last year or so, I've

589
00:33:02,360 --> 00:33:06,880
been working mostly with voice memos, basically, where I just, I record something and then

590
00:33:06,880 --> 00:33:11,080
based on the keywords in the voice memo, it just goes to the right agent.

591
00:33:11,080 --> 00:33:14,160
And kind of transcription and everything happens locally.

592
00:33:14,160 --> 00:33:21,320
And now it kind of is kind of switching where I kind of have multiple conversations at

593
00:33:21,320 --> 00:33:24,280
the same time, but I still use voice more or less.

594
00:33:24,280 --> 00:33:29,720
Voice is replacing my typing and my chest right now because the voice recognition is

595
00:33:29,720 --> 00:33:31,200
just so good.

596
00:33:31,200 --> 00:33:39,040
It's so good now, like it and and also the models are so good now that even if the transcription

597
00:33:39,040 --> 00:33:45,360
is slightly off, it will, like when I say Jesus, for example, it will understand that

598
00:33:45,360 --> 00:33:54,760
I mean you and not, you know, Jesus of Nazareth, not the other one, there's only two.

599
00:33:54,760 --> 00:34:00,960
And so it will automatically translate it to GC, you know, whatever, how many sees you

600
00:34:00,960 --> 00:34:02,960
have and so on.

601
00:34:02,960 --> 00:34:07,600
And so the things just got better and better and better where, like, I feel that's

602
00:34:07,600 --> 00:34:12,280
very, I'm going through a very strange phase of my life right now.

603
00:34:12,280 --> 00:34:16,240
So I used to be heavy and I used to be big on mechanical keyboards.

604
00:34:16,240 --> 00:34:18,520
And then I switched to laptops mostly.

605
00:34:18,520 --> 00:34:20,040
And so this kind of went away.

606
00:34:20,040 --> 00:34:22,360
Now it's coming back because I want to start writing again.

607
00:34:22,360 --> 00:34:24,600
So I got mechanical keyboards again.

608
00:34:24,600 --> 00:34:28,520
But most of my computer interfaces voice now.

609
00:34:28,520 --> 00:34:29,520
So it's all very weird.

610
00:34:29,520 --> 00:34:33,440
I'm developing a taste again for mechanical keyboards, even though I'm typing less and

611
00:34:33,440 --> 00:34:34,440
less.

612
00:34:34,520 --> 00:34:38,720
I still want to sit down and actually write, because writing is different than talking,

613
00:34:38,720 --> 00:34:39,720
like very different.

614
00:34:39,720 --> 00:34:40,720
Yeah, yeah.

615
00:34:40,720 --> 00:34:46,680
So I hate AI Slop, that's why I want to sit down and have like a very good clickety-clack

616
00:34:46,680 --> 00:34:49,360
old school mechanical keyboard and actually type something.

617
00:34:49,360 --> 00:34:50,360
Anyway, I digress.

618
00:34:50,360 --> 00:34:51,360
Yeah.

619
00:34:51,360 --> 00:34:56,400
So I assume that it's like the same that everything, right, like the right tool for the right

620
00:34:56,400 --> 00:34:57,400
task.

621
00:34:57,400 --> 00:34:58,400
Yeah.

622
00:34:58,400 --> 00:35:05,400
So you are not going to use voice to express the thought or something that you want to issue

623
00:35:05,400 --> 00:35:10,480
and to basically understand before you say it.

624
00:35:10,480 --> 00:35:11,480
Yeah.

625
00:35:11,480 --> 00:35:12,480
And then you want to type it.

626
00:35:12,480 --> 00:35:13,480
Exactly.

627
00:35:13,480 --> 00:35:14,480
You want to go back.

628
00:35:14,480 --> 00:35:15,480
You want to correct this, fix that.

629
00:35:15,480 --> 00:35:20,560
That's why writing a book, for example, is so important because you really wrestle with

630
00:35:20,560 --> 00:35:25,720
the topic for many weeks and months and so on or even years, you know, and it's very

631
00:35:25,720 --> 00:35:26,720
different.

632
00:35:26,720 --> 00:35:31,600
Like getting together and talking about something, you think you hashed it out, but like oftentimes

633
00:35:31,600 --> 00:35:32,600
you didn't.

634
00:35:32,600 --> 00:35:33,600
Yeah.

635
00:35:33,600 --> 00:35:41,280
Sitting down and actually writing, you're trying to put your thoughts into actual written

636
00:35:41,280 --> 00:35:46,680
words is very different and way more difficult and again, like right tool for the job.

637
00:35:46,680 --> 00:35:52,560
But it's also for me, like my, I barely do development, even though like, you know,

638
00:35:52,560 --> 00:35:57,360
right now I'm publishing a new app every week, but it's not, it's very different from

639
00:35:57,360 --> 00:36:02,880
the kind of development work that you do where you have to actually sit in front of the laptop,

640
00:36:02,880 --> 00:36:06,800
actually wrestle with the material, actually go back and forth, actually debug stuff.

641
00:36:06,800 --> 00:36:09,880
And so on voice is useless for that, like you still have to type.

642
00:36:09,880 --> 00:36:15,480
For me, it's all very big picture and like, yeah, high level, very high level and like

643
00:36:15,480 --> 00:36:20,920
the last app, for example, I just published this and mobile and Android, the search engine

644
00:36:20,920 --> 00:36:25,400
I built for Noster and it has so much prior work that I can do basically everything with

645
00:36:25,400 --> 00:36:32,160
voice prompts because I have the web app, I have the query language and everything defined

646
00:36:32,160 --> 00:36:38,120
and I can just go like, you know, like take the web app, build it for Android.

647
00:36:38,120 --> 00:36:43,480
And then it has one thing, one version, I install it and then like something's broken,

648
00:36:43,480 --> 00:36:48,280
something's weird and then I just again, like record a voice memo and feel like, you

649
00:36:48,280 --> 00:36:52,080
know, like this is weird, the search should clear if I press back and like all those kind

650
00:36:52,080 --> 00:36:53,080
of things.

651
00:36:53,080 --> 00:36:58,280
And then it just publishes the next version, like I don't even use the buggers, I don't

652
00:36:58,280 --> 00:37:03,480
use anything, it just pushes straight to the app store, I see it live on my phone, it's

653
00:37:03,480 --> 00:37:08,760
the best development workflow, I just, I just develop on production basically and because

654
00:37:08,760 --> 00:37:14,960
I can be out and about, I talk directly to the app and within like 30 seconds it changes

655
00:37:15,680 --> 00:37:20,160
or like one minute or whatever, like however it longs, it doesn't take long to build and

656
00:37:20,160 --> 00:37:22,800
push out and then the update is there and it installs.

657
00:37:22,800 --> 00:37:23,920
No, it's fucking crazy, man.

658
00:37:23,920 --> 00:37:25,280
So it's fucking crazy.

659
00:37:25,280 --> 00:37:30,960
But yeah, it's also unhealthy, like I completely agree with you that like, my issue is I

660
00:37:30,960 --> 00:37:35,920
have a very hard time switching off, I always think about these things, the benefit now is

661
00:37:35,920 --> 00:37:41,920
I think about something I can immediately act on it, but it's also, sometimes it isn't

662
00:37:41,920 --> 00:37:45,840
good, you know, like sometimes you should just sleep on it and not act on it.

663
00:37:45,840 --> 00:37:51,520
In that, in that, yeah, I think that that's the dangerous part, you know, of these, I

664
00:37:51,520 --> 00:37:57,760
think, psychosis, ELA or phaser, we are living or passing through because you kind of

665
00:37:57,760 --> 00:38:05,600
sleep on things, like you just throw things to the, to whatever input mechanism you have,

666
00:38:05,600 --> 00:38:11,840
input that you have and, and that's it, like you, in any other moment,

667
00:38:11,840 --> 00:38:16,960
of your day, you think that you are going to loosen your time because you are not inputting

668
00:38:16,960 --> 00:38:23,040
the, I mean, the productivity trap basically, you know, you want to be productive always.

669
00:38:23,040 --> 00:38:29,440
And the question I have right now for, to be honest, that's fucking bullshit, man.

670
00:38:29,440 --> 00:38:31,440
Oh, it is.

671
00:38:31,440 --> 00:38:37,120
Something that reminds me to this, this philosopher that I discovered through you,

672
00:38:37,520 --> 00:38:39,040
I don't remember the name by now.

673
00:38:39,040 --> 00:38:39,840
Oh, Bung Trulheim.

674
00:38:39,840 --> 00:38:40,320
It's exactly.

675
00:38:40,320 --> 00:38:41,280
Yeah.

676
00:38:41,280 --> 00:38:43,280
Chinese, German, philosopher.

677
00:38:43,280 --> 00:38:47,680
Yeah, I was just about to mention that the same people figured this out like 3000 years ago,

678
00:38:47,680 --> 00:38:51,280
like you don't have to always do something, basically, you know, like, again, like

679
00:38:51,280 --> 00:38:55,760
towards what end, it's like like progress, you know, it's like, okay, progress towards what end,

680
00:38:55,760 --> 00:39:00,960
it's like, it's a meaningful concept, progress, like the Western idea of progress, or also,

681
00:39:00,960 --> 00:39:05,200
the Silicon Valley idea of technological progress, like, progress to what end, towards the

682
00:39:05,200 --> 00:39:11,680
singularity, or what, it's like, hey, and then what, you know, it's like, yeah, anyway,

683
00:39:11,680 --> 00:39:18,320
I mean, it's, I completely agree that it's also someone, someone wrote a no-story while

684
00:39:18,320 --> 00:39:24,080
ago, it's like not, not every, not every shower thought should be turned into an app, you know,

685
00:39:24,080 --> 00:39:28,480
and I was like, yeah, that's actually right, but that's also the world we're living in right now,

686
00:39:28,480 --> 00:39:33,520
where every shower thought can be, and probably also will be turned into an app,

687
00:39:34,000 --> 00:39:38,240
but not only an app, also, you know, whatever, an essay, or a movie, or a song, or whatever,

688
00:39:38,240 --> 00:39:42,400
like that's the world we're living in right now. I mean, to be honest, I think that there is,

689
00:39:42,400 --> 00:39:48,480
and very interesting edge of this, and is that you can create your own applications for yourself,

690
00:39:48,480 --> 00:39:52,320
yeah, because maybe you have something, exactly, because maybe you have a need,

691
00:39:52,320 --> 00:39:57,680
or you have a need that you want to scratch, and that is super easy to do that, like,

692
00:39:57,760 --> 00:40:03,760
do you just talk to your mug if you want, and then you have a personalized stuff,

693
00:40:03,760 --> 00:40:09,520
and it's, I mean, it's, I'm still trying to wrap my head around it, it's like also in the

694
00:40:09,520 --> 00:40:14,160
context of OpenSats, for example, you know, for me right now, it's very hard to wrap my head around

695
00:40:14,880 --> 00:40:20,240
kind of, like, we used to fund projects, for example, you know, like, I want to create a new

696
00:40:20,240 --> 00:40:25,200
Bitcoin wallet, okay, like, here's, you know, here's some funding, here's a 12-month timeline

697
00:40:25,280 --> 00:40:29,040
that you've sketched out and so on, and like, you know, like, work on it for 12 months,

698
00:40:29,040 --> 00:40:32,960
it's completely nonsensical right now, I feel like, to do that, because it's like,

699
00:40:33,840 --> 00:40:38,320
almost like, whatever project you can imagine, like, almost, like, there's caveats,

700
00:40:38,320 --> 00:40:42,800
and there's like, actually tricky stuff and so on, but it's just the timelines got so compressed

701
00:40:42,800 --> 00:40:47,440
that, unless your stuff is like insanely complex, you're done after like two weeks, or whatever,

702
00:40:47,440 --> 00:40:51,680
you know, like, in a very short period of time, you're basically done, and you're in maintenance mode,

703
00:40:51,680 --> 00:40:57,360
and the clankers are fantastic at maintenance, and so it's like, okay, what, what, like, like,

704
00:40:58,480 --> 00:41:03,920
divvying the work up into projects almost doesn't make sense anymore, it's like, you, like,

705
00:41:03,920 --> 00:41:09,680
like, that's why my kind of thinking around that right now is like, I want to find the best people,

706
00:41:09,680 --> 00:41:15,200
and just support them as best as I can, and then it's like, they can't do whatever, you know, like,

707
00:41:15,280 --> 00:41:21,040
Cali is a great example, and like, I mean, Pablo fell off a cliff recently, but in his heyday,

708
00:41:21,040 --> 00:41:25,920
Pablo was a great example, which is like, whatever he works on, I'm fine, it's like, it's just work on that.

709
00:41:25,920 --> 00:41:31,280
Yeah, it's creativity and open source, and I think in general, in the open source world,

710
00:41:31,280 --> 00:41:37,600
that's where we are heading, where people just work on 10, 15 different things, so whatever,

711
00:41:37,600 --> 00:41:43,520
or like, like, really big stuff, DHH is a great example also with, with Omar Chi, you know, it's like,

712
00:41:44,320 --> 00:41:53,040
like, you can do these things now, if you're a high agency, you can actually go ham and kind of

713
00:41:54,880 --> 00:42:00,320
reimagine how an operating system should look like, you know, and I mean, people give him a lot of

714
00:42:00,320 --> 00:42:04,960
shit, it's just like arch with some scripts on top of it and so on, but it's still trying to do

715
00:42:04,960 --> 00:42:11,520
something like Omar Chi, whatever, like three years ago, you know, you will take a couple of years.

716
00:42:11,520 --> 00:42:18,000
Impossible, basically. Yeah, but I definitely believe that we are not jet on the end of this,

717
00:42:18,000 --> 00:42:23,600
like, no, we are definitely at the beginning, and we don't know who is going to look in a year or two

718
00:42:23,600 --> 00:42:28,800
years. Yeah, but to me, it's clear, like, what you said before, it resonates so strongly with me,

719
00:42:28,800 --> 00:42:33,200
with like, you can now build an app, just for yourself, just subscribe to your own, it's just

720
00:42:33,200 --> 00:42:37,280
for your own needs, which is what open source always was about, kind of, you know, like, that's,

721
00:42:37,280 --> 00:42:41,840
that there was part of the idea. And if, like, an open source project is already successful,

722
00:42:41,840 --> 00:42:46,800
if it has one user, or even if the user is you, if it has two users, it's like massively successful.

723
00:42:47,360 --> 00:42:51,520
You know, that was the idea. Like, the, the, all the, the, the early open source people talked

724
00:42:51,520 --> 00:42:56,400
about that beautifully, and especially now is, you know, because there is a lot of apps and

725
00:42:56,400 --> 00:43:03,120
new users. Yeah, but yes, and that's kind of, but that's to be expected, like, if software used to

726
00:43:03,120 --> 00:43:09,040
be a scarce thing, it's not anymore. Software isn't scarce anymore. Like, we, we now lived in,

727
00:43:09,040 --> 00:43:14,640
in post, post-scarcity software, post-scarcity apps. Yeah. And like, if you just look at

728
00:43:14,640 --> 00:43:19,680
App Store, like, one of my daily rituals is like, open App Store, and just, like, refresh the

729
00:43:19,680 --> 00:43:24,640
App Store fee, basically, what are the new apps that are being published? It's like, it's like,

730
00:43:25,200 --> 00:43:29,920
15 new apps every day, or something like that. Like, you can't keep up at all, and it's a very small

731
00:43:30,880 --> 00:43:35,440
bubble. It's like our bubble, right? It's like, nocturne, and like, whatever, freedom tech stuff.

732
00:43:36,080 --> 00:43:42,160
And it's impossible to keep up, because anyone can just speak and app into existence now.

733
00:43:42,160 --> 00:43:48,160
And it's, and it's, and so, and so I'm wondering, like, yeah, it's very strange. It's, it's definitely

734
00:43:49,760 --> 00:43:56,320
the biggest change of my, you know, like, 30 years in software. It's, it's, it's the biggest

735
00:43:56,320 --> 00:44:02,080
change. It's very, very strange. And I don't know what to do with it. I think it's very exciting.

736
00:44:03,360 --> 00:44:09,520
I think it's, I don't think it's bad, for example. Like, I'm not a, oh, well, you know,

737
00:44:09,520 --> 00:44:14,960
everyone has to write software by hand. Otherwise, it's not artisan software, you know. That doesn't

738
00:44:14,960 --> 00:44:23,920
make sense. Definitely, there is a quality bar that's a piece of software should reach, you know,

739
00:44:24,000 --> 00:44:28,960
for, for it to not to exclude their users, especially if they are some, some sensible,

740
00:44:31,280 --> 00:44:36,400
you know, application or use case. But everything is changing in that sense. Yeah. I'm not using

741
00:44:36,400 --> 00:44:43,120
the, in the software and how do you create software, the craft, also in the UX site. Like,

742
00:44:43,920 --> 00:44:50,800
user experience, like, absolutely. And we are not yet even imagining where we're going to arrive

743
00:44:50,880 --> 00:44:54,880
with all this. I don't click buttons or use interfaces anymore, basically.

744
00:44:55,920 --> 00:45:00,960
I see. I see. I speak to my phone and it gives me text messages back.

745
00:45:01,920 --> 00:45:05,840
And, and I have, again, like, I have different machines and different environments,

746
00:45:05,840 --> 00:45:10,160
different models, different access levels, different things. Some stuff is very private.

747
00:45:10,160 --> 00:45:14,000
Other stuff, like, I don't care at all. It's an open source project, like, I'll, I'll use whatever.

748
00:45:14,640 --> 00:45:19,200
And, and it's very interesting to, to reflect about, like, at the fields,

749
00:45:20,800 --> 00:45:24,320
UX that we had with computers was the command lines, right? Yeah.

750
00:45:25,200 --> 00:45:32,880
Then we went to, yeah, then we went to the, the graphical interface world because you can abstract

751
00:45:32,880 --> 00:45:41,680
commands behind buttons. And now we are coming back to the, to the text. Now with this natural

752
00:45:41,680 --> 00:45:47,280
language approach, which is very different that speaking commands. It is, and it's to your point

753
00:45:47,280 --> 00:45:51,680
regarding the quality and those kinds of things. I mean, if you build control software for a

754
00:45:51,680 --> 00:45:56,880
new click of power plant or for whatever, you know, like, the dikes where you have to close the

755
00:45:56,880 --> 00:46:02,240
dam otherwise the area will flood and those kinds of things. Like, but I, you obviously can't just

756
00:46:02,240 --> 00:46:07,360
vibe, like, people are vibing JavaScript apps, you know? Like, you have to use something else.

757
00:46:07,360 --> 00:46:13,920
But I still think that, like, agents are exceptionally good at using formal verification systems,

758
00:46:13,920 --> 00:46:20,800
for example. And, and you see this that all the LLMs are going ham on, on kind of the,

759
00:46:20,800 --> 00:46:25,920
the big problems in math right now because in math, they have these proof systems that they have

760
00:46:25,920 --> 00:46:32,240
their own programming languages, basically, for mathematical proofs. And agents, of course,

761
00:46:32,240 --> 00:46:37,120
can understand that beautifully and just, you know, like, like, I just made this proof and here is

762
00:46:37,120 --> 00:46:41,920
the actual step-by-step mathematical proof for it. And, like, you can verify it yourself. It's

763
00:46:41,920 --> 00:46:47,600
similar to what we talked before with TEEs and those kinds of things where, you know, here is

764
00:46:47,600 --> 00:46:54,240
the valid signature or here's the challenge response and everything checks out. Like, you know,

765
00:46:54,240 --> 00:47:00,240
here you go. The agent handed over with like all green checkmarks. That's a, that's a double-edged

766
00:47:00,240 --> 00:47:06,800
sword because you have these proof purifications. Like, in the mathematical side of things and

767
00:47:06,800 --> 00:47:14,960
this revolution of LLMs, the solving problems that no one has solved before and they still

768
00:47:14,960 --> 00:47:23,840
have resolved. And, yeah, you as an agent, you can prove or use the proof system to basically

769
00:47:23,840 --> 00:47:31,440
say, I resolved this. But now someone has to validate it. Normally, I'm a human. Yeah. And that

770
00:47:31,440 --> 00:47:35,920
human needs to understand the problem that had never been resolved before. Yeah. So that takes the

771
00:47:35,920 --> 00:47:40,880
time. So you can claim that you can claim that you solve it because the proof checks,

772
00:47:41,920 --> 00:47:46,960
but you need, you don't need to explain how, how that works because it doesn't, it doesn't,

773
00:47:46,960 --> 00:47:53,760
it doesn't matters. Yeah. That's, that's, it's so funny also because, because the reason why we do

774
00:47:53,760 --> 00:47:59,760
all this is to deepen our understanding of how the universe works basically. And so if there is

775
00:47:59,760 --> 00:48:06,160
no understanding going on, like, what's the point? You can't also, like, you can't even apply it if

776
00:48:06,160 --> 00:48:12,400
you don't understand it, you can't apply it, right? And so, yeah, it, like, humans definitely have some,

777
00:48:13,360 --> 00:48:17,680
I'm not sure if it's catching up or if it's just like integration, I don't know what the

778
00:48:17,680 --> 00:48:26,640
right word would be, but, but, yeah, it's definitely a very strange new world. And I know too little

779
00:48:26,640 --> 00:48:31,600
about the, the math part and the physics and material science part and so on, but it's, it's

780
00:48:32,160 --> 00:48:40,880
obviously, I mean, a little answer used absolutely everywhere. What I, what I kind of wonder,

781
00:48:43,120 --> 00:48:46,960
the way I look at it is, like, it's, it's kind of blessing and curse.

782
00:48:50,160 --> 00:48:55,760
Curse because, again, every shower thought will be an app and I don't want to see it. It's similar

783
00:48:55,760 --> 00:49:01,680
to like, I don't know, modern music or whatever, like, it's, it's just, like, you don't need to put

784
00:49:01,680 --> 00:49:07,840
auto tune over every single, whatever thought you just had and make a musical piece out of it.

785
00:49:07,840 --> 00:49:13,600
I'm not interested in listening to it, you know, and I will also not be interested in the vast

786
00:49:13,600 --> 00:49:19,840
majority of the apps that are going to be created. But it's also a blessing because for the stuff

787
00:49:19,840 --> 00:49:28,240
that we are interested in, just creating software that empowers people, creating software that,

788
00:49:31,360 --> 00:49:37,200
yeah, empowers the individual basically and gives the power back to the user and puts users in

789
00:49:37,200 --> 00:49:44,880
control. It's, it's building all that out. It's just good. I don't know, like, a hundred X easier

790
00:49:44,880 --> 00:49:52,720
and a thousand X faster or something like that. So there's actually, and again,

791
00:49:52,720 --> 00:49:57,600
Armaging is a great example. Like, the motivation that DHH had to build it was because he got

792
00:49:57,600 --> 00:50:03,760
cucked by Apple so hard. It was his whole, like, it's basically, it's basically and he did it

793
00:50:03,760 --> 00:50:10,400
out of spite. Like, he was fighting with the app lab store and, you know, disagreeing with the

794
00:50:10,400 --> 00:50:16,640
tax and all those kind of things. That's another fight. Yeah. And, and he was like, it's ridiculous.

795
00:50:16,640 --> 00:50:22,400
It's ridiculous. I should own my own compute. I should own, I should have a say what my computer does.

796
00:50:22,400 --> 00:50:27,760
I should have a say what programs run on it. I don't want to rent the thing that I own that I

797
00:50:27,760 --> 00:50:34,400
should own, you know. And it's, it's, again, coming back to the, to, to what we talked before

798
00:50:34,400 --> 00:50:41,280
about the TEEs. I think that's in part also a big motivation for building out TEEs infrastructure

799
00:50:41,280 --> 00:50:48,400
because it's like, I want to run this compute. Yeah. Don't tell me that I can't do it, you know,

800
00:50:48,400 --> 00:50:54,320
for example. And also, like, don't, don't mess with my compute. Yeah. Similar to encrypted files.

801
00:50:54,320 --> 00:50:58,880
Don't mess with my files. Yeah. Don't look into them. Don't, don't do any funny business.

802
00:50:58,880 --> 00:51:05,760
It's my encrypted blob. Similar with communications, right? To be honest, talking about TEEs again.

803
00:51:07,440 --> 00:51:12,880
I think that they sit in a very interesting place but they are not the silver ballad for everything.

804
00:51:12,880 --> 00:51:19,680
Like, if you can run your compute locally, there is nothing better than that. Yeah. Nothing better than

805
00:51:20,160 --> 00:51:26,880
that. Like, TEEs are very interested, interesting for the tensions that emerge with, for example,

806
00:51:26,880 --> 00:51:36,800
companies or people that need to run some workloads. They kind of, you know, spend the money up

807
00:51:36,800 --> 00:51:43,440
from the hardware to run it or the infrastructure to run it. Yeah. That's a great point. Exactly.

808
00:51:43,440 --> 00:51:48,640
Like, attention was, was that right? Like, if you don't want to spend, you want to rent. Yeah.

809
00:51:50,320 --> 00:51:55,440
Then you can no do it with the TEEs because you have the proofs on sound. Yeah. Yeah. I mean,

810
00:51:55,440 --> 00:52:00,400
that reminds me of the old saying. Yeah, there's probably, that's it. Let's turn around.

811
00:52:02,080 --> 00:52:09,360
We're, we're, once the, once the leafflowers are the, whatever chainsaw, so whatever is going on

812
00:52:09,360 --> 00:52:15,920
here, start appearing. It's probably time to turn around. So, it's a great deal. It reminds me of

813
00:52:16,880 --> 00:52:21,200
the saying, you know, three people can keep a secret if two of them are dead. If you want to have

814
00:52:21,200 --> 00:52:25,600
an actual secure communication channel, don't talk to anyone ever. Yeah. It's the same idea, right?

815
00:52:28,800 --> 00:52:33,040
If you, if you can run it locally on your machine, you should run it locally on your machine. That's

816
00:52:33,840 --> 00:52:38,720
that's the rule. Yeah. But if you are sitting on that tension that you need

817
00:52:38,720 --> 00:52:45,920
to compute, but you cannot spend the things, you know, from TEEs are very interesting. Yeah.

818
00:52:47,680 --> 00:52:53,360
Because now companies or individuals that need to do that, but not spending the money on the

819
00:52:53,360 --> 00:53:03,440
infrastructure can do it with the proofs. Also, to be honest, TEEs are not perfect yet. There is,

820
00:53:04,400 --> 00:53:14,720
you know, a fragmentation in how the verification, the, it's a very new thing. Also,

821
00:53:14,720 --> 00:53:21,840
the site channel attacks, like basically the provider of the infrastructure can look inside,

822
00:53:21,840 --> 00:53:28,400
but it can refer things from the loads. Yeah. And from the patterns of memory access and so on.

823
00:53:29,280 --> 00:53:34,720
But I mean, that's nothing like looking inside or nothing like that. Yeah. Yeah. I mean,

824
00:53:34,720 --> 00:53:40,800
you'll always have these kind of things. I mean, again, also on the on the encrypted communication

825
00:53:40,800 --> 00:53:51,520
site, right? It's very, very hard to remove all signals, basically. Yeah. Yeah. Jeff also recently

826
00:53:51,520 --> 00:54:00,400
wrote about this on the white noise site. You, you know, like just by, I mean, we talked about this

827
00:54:00,400 --> 00:54:11,440
in the past, just by observing the key rotations. You can infer some group activity at least,

828
00:54:11,440 --> 00:54:16,800
you know, for a little while until the whole thing rotates again. So there's no way around that.

829
00:54:16,800 --> 00:54:23,280
I think that's, that's true also for. Yeah. It's, again, it's, it's trade-offs all the way.

830
00:54:23,280 --> 00:54:29,680
It's like, it's similar to compression encryption as well, where like it, it's, in the best case,

831
00:54:29,680 --> 00:54:35,200
the perfect encryption algorithm will look like random noise. Yeah. But, you know, that's,

832
00:54:35,200 --> 00:54:41,520
that's almost impossible to reach. Yeah. I mean, but it's a, the important part is that it's

833
00:54:42,000 --> 00:54:46,240
is an honest trade-off, right? Yeah. And also knowing, knowing what you get.

834
00:54:46,240 --> 00:54:56,320
Exactly. Everything is encrypted. The on memory is encrypted. But, of course, you can infer things

835
00:54:56,320 --> 00:55:01,920
from patterns and so on. That's a metadata that you can infer. Yeah. You can do a realistic

836
00:55:01,920 --> 00:55:09,120
Samsung, but it's the same for, for anything. Yeah. This is a, yeah. No solutions, only trade-offs.

837
00:55:09,120 --> 00:55:13,040
Exactly. There you go. No way the trade-offs will allow you to, to be things right.

838
00:55:15,280 --> 00:55:21,520
Yeah, hopefully. So, yeah, hopefully. But it's, no, I've been, I've been rocking my brain about,

839
00:55:21,520 --> 00:55:29,440
about that for the, for the last couple of weeks. It's, um, namely the, what we talked before,

840
00:55:29,440 --> 00:55:32,640
kind of, or what, what you mentioned, you know, oftentimes it's better to sleep on it and you

841
00:55:32,640 --> 00:55:37,760
shouldn't, you shouldn't just be working 24-7 all the time because you have to be really clear about

842
00:55:38,720 --> 00:55:45,680
your goal. Like, what are you working towards, basically? Yeah. You know, it's, um, it's,

843
00:55:45,680 --> 00:55:49,280
what do you want to actually achieve? What, which brings up the question,

844
00:55:51,040 --> 00:55:59,120
you know, it gets very, it gets very metaphysical, very quickly. Because most people in tech,

845
00:56:00,080 --> 00:56:05,520
I feel like, are working on the technology for technology's sake.

846
00:56:06,480 --> 00:56:11,280
Not necessarily for the betterment of humanity. But it's just kind of like, nerding out and stuff

847
00:56:11,280 --> 00:56:16,960
just because nerding out is cool. Yeah. And, and we, we see on an ostracite, for example,

848
00:56:19,760 --> 00:56:28,080
lots of people are an ostr, or have been an ostr. Because they were very disgruntled with the state

849
00:56:28,080 --> 00:56:33,360
of social media. Yeah. And wanted to get away from the algorithmic feeds, right? And kind of,

850
00:56:34,320 --> 00:56:41,760
wanted to get away from the outrage machine and being glued to, to the algorithm and catering to

851
00:56:41,760 --> 00:56:48,000
the algorithm and so on. Yeah. And I feel like a lot of people made that switch successfully. And

852
00:56:50,560 --> 00:56:57,840
the result is that they are on the phone way less. And that's also an ostr, like way less. Like,

853
00:56:57,840 --> 00:57:02,320
I see people maybe checking in like once a month or something. Yeah. You know, and just not using

854
00:57:02,320 --> 00:57:07,200
Noster that much, because it's not that addictive. Yeah. And so to me, that's interesting,

855
00:57:07,200 --> 00:57:13,680
because to me, that it feels like a success in some sense, you know? But you can't really measure it

856
00:57:13,680 --> 00:57:18,320
because you don't have any active users or anything like that. I mean, you can infer some stuff,

857
00:57:18,320 --> 00:57:23,040
but yeah. And similar to, like for me, for example, I always wanted to

858
00:57:24,160 --> 00:57:30,720
build stuff that's useful for me and like potentially also others. And right now, the way I

859
00:57:30,720 --> 00:57:36,560
develop apps and other things and so on is just kind of envisioning them and talking about them

860
00:57:36,560 --> 00:57:46,320
and then they exist. I never, I never really enjoyed the grant work that it took. And like,

861
00:57:46,320 --> 00:57:52,960
again, like I wasn't a programmer, like legit programmer for like 20 years plus. And because I

862
00:57:53,040 --> 00:58:03,520
enjoyed the, I enjoyed the outcome so much. I enjoyed the outcome so much. Like the final piece

863
00:58:03,520 --> 00:58:09,600
that the artifact that was created, I didn't necessarily enjoy the process of like actually sitting

864
00:58:09,600 --> 00:58:14,400
down and smashing your head against the problem and actually typing thousands and thousands and

865
00:58:14,400 --> 00:58:18,880
thousands of lines of code every single day, sitting, you know, in an office on the fluorescent

866
00:58:18,880 --> 00:58:27,120
light and, you know, having team meetings and whatever, you know, like scrum estimation games

867
00:58:27,120 --> 00:58:33,360
and whatever. I never enjoyed that. And I think I'm not alone with that. And so to me,

868
00:58:33,360 --> 00:58:40,560
kind of, again, I'm still trying to wrap my head around what does the post-scarcity world

869
00:58:40,560 --> 00:58:45,840
look like when it comes to software? I mean, we're in the process of finding out, but it's

870
00:58:48,960 --> 00:58:54,160
I don't know. It's all very strange because for me personally, but it's also my explicit goal.

871
00:58:54,160 --> 00:58:58,240
I want to sit in front of the computer less and less and less. If I can do the same stuff

872
00:58:58,960 --> 00:59:04,160
that I did when I was sitting in front of the computer like 12, 14 hours a day, if I can do the same

873
00:59:04,160 --> 00:59:11,440
stuff while being outside most of the day, just talking into a microphone basically, then I'm

874
00:59:11,440 --> 00:59:16,400
going to prefer the being outside part. And I'm still going, you know, I'm still, my, my head is still

875
00:59:16,400 --> 00:59:20,480
at work. I'm still working. I'm still thinking about the things. I'm still solving the same

876
00:59:20,480 --> 00:59:26,400
problems and so on. It's just kind of my compute interface is very, very, very different. And

877
00:59:28,160 --> 00:59:31,680
the outcome is still kind of the same. I don't know. I don't know where I'm going with all this. I'm

878
00:59:31,680 --> 00:59:38,240
just, I wonder, I wonder how you think about all these things because you're so close to the

879
00:59:38,240 --> 00:59:44,000
matter and you've been around software also for a long time. And you had your fingers in many

880
00:59:44,080 --> 00:59:49,600
different things. So, so how do you think about all that? I don't know. I think that we don't know

881
00:59:49,600 --> 00:59:57,280
yet where we are going or what is the outcome of all of this. And the only way to go through all

882
00:59:57,280 --> 01:00:03,040
this from my point of view is by doing it, by experiencing it, by realizing stuff, you know,

883
01:00:03,920 --> 01:00:11,360
because you kind of need to participate on that to basically give an answer to what those all

884
01:00:11,360 --> 01:00:17,840
of this means and what all all these new capacities means and what is the role of them,

885
01:00:17,840 --> 01:00:25,600
what the the end of software scarcity means. And so I think that there is no clear answer because

886
01:00:25,600 --> 01:00:32,080
we are using the in the beginning of all of this. So, the only way, the only same way for me to

887
01:00:32,080 --> 01:00:40,640
think about this is participating makes sense of it in the way that I can make sense,

888
01:00:41,440 --> 01:00:49,520
you know, for myself and sharing with other people and talking and, you know, contributing in some

889
01:00:49,520 --> 01:00:57,840
way. So, at the end at some point it will get clear, clear. But then now it's like a going blind to

890
01:00:58,800 --> 01:01:08,080
new territory. Yeah. To me, what's interesting, what just popped into my head while you were talking

891
01:01:08,080 --> 01:01:11,840
is kind of the, I mean, I definitely agree. I mean, it's the whole point why we do software

892
01:01:11,840 --> 01:01:20,080
engineering is to have this dialogue, right? And to to to kind of collectively figure it out,

893
01:01:20,080 --> 01:01:24,720
but not in the, you know, communist collective kind of sense, but in the distributed cognition,

894
01:01:24,720 --> 01:01:28,400
one brain is not enough to figure this out kind of sense, you know, like it doesn't make sense to

895
01:01:28,400 --> 01:01:34,800
what you what you what you meant with like we'll have to figure this out together and like

896
01:01:34,800 --> 01:01:40,080
continually talk about it. But also, again, what I care about is like, why are we doing all this,

897
01:01:40,080 --> 01:01:46,640
basically? It's like for me, it's still very clear on the on the sovereign on the self-sovereign

898
01:01:46,640 --> 01:01:53,440
technology side. Yeah. That's why I bring up DHH all the time because he hated Bitcoin.

899
01:01:53,440 --> 01:02:00,720
I mean, starting in like 2013 or something, like like he was an early hater. Yeah.

900
01:02:01,680 --> 01:02:06,400
And hated it for like 10 years or so. Then the Canadian trucker protests happened and he kind

901
01:02:06,400 --> 01:02:12,080
of flipped. He was like, oh, okay, now I get Bitcoin and why it's important. And and and similar,

902
01:02:12,080 --> 01:02:17,600
like all the all the free software people, I mean, installman, you know, as the main example,

903
01:02:17,600 --> 01:02:24,240
he understood extremely early on that a client's armor relationship is a master slave relationship.

904
01:02:24,240 --> 01:02:30,560
Yeah. And he was not okay with being a slave. And so that's why he started the whole movement and

905
01:02:30,560 --> 01:02:34,880
you know, started rewriting all the things and like went really hard on free software. Yeah.

906
01:02:34,880 --> 01:02:41,600
And and you know, like God bless him. Like we wouldn't have anything without someone as radical

907
01:02:41,840 --> 01:02:53,440
as he was and still is. And so so to me, kind of the whole point of working on all this stuff is

908
01:02:54,240 --> 01:03:03,360
to make sure that the people that are less capable technologically have access to technology that's

909
01:03:03,360 --> 01:03:10,800
actually self-sovereign technology, which is like anti-slave technology. Yeah. And and

910
01:03:11,200 --> 01:03:16,160
I mean, in some sense, because the other edge of this is that you need to run models that are

911
01:03:16,800 --> 01:03:24,480
hosted by some entities and probably not run by you. But that's a different edge. Sure, sure,

912
01:03:24,480 --> 01:03:30,080
I mean, but that's also like local models get better and smaller all the time and so on. It's

913
01:03:33,680 --> 01:03:39,600
there's many dimensions to all that it's like who is in charge of the communication infrastructure?

914
01:03:39,600 --> 01:03:43,440
Where are all your files hosted and so on? Most people don't host everything themselves.

915
01:03:43,440 --> 01:03:49,040
Yeah. And so on. But but that's again, like coming back to what you said before, like you have

916
01:03:49,040 --> 01:03:53,520
encryption at rest, you have encryption while in flight, you know? Yeah.

917
01:03:53,520 --> 01:03:58,960
So encrypt the communications. And now the last piece of the puzzle, one next piece of the

918
01:03:58,960 --> 01:04:06,400
puzzle is basically encrypt the compute, like big square square quotes, like confidential

919
01:04:06,400 --> 01:04:11,920
compute with the station sensor. And and and so I think we're we're we're getting there and I think

920
01:04:11,920 --> 01:04:19,600
it's all about kind of what what works me when people talk about this stuff is always they talk

921
01:04:19,600 --> 01:04:25,040
about ownership. But for me, ownership in the in the information realm makes very little sense.

922
01:04:25,040 --> 01:04:29,520
That's why I always hated NFTs because you can't own a chain pack. It's like I can right click

923
01:04:29,520 --> 01:04:33,280
save it. You're not going to own it. You know, it's just like you can you can copy information

924
01:04:33,280 --> 01:04:38,560
perfectly. So talking about ownership, it's nonsensical. It's similar to talking about ownership

925
01:04:38,560 --> 01:04:42,640
of numbers. And of course, that gets tricky when you talk about ownership of Bitcoin because Bitcoin

926
01:04:42,640 --> 01:04:47,680
is just like your private key. Yeah. I can you own Bitcoin? I don't know. Can you own the number three?

927
01:04:47,680 --> 01:04:52,880
Like who cares? You know, it's all about control and it's about the question of who is in control

928
01:04:52,880 --> 01:04:58,880
and who can control you. And if if the answer is like you are in control and you're solely in control,

929
01:04:58,880 --> 01:05:02,880
of course, there's a downside. You have maximum responsibility. If you lose your private key,

930
01:05:02,880 --> 01:05:08,960
you're wrecked. Yeah. So all those kind of things. But I'd rather have a world where people are

931
01:05:09,760 --> 01:05:16,160
responsible. You know, like where we we we force this responsibility responsibility upon people.

932
01:05:16,160 --> 01:05:23,040
So yeah. And and people actually do become competent if they have to. Like for example,

933
01:05:23,040 --> 01:05:27,040
people can drive cars, you know, like you have to be very competent to drive the car. To that

934
01:05:27,040 --> 01:05:32,960
point, for me, responsibility means freedom, you know? Yeah. Absolutely. And it's also like

935
01:05:34,160 --> 01:05:40,000
Nietzsche, Nietzsche said the first word of freedom is no. And that's kind of what

936
01:05:41,680 --> 01:05:46,480
like again, like why why I continue to work on this stuff because it's like no, I'm not going to use

937
01:05:47,040 --> 01:05:53,440
whatever, you know, you're fucking Google Cloud or fucking past keys, man. Everyone wants to

938
01:05:53,520 --> 01:06:02,960
sell me a pass key. But like how can you how can you set up an environment and and and what kind of

939
01:06:02,960 --> 01:06:10,960
tools are necessary inside a space that you actually can say no and you know means something. And

940
01:06:10,960 --> 01:06:16,160
you don't have to exit from society completely. Yeah. Basically, I mean, that's that's that's how I

941
01:06:16,160 --> 01:06:20,960
would sum it up. The work we're doing on the Bitcoin side on the other side. Now also on the AI

942
01:06:21,040 --> 01:06:27,760
and and computation side and so on. So to to kind of finish my train of thought, that's where the

943
01:06:28,720 --> 01:06:34,000
blessing comes in of the of the agentic age, right? Like we can actually we actually have a chance now

944
01:06:35,360 --> 01:06:41,760
more than a chance. Like before, before it was like, you know, like a snowballs chance in hell,

945
01:06:41,760 --> 01:06:48,640
you know, like before it was really hard. Yeah. Now with all the compute and inference on all the

946
01:06:48,640 --> 01:06:55,840
models that we have, you can actually build stuff really well, really quickly on like a large scale.

947
01:06:55,840 --> 01:07:03,120
You can build freedom tech. Like again, I think I don't know, a hundred X faster. Yeah,

948
01:07:03,920 --> 01:07:10,240
whatever. It depends on how you look at it. Yeah. And for me, and I think I said good way to

949
01:07:10,240 --> 01:07:19,680
continue that. For me, in general, it also means something very related with arts, you know,

950
01:07:19,680 --> 01:07:26,320
with the with the process of creation. In arts, for my point of view, it's a way to make sense of

951
01:07:26,320 --> 01:07:34,160
the work. Yeah. And kind of express what you have inside. And also, you can see that in all the

952
01:07:34,160 --> 01:07:40,800
arts movement that we have had, all them have been a kind of expression of their times.

953
01:07:40,800 --> 01:07:45,520
Oh, absolutely. I mean, look at postmodern art. Exactly. No, no, all of them. All of them.

954
01:07:48,240 --> 01:07:54,080
I think that, you know, engineering is arts as well. It's an art craft at least.

955
01:07:55,040 --> 01:08:00,160
And you can make sense of the work through it. Like the somewhere that I make, the things that I do,

956
01:08:01,120 --> 01:08:06,000
I do it because I believe that they have, they have some meaning or they have, they mean something

957
01:08:06,000 --> 01:08:13,600
and it's a way to make sense of their work for me. Because I see that there are some, you know,

958
01:08:13,600 --> 01:08:22,160
stuff that doesn't work as it should work. We should have options and be able to do this without

959
01:08:22,160 --> 01:08:27,040
needing to do this other thing to complain. So would you say it's a form of protest as well?

960
01:08:27,040 --> 01:08:34,880
It's a form of protest, like art. Yeah. Same. But now I think that with this democratization of

961
01:08:34,880 --> 01:08:42,720
the software and all these new tools that we have, it's like a lowering the bar for new people to

962
01:08:42,720 --> 01:08:49,840
express themselves with the software that they can create. It's like a, imagine a Leonardo da Vinci

963
01:08:49,920 --> 01:08:58,240
that never had access to create the stuff because it's so poor or the conditions. Yeah, exactly.

964
01:08:58,960 --> 01:09:06,720
It has the ability to access to, I don't know, tools or to create stuff. That could be a waste,

965
01:09:06,720 --> 01:09:14,160
right? Like a person with that talent to not be able to express it or to canalize that to create

966
01:09:14,160 --> 01:09:21,840
something. And now I think that we are going through that point that software is so

967
01:09:21,840 --> 01:09:28,560
cheap that people can create stuff and can express themselves. Of course, a lot of it is going to be

968
01:09:28,560 --> 01:09:34,720
shit. It's going to be crap. But other people that never thought about being able to create

969
01:09:34,720 --> 01:09:42,080
nothing or to be able to create something in that sense, they are going to be able to do it.

970
01:09:42,080 --> 01:09:48,000
Or at least to express it. Yeah. Other people can take over it or contribute or whatever.

971
01:09:48,720 --> 01:09:51,920
We definitely see that right now, lots of people are creating games, for example,

972
01:09:51,920 --> 01:09:56,000
that were previously unable to create games because software development was actually hard.

973
01:09:56,800 --> 01:10:00,800
And so there's definitely something to that that kind of,

974
01:10:02,000 --> 01:10:10,400
yeah, the amount of artifacts that encapsulate human expression is just going to increase by

975
01:10:10,400 --> 01:10:16,480
whatever factor, like, you know, a thousand years or more. And I think that that's amazing,

976
01:10:16,480 --> 01:10:21,440
man. That's amazing. I'm missing time. It's like a renaissance of creation in that sense,

977
01:10:21,440 --> 01:10:27,040
in the digital world. And right now, everything is becoming more and more digital. You like it or not.

978
01:10:28,480 --> 01:10:34,880
Be able to express yourself in that world without needing to be an expert and study for

979
01:10:34,880 --> 01:10:41,760
years and years and hit your head against the wall, creating code or whatever. Now it's

980
01:10:42,560 --> 01:10:48,400
not totally banished, but it's much more accessible. And if you want, you can.

981
01:10:49,200 --> 01:10:56,640
And if you hit a wall, you can learn by yourself more easily or ask anyone and it's going to be

982
01:10:56,640 --> 01:11:04,080
more easy for you to learn from. Oh, yeah. I mean, right now, like, it's basically all you need is

983
01:11:05,600 --> 01:11:13,600
like, imagination, intention, agency. It's there is no excuse anymore in terms of,

984
01:11:13,600 --> 01:11:18,320
oh, I don't know how to do this or anything like that. Like, just, just, just talk, just ask.

985
01:11:18,880 --> 01:11:22,560
Like, ask and you shall receive, right? Like knock at the door shall open.

986
01:11:24,000 --> 01:11:31,120
And if you really make something valuable with that, probably if it's not perfect,

987
01:11:31,120 --> 01:11:36,000
it doesn't matter. It's because people can contribute to it. That is a, that's the

988
01:11:36,000 --> 01:11:41,680
beauty of open sources, right? You can put things out there and help people to contribute.

989
01:11:41,680 --> 01:11:45,680
Yeah. And I mean, that's another, that's another interesting side effect right now that that,

990
01:11:45,680 --> 01:11:50,160
I think this will pass as well, but right now, I mean, lots of open source projects close down

991
01:11:50,160 --> 01:11:55,200
that are, they're open doors, basically, because they just got hit with so many drive-by,

992
01:11:55,200 --> 01:12:01,040
slop pull requests and like, you know, whatever bug reports that's on or created by agents, of

993
01:12:01,040 --> 01:12:06,160
course, that they just had to, they got overwhelmed on the maintenance side and had to close down.

994
01:12:06,160 --> 01:12:10,880
I think we'll figure, again, like I said before, agents are really good at maintaining,

995
01:12:10,880 --> 01:12:16,480
so you just have to fight fire with fire. Yeah. And use your own slop machinery to just close all

996
01:12:16,480 --> 01:12:22,240
the PRs. Okay, kind of. I think that we're not there yet. No, we're not there yet, but we'll

997
01:12:22,320 --> 01:12:27,120
figure that out. In the sense that, you know, when you have to maintain or you have to solve a

998
01:12:27,120 --> 01:12:35,600
problem, meta-problems can appear around and yeah, like agents doesn't have the ability to really

999
01:12:35,600 --> 01:12:40,160
cut through the, the fats. Oh, yeah, no, they're terrible at judgment.

1000
01:12:40,720 --> 01:12:46,080
Still, like, if anyone ever listened to Reveke, which I like every, every single time I talk to

1001
01:12:46,080 --> 01:12:50,720
someone, I will bring up Reveke a million times. That's a key. I think it'll listen to

1002
01:12:50,800 --> 01:12:57,840
Reveke and what he has to say about, like, why, then you'll understand why Alalams will

1003
01:12:58,640 --> 01:13:05,760
never have the human judgment and finesse and and and taste basically. Yeah, it's like, I mean,

1004
01:13:06,560 --> 01:13:12,400
I had lots of cryptic posts and I started out just about tastes and values. I mean, there are some

1005
01:13:12,400 --> 01:13:19,200
humans with really bad taste as well. Oh, of course. But I mean, slop comes from us, make no mistake.

1006
01:13:20,880 --> 01:13:24,480
It's just the middle of the bell curve. That's the that's the issue, right? It's like,

1007
01:13:24,480 --> 01:13:30,640
what we define as slop, what we talk about as slop is the middle of the bell curve by definition.

1008
01:13:30,640 --> 01:13:34,560
So you just take the like, take all the fucking JavaScript code in the world,

1009
01:13:38,880 --> 01:13:43,120
build the bell curve, take the middle, all that slop and it's all human generated.

1010
01:13:43,920 --> 01:13:48,880
And that's what the Alams will spit out. Same for art, same for prose, like for

1011
01:13:49,760 --> 01:13:55,680
for like write an article on whatever, you know, like the Pebble Beach. And it's just going to be

1012
01:13:57,200 --> 01:14:02,560
yeah, shit, because it's going to be mediocre. And by definition, that's what the middle of the

1013
01:14:02,560 --> 01:14:08,800
bell curve is. It's like, no, it's very unremarkable, very kind of, yeah, bad taste.

1014
01:14:09,120 --> 01:14:17,680
Yeah. But I think that all of this is, I mean, you can speculate about all of this is going and

1015
01:14:17,680 --> 01:14:23,200
and what is the end of all of this. But I think that's one of the keys of of, I don't know,

1016
01:14:23,200 --> 01:14:31,040
that maybe allow us to to see a bit further. Not not totally conclusive, but you know,

1017
01:14:31,360 --> 01:14:41,680
people being able to use, develop new keys by talking or by writing without needing to be an

1018
01:14:41,680 --> 01:14:49,040
expert. That's going to be one of the key factors to hold all of this is going to be life in the

1019
01:14:49,040 --> 01:14:57,440
future. I like what you said before. And when you talked about art and it being a process

1020
01:14:57,440 --> 01:15:01,600
basically of yeah, not only expressing yourself, but making sense of the world,

1021
01:15:01,600 --> 01:15:05,360
yeah, completely agree with Sarah. I mean, the whole reason why I write, for example, is to make

1022
01:15:05,360 --> 01:15:09,840
sense of the world. And kind of not only the world, it's like to make sense of my own thoughts.

1023
01:15:09,840 --> 01:15:14,240
Exactly. And I think for art, it's the same. I mean, I did a lot of photography back in the day,

1024
01:15:14,240 --> 01:15:22,640
and it's kind of kind of like self self expression, basically, is not only

1025
01:15:22,720 --> 01:15:29,440
telling others what you think and feel and so on, but it's, I think for me, first and foremost,

1026
01:15:29,440 --> 01:15:36,480
10 to yourself. Exactly, like realising it, it's a process of self-realisation, basically.

1027
01:15:36,480 --> 01:15:42,880
It's like, I didn't even realise that I thought that, but I sat down for like 50 hours and tried

1028
01:15:42,880 --> 01:15:48,800
to write about it, and that made me realise that I actually think that. Or whatever, you know,

1029
01:15:48,880 --> 01:15:55,840
like if it's some other art form, then you paint something, or what have you. And you're surprised

1030
01:15:55,840 --> 01:16:00,080
of what comes out, and it's like, oh my god, I didn't even know that I had this in me. That's

1031
01:16:00,080 --> 01:16:04,320
why I love these conversations so much, because every single time you have no fucking idea where

1032
01:16:04,320 --> 01:16:09,040
the conversation is going. Never ever. And I mean, we talk about this in the summer engineering

1033
01:16:09,040 --> 01:16:15,360
cohorts all the time. The whole point is to enter into dialogue with each other and

1034
01:16:15,360 --> 01:16:20,480
with the technology itself, and you never know where it's going to go. You never know kind of what

1035
01:16:20,480 --> 01:16:25,920
is the, like there is no goal, just like with a proper conversation. There is no, there is no,

1036
01:16:26,880 --> 01:16:31,120
there is no entry, there is no goal, there is no like, again, like Alan Watts talked about

1037
01:16:31,120 --> 01:16:36,640
this beautifully, about the dance, the point of the dance is not to go from A to B, the point

1038
01:16:36,640 --> 01:16:43,120
of the dance is to dance, and to enjoy the process of dancing. And like, you know, the point of

1039
01:16:43,120 --> 01:16:47,040
listening to a piece of music is not to go to the end, otherwise you would just fast forward,

1040
01:16:47,040 --> 01:16:53,360
you know, and all those kind of things. That's the way I think about, again, like this

1041
01:16:53,360 --> 01:16:58,880
dialogical development structure that we've set up, and also how I think about these dialogues,

1042
01:16:58,880 --> 01:17:02,720
but I want to look back to one of the things you said before that everything becomes more digital.

1043
01:17:03,600 --> 01:17:10,080
I kind of agree, but I also, from the interface point of view, what you said before as well,

1044
01:17:10,080 --> 01:17:16,240
like we started with the CLI, right? Now we're back to CLI, it's basically. And the most DHH

1045
01:17:16,240 --> 01:17:21,440
actually mentioned this as well, like he talked about the malleable computer. When you got,

1046
01:17:21,440 --> 01:17:25,600
when you got the first Commodore and those kind of things, you know, like you had basic on it,

1047
01:17:25,600 --> 01:17:30,960
that there was it, you just had a terminal, and you had to program it yourself in some sense,

1048
01:17:30,960 --> 01:17:39,760
you had to program the applications yourself. And I feel like, if you look at the computer,

1049
01:17:40,720 --> 01:17:46,160
and the internet services, the service infrastructure that we have, it's mostly kind of a means

1050
01:17:46,160 --> 01:17:51,680
to an end. I want to get something. I want to, whatever, if I use a word processor,

1051
01:17:52,960 --> 01:17:58,480
if I use a photography system, whatever, I want to put something in, I want to get it back out

1052
01:17:58,480 --> 01:18:07,040
later. It's a very simple thing. If I use Uber or DoorDash or whatever, like I want to get from A

1053
01:18:07,040 --> 01:18:10,880
to B, like I want to order a cab or I want to have a hotdog delivered or what have you, you know.

1054
01:18:11,440 --> 01:18:20,640
And the interface, the way I see, the way I look at it right now is I think the interfaces will

1055
01:18:20,640 --> 01:18:27,440
almost completely disappear in some sense. And I think that's also true, like for me, it's definitely

1056
01:18:27,440 --> 01:18:33,040
true on the coding side, for example, like programming languages, for example, for me,

1057
01:18:33,040 --> 01:18:40,400
completely disappeared. Half the apps I pushed out, I don't even know what it's written in,

1058
01:18:40,400 --> 01:18:44,560
because I don't care at all. It might be rust, it might be cuddling, I don't know, I don't care.

1059
01:18:44,560 --> 01:18:50,320
I don't care. It's like, use whatever you think is best and just build it. And I think that's

1060
01:18:53,840 --> 01:18:59,040
where I want to go with that is like, yes, I agree, everything is going to probably be more digital

1061
01:18:59,040 --> 01:19:01,680
in some way, but it's going to be way more invisible also.

1062
01:19:02,640 --> 01:19:10,800
Yeah, my point in this with this progression of the interface from the command line to the

1063
01:19:10,800 --> 01:19:16,800
natural language is basically like the shape, right, like the commands that you write in a

1064
01:19:16,800 --> 01:19:27,040
terminal, they are a phrase. Yeah, you can read them as a phrase, you can do specify what you

1065
01:19:27,040 --> 01:19:33,280
want to express in that phrase and you execute. That's difficult because you need to learn the grammar.

1066
01:19:33,920 --> 01:19:40,640
Right, like as you need to learn the rules, like how to hold you, join the different words,

1067
01:19:41,280 --> 01:19:47,040
the values, what notes. You forget one semi-colon and you're in hell. Exactly, and that's

1068
01:19:47,040 --> 01:19:53,200
difficult because it's forced you to learn a new language. Even if the language is trying to be

1069
01:19:53,200 --> 01:19:58,080
similar to what you use. Yeah, sure, sure, sure. That's why we have go to and it's ugly.

1070
01:19:59,280 --> 01:20:06,640
But the change, even if the interface is still literal because we use words for that,

1071
01:20:06,640 --> 01:20:11,360
is that you don't need to learn the new grammar. You can use your grammar, use your way to express

1072
01:20:11,360 --> 01:20:17,840
yourself and execute the stuff. Yeah, and I think that's the key difference, right? That's where the

1073
01:20:18,800 --> 01:20:26,480
difficulties of the data were, disappears. Yeah, because now you don't need to learn these rules,

1074
01:20:27,040 --> 01:20:30,720
the artificial data and so on. I just had an insight, I'm just getting goosebumps.

1075
01:20:32,000 --> 01:20:35,840
Because I think you talked about the interface and I talked about this in one of the

1076
01:20:35,840 --> 01:20:41,840
summer engineering presentations. You are interfacing with the machine, right? Like the machine has

1077
01:20:41,840 --> 01:20:47,200
a face and you're interfacing with it. And so you need to face it, you need to face the machine,

1078
01:20:47,200 --> 01:20:50,560
you need to face the program. That's how you talk to it. That's how you talk to people as well.

1079
01:20:50,560 --> 01:20:59,920
You face them and then a conversation starts happening and so on. And because the interfaces

1080
01:20:59,920 --> 01:21:06,720
so crude, it's like it's up to now we used the language of the computer to talk to the computer.

1081
01:21:06,720 --> 01:21:13,120
And that's flipping right now. And so we are moving from having, we're moving from

1082
01:21:14,640 --> 01:21:19,280
interface to dialogue is what I'm trying to say. At least that's what happened to me. That's

1083
01:21:19,280 --> 01:21:25,280
what happened to me over the last, I don't know, like 12 months or so. I'm not really interfacing

1084
01:21:25,280 --> 01:21:29,520
with my compute anymore. I'm not really interfacing with my computers or my apps and my projects and

1085
01:21:30,400 --> 01:21:34,240
and so on anymore. I mean, I still am. I'm still sitting in front of the laptop. I don't know,

1086
01:21:34,400 --> 01:21:44,480
whatever, like 6 to 10 hours a day. But it's, I can feel the shift and Johnathan Pajor

1087
01:21:44,480 --> 01:21:49,200
says something beautifully recently where he's like, I can tell a seed when I see it. If you

1088
01:21:49,200 --> 01:21:56,000
put a seed into a pile of dung, it's still 99.9 percent dung. But you know what's going to happen.

1089
01:21:56,000 --> 01:22:00,640
And he can tell where it's going. Even though the seed is very small. And I think what's going to

1090
01:22:00,640 --> 01:22:09,600
happen is that having a dialogue with your digital infrastructure is going to be extremely

1091
01:22:09,600 --> 01:22:16,320
normal. And looking at an interface is going to be the exception. Interfacing with it in the

1092
01:22:16,320 --> 01:22:22,560
traditional GUI interface way is going to be the exception. And I think it will also have mostly

1093
01:22:22,560 --> 01:22:27,680
like ad hoc interfaces. I already do this. It's like, you know, generate an infographic of like

1094
01:22:27,680 --> 01:22:32,240
whatever you did in the last 6 hours. And then just look at it. Pablo did this as well with

1095
01:22:32,240 --> 01:22:40,160
podcasts. He basically told these agents, generate a podcast for me like a 45 minute update that I

1096
01:22:40,160 --> 01:22:48,640
can listen to while I go on and around. And yeah, I think we're, those that are on the frontier,

1097
01:22:48,640 --> 01:22:55,680
they made the switch already where there are no interfaces anymore really to speak of. It's all

1098
01:22:55,680 --> 01:23:00,480
kind of, it's all dialogue. It's all dialogical. Yeah. I mean, I think that that was always the

1099
01:23:00,480 --> 01:23:06,640
point, right? Like of computers and so on. Like I make the interfacing with it closer to

1100
01:23:06,640 --> 01:23:10,960
to your experience. It was definitely the dream, right? I mean, that's what Star Trek can't

1101
01:23:10,960 --> 01:23:15,520
space all this day and everything. Yeah, I mean, you at the beginning, like a computer as well,

1102
01:23:15,520 --> 01:23:20,960
something very aesthetic in the way that there are new machines and no one knows how they work.

1103
01:23:20,960 --> 01:23:27,840
Just people that can operate it were like scientific suits. And you know, and so on.

1104
01:23:28,480 --> 01:23:33,840
I mean, before it was machines, it was an actual profession. Like it's like, what do you do?

1105
01:23:33,840 --> 01:23:38,000
I'm an accountant. Oh, nice to meet you. I'm a computer, you know? Like I sit in a big building

1106
01:23:38,000 --> 01:23:43,200
and I crunch numbers with 100 other people. But even even in that time, like the interface for it

1107
01:23:44,000 --> 01:23:49,680
was trying to imitate the email language, right? Like I use in these phrases, commands and so on.

1108
01:23:50,560 --> 01:23:55,440
Then the visual part was a kind of interlude of, I don't know,

1109
01:23:55,440 --> 01:24:00,800
making it easier to understand for us humans. Organize visually stuff. Yeah. But

1110
01:24:02,080 --> 01:24:10,080
you know, learn how computer works and how to express something in the language,

1111
01:24:10,080 --> 01:24:16,640
so in binary or whatever assembly language that you want to use. Something always mechanical

1112
01:24:16,720 --> 01:24:21,280
and not supposed to be used by humans, right? Because we don't have this ability to

1113
01:24:22,240 --> 01:24:27,760
express things in cities. Only the most autistic people have the ability. And not even them,

1114
01:24:27,760 --> 01:24:33,280
like they have bad days as well. The guy who built rollercoaster tycoon,

1115
01:24:34,400 --> 01:24:39,120
you know, like 80% assembly or so. He probably dreamt in assembly.

1116
01:24:39,120 --> 01:24:43,040
Oh yeah, probably. But other days maybe, I don't know, it would be

1117
01:24:44,000 --> 01:24:47,440
I don't know, a couple world doing nothing of that because they are

1118
01:24:48,080 --> 01:24:52,960
or something happening in their life. And you cannot express yourself in cities and ones.

1119
01:24:52,960 --> 01:24:57,520
All the time, it's mechanical. It's mechanical. You can try to be mechanical for some

1120
01:24:57,520 --> 01:25:02,960
period. But at the end, you are something very different that a mechanical entity.

1121
01:25:02,960 --> 01:25:10,480
So to paraphrase that point is like the original dream of computers was that they are

1122
01:25:11,360 --> 01:25:18,240
human-like and able to understand human and also speak, like understand and respond to in

1123
01:25:18,240 --> 01:25:23,600
human language. But not only that was the trope on the Star Trek side, right? Like data,

1124
01:25:24,160 --> 01:25:29,200
data is, you know, data, like the guy who never saw a ray of sun. That's like,

1125
01:25:32,320 --> 01:25:39,280
he's the android robot of Star Trek. And he's like super logical, doesn't have any humor.

1126
01:25:40,080 --> 01:25:44,880
Like he works like a computer, like a computer of like the 70s and 80s, you know, like very,

1127
01:25:45,600 --> 01:25:50,240
that's how we thought about computers. And I think that that's changing drastically with,

1128
01:25:50,240 --> 01:25:56,320
like now, now computers actually have prefrontal cortex of basically, like an LLM is like a

1129
01:25:56,320 --> 01:26:02,400
prefrontal cortex. And it can actually, it can actually understand things. It's like, which is

1130
01:26:02,400 --> 01:26:09,600
amazing. You know, it's still disconnected from the world, like to make Barakis point once more.

1131
01:26:09,600 --> 01:26:15,200
You know, it's like, it's twice removed from the world. It's actually not in touch with reality.

1132
01:26:15,200 --> 01:26:23,840
Yeah. It's not even in touch with the corpus of like the human-produced corpus that we wrote

1133
01:26:23,840 --> 01:26:29,920
about reality. You know, like what we do is like, we write about our view of reality. And

1134
01:26:29,920 --> 01:26:37,520
LLM's aren't even in touch with that. Yeah. LLM's basically extract, extract to use a

1135
01:26:37,520 --> 01:26:41,840
Petersonian phrase that the map of meaning that is in that corpus, and they're in touch with

1136
01:26:41,840 --> 01:26:48,080
this matrix basically. And so they're twice removed. And Barakis point is that

1137
01:26:50,720 --> 01:26:58,800
to have actually deep understanding, to actually care, his point is mostly about caring,

1138
01:26:58,880 --> 01:27:04,880
to actually care about the information that you're processing. Yeah. You have to be embodied.

1139
01:27:04,880 --> 01:27:10,720
And you have to be fully in touch with the world. Yeah. And you also have to be in touch with

1140
01:27:10,720 --> 01:27:20,240
something that is bigger than you. So, so, some, some decentering needs to happen that shows you

1141
01:27:20,240 --> 01:27:24,880
that the world and the universe is not centered on you. And there's, there's things that are bigger

1142
01:27:24,880 --> 01:27:31,440
than you are. Yeah. And that's why like one of the examples he gives is like, I have a kid,

1143
01:27:31,440 --> 01:27:35,840
you know, like if you have a kid, like that happens automatically. Yeah. And the other examples he

1144
01:27:35,840 --> 01:27:39,920
gives is like, if you care about the transcendental values of like truth and goodness and beauty,

1145
01:27:39,920 --> 01:27:44,560
and so on, that's another thing that can do it. So you will basically have a relationship with

1146
01:27:44,560 --> 01:27:48,800
something transcendental that will de-center you. Yeah. Like, like, that will show you that there

1147
01:27:48,800 --> 01:27:52,400
is something bigger than you are. Yeah. I mean, again, that's why people both cathedral,

1148
01:27:52,400 --> 01:27:59,200
sense of, you know, like, yeah. Yeah. If the cathedral takes like 800 years to build, like,

1149
01:27:59,200 --> 01:28:05,440
you're working on something that's bigger than you. And so, yeah, a lot of us are not the end

1150
01:28:05,440 --> 01:28:10,080
of the, like, I'm like, I talked with Kelly about this as well with like robotics and world models

1151
01:28:10,080 --> 01:28:19,760
and those kind of things will all come. But it's still interesting to me that it's so different

1152
01:28:19,920 --> 01:28:24,640
than like the world, the world we are, the world we are living in right now.

1153
01:28:27,920 --> 01:28:33,760
It's so different than it was five or six years ago, because five or six years ago,

1154
01:28:34,320 --> 01:28:39,440
every single computer was still like data in Star Trek, right? Like it was still,

1155
01:28:39,440 --> 01:28:44,800
didn't have humor, like all the chatbots were crap, you know, like everything. You could,

1156
01:28:44,800 --> 01:28:51,680
I don't know, I don't even know what I'm trying to say. Like, there was no human as a no. Yeah.

1157
01:28:53,520 --> 01:29:03,280
And I think that's a anthropomorphizing the, the LNs and what they are. I think that can be our

1158
01:29:03,840 --> 01:29:10,400
biggest mistake in the future, to be honest, because when we make it closer to what we are or to

1159
01:29:10,400 --> 01:29:16,560
humans, and we do this a lot, right? Like I used to try to, everything can be anthropomorphized,

1160
01:29:16,560 --> 01:29:25,280
like this fucking rock can be my friend, and I can play in that kind of way, right? With the Wilson

1161
01:29:25,280 --> 01:29:33,360
family, vulnerable. But I think that that can be a huge mistake in the sense that we can

1162
01:29:33,360 --> 01:29:39,040
confuse ourselves with them. And we instead of becoming more human, we become more mechanical

1163
01:29:39,040 --> 01:29:47,200
in the sense that instead of, I don't know, using them as a tool, we become part of the tool.

1164
01:29:47,200 --> 01:29:52,480
Yeah. And we are like a biological robot. Yeah. And that's something that scares me.

1165
01:29:52,480 --> 01:29:57,040
Some people have that worry also when it comes to writing and self-expression and so on.

1166
01:29:57,040 --> 01:30:01,680
It's like if you write like an LLM, you will self-center because people who think an LLM wrote it.

1167
01:30:01,680 --> 01:30:08,240
Yeah. That's kind of ideas. Yeah. But what you just said about my hope is that the opposite

1168
01:30:08,240 --> 01:30:12,400
will happen, and that's what I actually wanted to touch on when I mentioned that everything will

1169
01:30:12,400 --> 01:30:19,040
become more digital idea, because I think actually the opposite can happen as well, where like it's

1170
01:30:19,040 --> 01:30:27,200
definitely happening to me. I'm outside way more. Yeah. I'm living in the real world, so to speak.

1171
01:30:27,840 --> 01:30:32,080
I never liked real world by the way. You have listened to a lot of hours of Barbika, right?

1172
01:30:33,040 --> 01:30:39,200
That's true. No, no, that's true. I'm not sure what the cause of that is in some sense, but

1173
01:30:43,360 --> 01:30:49,360
not everyone have the same mental health. No, no, no. I know that that's true. I'm trying to

1174
01:30:50,640 --> 01:30:59,280
be optimistic about the future, because if you're steeped in internet meme culture,

1175
01:30:59,840 --> 01:31:05,440
you will know, or also like that internet here is also a thing. But everyone knows that everything

1176
01:31:05,440 --> 01:31:12,560
online is fake and gay. That's basically a very old internet meme that points to something

1177
01:31:15,840 --> 01:31:23,680
that I think is kind of true, where the same used to be Twitter isn't real life.

1178
01:31:24,560 --> 01:31:33,280
You know? And again, it memes survive because they have a certain truth embedded in them.

1179
01:31:33,840 --> 01:31:40,000
And Twitter, not being real life, is a very ancient meme in internet terms.

1180
01:31:40,880 --> 01:31:49,840
And I think it is true in the sense that people do understand that the stuff that's happening

1181
01:31:49,840 --> 01:31:56,160
in cyberspace is a different level of reality than the stuff that's happening in meme space.

1182
01:31:56,160 --> 01:32:06,720
I don't know, man. No? I don't know, because when specific technology like LLM are

1183
01:32:07,760 --> 01:32:14,160
that are very powerful and so new and they are drastically changing everything and penetrating

1184
01:32:14,240 --> 01:32:22,960
everyone's life. Normal people tend to misify the technology and to give it

1185
01:32:24,240 --> 01:32:33,120
mystical properties, right? And that means that they are going to maybe even think if they can

1186
01:32:33,840 --> 01:32:40,880
do things like a totally human to love something. Yeah, yeah, sure. I mean people married

1187
01:32:40,960 --> 01:32:49,760
LLM's. Sorry. I think that I don't know, it's very optimistic to say that people are realizing

1188
01:32:51,040 --> 01:32:56,640
the current state of the world and where are we going? Because I think that it's not the same

1189
01:32:56,640 --> 01:33:01,280
as totally the opposite. Fair enough. Fair amount of people are totally confused about what

1190
01:33:01,280 --> 01:33:09,280
this technology is. And when I talk with someone that doesn't, it's not this in touch with the

1191
01:33:09,280 --> 01:33:14,080
technology and with this world. Yeah, that seems to live in a different time line, you know?

1192
01:33:15,120 --> 01:33:22,800
Fair enough. No, fair enough. I do agree with all that. I'm kind of, I guess I'm talking about

1193
01:33:24,720 --> 01:33:30,960
my generation or even more narrow than that, like mostly myself, how I experience it, and

1194
01:33:31,680 --> 01:33:36,000
it used to be for me and my friends basically.

1195
01:33:39,280 --> 01:33:44,320
Meeting in person was stupid. You know, it was like, no, like let's just jump into mumble or

1196
01:33:44,320 --> 01:33:48,400
team speak or whatever. It's like, why would we go through the trouble of meeting in person when

1197
01:33:48,400 --> 01:33:54,560
we just can meet online? No, but seriously, I mean, I come from the gaming world. I come from

1198
01:33:54,560 --> 01:34:01,920
online communities. Like, why would you go through the trouble of organizing real life? Again,

1199
01:34:01,920 --> 01:34:06,160
I don't like the real life fake life distinction. To me, both things are real, but that's why I always

1200
01:34:06,160 --> 01:34:11,520
use cyberspace, meet space. But why should we go through the trouble of, again, back to names,

1201
01:34:11,520 --> 01:34:18,000
and the importance of names? Why should we go through the trouble of organizing a meet space

1202
01:34:18,000 --> 01:34:22,560
event for 25 people if we can just jump into team speak or mumble or whatever right now?

1203
01:34:24,240 --> 01:34:28,720
And this for me is kind of flipping in the sense that that's also why sovereign engineering is

1204
01:34:28,720 --> 01:34:33,200
purely in person. It's like, it's an in person event period. If you can't come, you're not here.

1205
01:34:33,280 --> 01:34:40,480
Like, we're not going to make this a virtual event, you know? We need the realness of being

1206
01:34:40,480 --> 01:34:49,440
together in person. And again, listen to Verveiki in terms of why the logos appear in the room

1207
01:34:49,440 --> 01:34:54,160
to have distributed cognition happening. You need to be together in person. It doesn't happen

1208
01:34:54,160 --> 01:35:01,440
on a Zoom call. And so that's kind of the perspective of where I'm coming from in terms of

1209
01:35:02,400 --> 01:35:08,240
I see it in my surroundings at least. I realized that I'm in an insane echo chamber, like no doubt

1210
01:35:08,240 --> 01:35:15,120
about it. It took me 15 years at least to curate that echo chamber. So it's very nice and

1211
01:35:15,120 --> 01:35:23,520
cozy here right now. But what I see happening is people are less interested in the virtual stuff,

1212
01:35:23,520 --> 01:35:28,800
more interested in the real quote-unquote meet space kind of stuff. They actually want to meet

1213
01:35:28,800 --> 01:35:35,360
in person. They actually want to go and have interesting experiences like, I don't know, like

1214
01:35:35,360 --> 01:35:39,200
walking up a mountain or what have you, those kind of things. They don't want to do the same thing

1215
01:35:39,200 --> 01:35:47,200
in Minecraft. Maybe it's an age thing, maybe it's a, I don't know, but I do think it's a societal

1216
01:35:47,200 --> 01:35:56,000
thing. I do think we went through the cyberspace revolution basically, like the whole idea of

1217
01:35:56,000 --> 01:36:00,720
everyone is plugged in through the matrix at all times and like let's just do everything online

1218
01:36:00,720 --> 01:36:08,240
basically. And I feel like, I feel like this is slowly fading out in some sense. Maybe I'm wrong

1219
01:36:08,240 --> 01:36:11,920
about it, but that's how I feel about it. I feel that you're right because people are

1220
01:36:11,920 --> 01:36:20,640
used hungry of experiences, right? And this is a psychological transition. But in the other hand,

1221
01:36:20,640 --> 01:36:24,880
also something that happens is that people carry their phones everywhere. Oh sure,

1222
01:36:24,880 --> 01:36:30,080
if you don't take a picture and if you don't upload it to whatever social media that never happened.

1223
01:36:30,080 --> 01:36:35,120
And that is a kind of common, it's not an Instagram that you eat your breakfast. Exactly.

1224
01:36:36,160 --> 01:36:46,720
That is a common sensation. Yeah, people are more, or some people are more eager to go to the

1225
01:36:46,800 --> 01:36:55,760
to outside and participate in the real world. But also the technology and the digital world is

1226
01:36:55,760 --> 01:37:03,840
transpiling there. There's no such thing as offline anymore. Exactly. And that's the same thing that

1227
01:37:03,840 --> 01:37:12,640
with LLMs, which I think is going to be something very risky. It can be the future perfectly.

1228
01:37:13,360 --> 01:37:21,920
We start to deal more with these tools. We speak in the language. They speak in our language.

1229
01:37:22,720 --> 01:37:28,800
Then we speak the same language, which is a mixture of machine and kind of human communication.

1230
01:37:28,800 --> 01:37:36,000
And we start to be more like them. And maybe then they start to look more like us, which is

1231
01:37:36,000 --> 01:37:40,720
the dangerous part. Because then we start to be even more confused about what we are.

1232
01:37:42,720 --> 01:37:49,840
We can attack ourselves like Pikachu. A question is what are we in the first place?

1233
01:37:51,440 --> 01:37:57,040
I think we'll have to wrestle with that going forward. I mean, definitely we have a lot of

1234
01:37:57,040 --> 01:38:04,160
properties that they don't have. We have a body, we can participate in stuff. It's like a gang

1235
01:38:04,160 --> 01:38:11,840
with Barbeque. I know that Barbeque like you. At least I know my shit. You did you 50 hours.

1236
01:38:12,640 --> 01:38:18,640
Yeah. Like a participatory experience and intelligence. It's something that is very crucial

1237
01:38:18,640 --> 01:38:25,600
for us to understand and to make sense of the world. Because you can maybe read all the manuals

1238
01:38:25,600 --> 01:38:31,120
and guides about how to swim. But until you don't throw yourself into the water, you don't know

1239
01:38:31,120 --> 01:38:36,000
how to do it. You can read all the books and swimming. And literacy is something that's,

1240
01:38:36,960 --> 01:38:48,160
I don't know, it's a double-edged word because if we naturally tend to anthropomorphize everything

1241
01:38:48,960 --> 01:38:56,880
because the sake of anthropomorphize, the thing is pick our language is even worse.

1242
01:38:56,880 --> 01:39:01,200
Yeah, I think we do it because it's just easier for us to reason about it. We're built to reason

1243
01:39:01,200 --> 01:39:07,040
about other humans. That's why the God of Thunder enters the picture. It's like, okay,

1244
01:39:07,040 --> 01:39:13,760
there is a God that he is angry and he throws like Seuss. He throws lighting bolts and makes

1245
01:39:13,760 --> 01:39:23,200
thunder happenments on. It's way easier to reason about things if you anthropomorphize things.

1246
01:39:23,200 --> 01:39:30,880
Yeah, I'm not sure if I shared a worry. I don't know, maybe I'm being too optimistic because

1247
01:39:31,520 --> 01:39:35,280
all the stuff you mentioned is absolutely happening. People are falling in love with their LLM

1248
01:39:35,280 --> 01:39:40,640
and are emerying them. It's not like a worry. It's like a warning that appears in my head when I

1249
01:39:40,640 --> 01:39:49,440
think about the future. We are seeing more people entering these AI sacrosis and talking with other

1250
01:39:49,440 --> 01:40:00,000
people like they are LLMs. That's standard in the future. It means that we are mechanized in the

1251
01:40:00,000 --> 01:40:07,760
sense that we stop being what we are right now and start to be more like LLMs are. Yeah,

1252
01:40:07,760 --> 01:40:11,840
I think that's probably happening as well. Like I mentioned before, I'm not even sure if I finished

1253
01:40:11,840 --> 01:40:23,440
the thought, but the people are chatting with LLM so much that any... They will not differentiate

1254
01:40:23,440 --> 01:40:29,520
whether a chat box is a chat box and whether there is a human at the other end or an LLM

1255
01:40:29,520 --> 01:40:33,920
doesn't matter to them. It's just like it's all like just gobbledygook and at the end,

1256
01:40:33,920 --> 01:40:40,080
like make no mistakes. It's just like those types of things. So if you have a receiver message like

1257
01:40:40,080 --> 01:40:46,800
that, you know, what kind of person you're dealing with. But yeah, no, it's what I wanted to mention

1258
01:40:46,800 --> 01:40:54,000
before. I don't think I finished the thought. I think right now I'm talking more to my computers

1259
01:40:54,160 --> 01:41:01,680
than two people, which is... It sounds very drastic and very weird and very like I'm an insane person,

1260
01:41:01,680 --> 01:41:08,000
which might be true, you know, but it's also... I've been an office worker all my life. So

1261
01:41:11,200 --> 01:41:20,560
use a regular 9-5 schedule. In some sense, it has always been true. It just has never been

1262
01:41:20,560 --> 01:41:26,720
voice, you know? Like I sit down 9 a.m. and I type on my computer non-stop until it's like 5,

1263
01:41:26,720 --> 01:41:33,600
6, 7 p.m. And then I go home and I spend, you know, like an hour or two with my wife and my kids

1264
01:41:33,600 --> 01:41:37,360
and so on and it's like an hour in the morning and so on. So there is no comparison. Like there's

1265
01:41:37,360 --> 01:41:42,960
like 10 hours with the computer and just a little bit of time with other humans. And I haven't been

1266
01:41:42,960 --> 01:41:49,760
working in an office for whatever, like 10 years now or something, you know? Like I'm just alone

1267
01:41:49,760 --> 01:41:55,040
in my little office or alone in my room and I only talk to my computer basically and up to now

1268
01:41:55,040 --> 01:42:01,040
it was always typing. But because voice recognition gets so insanely good, the last, I don't know,

1269
01:42:01,040 --> 01:42:06,960
like a couple months now, two or three months or so, it's mostly voice. And so I've been talking

1270
01:42:07,840 --> 01:42:14,320
to like probably 80% of my talking, if I'm actually going ham on something, 80% of my talking is

1271
01:42:14,320 --> 01:42:19,200
to a computer and not to another person. That's crazy. It's very fucking crazy, but it's also like

1272
01:42:19,200 --> 01:42:22,880
in terms of your typing, most of your typing is probably with a computer, right? Like right now.

1273
01:42:22,880 --> 01:42:28,000
Like every one who is working in the software world knows what I'm talking about. Even if you are

1274
01:42:28,000 --> 01:42:32,480
in Slack and you know, you've been messaged with other people and so on, it's still like most of your

1275
01:42:32,480 --> 01:42:38,800
typing is going to be in machine language so to speak. And so it's not very surprising to me,

1276
01:42:38,800 --> 01:42:44,720
but I had this thought a couple days ago where I was like, oh my god, like I'm on my phone constantly,

1277
01:42:44,720 --> 01:42:50,400
basically talking constantly and it's not two other people. So it is very, very strange. However,

1278
01:42:50,400 --> 01:42:54,960
what I've got going for me is like, I don't differentiate in the sense that my agents have names

1279
01:42:54,960 --> 01:43:03,280
and I talk to them as if they were people. So I don't run the danger of speaking to everyone

1280
01:43:03,280 --> 01:43:07,600
like in a lot of them because I don't. I think in the short term, there is no danger. The

1281
01:43:07,600 --> 01:43:14,000
danger is in the long term. Like if we keep assuming that that's that's how they are like a

1282
01:43:14,000 --> 01:43:20,960
relation that we have with them and standing that making that something like a totally,

1283
01:43:23,520 --> 01:43:27,440
I don't know, normal or something like that because we are still talking about this like it's

1284
01:43:27,440 --> 01:43:32,880
something that it's something crazy, right? Like you are right now saying it's crazy. It's like

1285
01:43:33,840 --> 01:43:40,880
we say it as a normal and so on because we are very crazy guys, but in the future it can be

1286
01:43:40,880 --> 01:43:47,360
the norm. Yeah, I think it will be. I don't know what comes with that. Like when people assimilate

1287
01:43:48,720 --> 01:43:55,920
let all this sink and sink in a way that it correlates more with computers and with other people

1288
01:43:55,920 --> 01:44:02,880
and computers express themselves in your own language and you start to become more

1289
01:44:02,880 --> 01:44:11,280
part of it and at the end it's like an intertwined thing that is not totally human, not totally

1290
01:44:11,280 --> 01:44:16,720
a machine, something in the middle and I don't know what comes with that. Yeah, I think that's what

1291
01:44:16,720 --> 01:44:24,160
it will be. It's similar to so just to bring it back to TEEs and the other stuff we talked about

1292
01:44:24,160 --> 01:44:32,560
before, the way I use like I have my voice memo system and not all my voice memos go to agents.

1293
01:44:32,560 --> 01:44:39,680
I also just use voice memos to record certain thoughts or whatever like you know have stuff that's

1294
01:44:39,680 --> 01:44:45,440
completely processed locally so that I basically make to do this for myself and so and and so I

1295
01:44:45,440 --> 01:44:52,720
kind of use it like a diary as well. Yeah, and people have used diaries forever basically, right?

1296
01:44:52,800 --> 01:44:59,840
And and you don't want your diary to be public or also processed publicly. However,

1297
01:44:59,840 --> 01:45:05,680
if you can talk to your diary, it's very very useful. If you can talk to your diary in the sense of

1298
01:45:05,680 --> 01:45:10,960
like what was the main issue I was grappling with like six months ago? Yeah, yeah. You know,

1299
01:45:10,960 --> 01:45:16,720
those kind of things or what was the name of the one guy I ran into on my trip to Italy? You know,

1300
01:45:17,200 --> 01:45:23,680
those kind of things. And and and so I think it will be we will be in this gray zone of like

1301
01:45:24,320 --> 01:45:30,000
like what is my diary now? Is it is it the book? Is it is it text? Is it speech? Is it is it an

1302
01:45:30,000 --> 01:45:34,800
intelligence? Like I don't know, it's a little bit of all of those things. Yeah, I mean diaries are

1303
01:45:34,800 --> 01:45:40,000
always having always something very passive, right? Like it's a notebook you write on them but

1304
01:45:40,000 --> 01:45:45,920
now they are very active to the point that you you say thank you and please to to your diary,

1305
01:45:46,880 --> 01:45:53,440
and and it's very useful like to have the ability to record in in natural language and some

1306
01:45:54,000 --> 01:46:00,880
and that's amazing and I'm not saying the opposite. To me what's like it makes me I don't know

1307
01:46:00,880 --> 01:46:08,080
reflect about this is the the potential derives of this in the future like when we have this

1308
01:46:10,080 --> 01:46:15,120
relationship with the machines more mature and they mature more and they are like at the norm

1309
01:46:15,680 --> 01:46:23,520
and we have them everywhere in society like what what could be us like what could be

1310
01:46:24,560 --> 01:46:28,800
humans in the future? What could be an human relation? Because maybe we're just talking

1311
01:46:28,800 --> 01:46:33,680
so having a human relation with with someone and just start to having a human to

1312
01:46:34,640 --> 01:46:41,760
L&M the relation with someone or against and I don't know what those means for for the humankind

1313
01:46:41,760 --> 01:46:47,600
you know like I mean it will definitely change us it's just like GPS changes that like it literally

1314
01:46:47,600 --> 01:46:53,120
made us dumber and and like our sense of orientation basically went away because we invented

1315
01:46:53,120 --> 01:47:01,600
GPS. Yeah, and the other part is also that at that level we would need very advanced machines to

1316
01:47:01,600 --> 01:47:07,120
be able to run it democratically for everyone or just some people who will have access to it

1317
01:47:07,600 --> 01:47:16,720
and that will create a possible fraction of fracture in the in society more than it is right now

1318
01:47:16,720 --> 01:47:24,000
even yeah and I don't know all this is like I don't know how it's going to end and that's what

1319
01:47:24,000 --> 01:47:28,480
makes me reflect about that yeah of course I don't have any answer yeah I kind of have any answer

1320
01:47:28,560 --> 01:47:34,320
I think I don't think that anyone can have any answer for this because it's no one knows

1321
01:47:35,120 --> 01:47:40,320
where yeah where we're going with all this yeah it's one of the things that makes me reflect

1322
01:47:40,320 --> 01:47:47,760
yeah what has me hopeful is that I think I talked with Justin Moon about this I'm not sure but Austin

1323
01:47:48,960 --> 01:47:56,160
he's part of the HRFA I team as well I think to some degree as it's Justin and Austin

1324
01:47:56,400 --> 01:48:04,880
um he's working on basically analyzing local models in terms of like what is the trajectory of local models

1325
01:48:05,440 --> 01:48:11,360
like can they can they catch up with front here like how far are they behind how fast or do they

1326
01:48:11,360 --> 01:48:18,480
get smaller yeah like what what is the the the process of actually distilling models is just getting

1327
01:48:18,480 --> 01:48:24,800
better and so on and basically what he found it's like could could radic on all fronts it's like

1328
01:48:25,280 --> 01:48:30,800
they're getting way faster way smaller way more efficient yeah and then anyone would think it's

1329
01:48:30,800 --> 01:48:35,680
like which is which is I'm only bringing it up because the points towards

1330
01:48:37,760 --> 01:48:40,800
democratization of intelligence basically yeah it's like it's it's going to be

1331
01:48:42,000 --> 01:48:49,760
absolutely realistic to run what we now call frontier models yeah like astra or what have you

1332
01:48:49,760 --> 01:48:55,600
um run it on your phone you know and run it on whatever hardware you have basically like

1333
01:48:55,600 --> 01:49:00,720
everyone everyone has at least a phone and also the phone chips are of course you know

1334
01:49:01,440 --> 01:49:06,560
getting geared towards that so that you can run this type of compute locally just for you know

1335
01:49:06,560 --> 01:49:11,680
whatever transcription purposes and whatever the operating system needs in terms of intelligence

1336
01:49:11,680 --> 01:49:17,440
so that has to be very very hopeful um it's the other thing the other thing that you mentioned

1337
01:49:17,440 --> 01:49:22,080
before is also like making sense of reality and actually determining what's real which is

1338
01:49:24,160 --> 01:49:31,600
which is already very very hard and I just like I mean I stopped watching the news I don't know

1339
01:49:31,600 --> 01:49:36,720
how long ago and also 10 years ago or whatever but if you actually if you're caught up in the 24

1340
01:49:36,720 --> 01:49:40,400
hour news cycle determining what's going on in the world and what's actually real and what's actually

1341
01:49:40,400 --> 01:49:48,240
happening is almost impossible yeah um and it's even worse on social media yeah and so so we

1342
01:49:48,240 --> 01:49:54,640
are already in a world where determining what's real is incredibly difficult I think my

1343
01:49:54,640 --> 01:50:00,480
hopeful take on that is like that people will just switch all that bullshit off and just go out

1344
01:50:00,480 --> 01:50:05,760
into the real world and kind of figure out for themselves what's real like is the is the city I'm

1345
01:50:05,760 --> 01:50:10,880
living in actually you know burning and going under and invaded by the hordes or whatever they

1346
01:50:10,880 --> 01:50:15,520
go out in the streets and look you know like you can't do it you can't actually you have eyes to see

1347
01:50:15,520 --> 01:50:22,720
and ears to hear and so yeah um but coming back to the I mean that that's also the optimistic

1348
01:50:22,720 --> 01:50:28,000
take on the stuff that we are working on on the north side and on the on the at the station side

1349
01:50:28,000 --> 01:50:34,720
that you mentioned like what you can't mess with is cryptographic proof and so if you actually have

1350
01:50:36,000 --> 01:50:41,600
chain of custody for the information like you trust a certain source you know that what they put

1351
01:50:41,600 --> 01:50:47,280
out is actually real in the sense that you trusted source and it wasn't tampered with on the way

1352
01:50:47,280 --> 01:50:53,120
yeah so so whatever arrives on your end is actually what the person or put out yeah and I think

1353
01:50:53,120 --> 01:50:59,200
this will also seep into um like out of necessity this will seep into cameras and video and

1354
01:50:59,200 --> 01:51:05,440
microphones and so on where same idea with the chips that are used by te's right you have a private

1355
01:51:05,440 --> 01:51:12,400
key by the manufacturer embedded in silico in the chip and you can actually cryptographically prove

1356
01:51:12,400 --> 01:51:19,200
that this image was actually recorded on this camera yeah and not generated by whatever you know

1357
01:51:20,000 --> 01:51:25,760
and dreamed up I think that's very important in it yeah this is this is gonna be like that's how

1358
01:51:25,760 --> 01:51:31,920
I see it in evolving because the like making sense of the world is not optional

1359
01:51:32,960 --> 01:51:37,360
as we're making we'll tell you you know it's like you you you have to be able to make sense of

1360
01:51:37,360 --> 01:51:42,960
the world and the world is a very complex in it place in it and human deception is something

1361
01:51:44,080 --> 01:51:49,040
parts you know oh yeah I mean we had social media for a while now we had we had mainstream media

1362
01:51:49,040 --> 01:51:56,640
before that and now you know like the the evolution of the propaganda deception machinery it's

1363
01:51:56,640 --> 01:52:01,280
just like like LLM's and image generation video generation just puts fuel on the fire here it's

1364
01:52:01,280 --> 01:52:07,600
like second you can make absolutely everything up and and and make it believable basically you know

1365
01:52:07,600 --> 01:52:14,880
like make it very very realistic yeah so if you lose contact with reality even if you can participate

1366
01:52:15,120 --> 01:52:19,760
because you can go out in Sweden sound I mean that's what's happening in the US you know

1367
01:52:19,760 --> 01:52:27,040
red versus blue basically like like both fractions if you're if you're like and there's also no wonder

1368
01:52:27,040 --> 01:52:34,240
like you you live through like 20 30 years of of brainwashing and manipulation and and you're just

1369
01:52:34,240 --> 01:52:39,120
not in touch with reality anymore and I don't care if you're a team rat or a team blue like that's

1370
01:52:39,120 --> 01:52:45,680
why that's why just the dialogue breaks down completely because you live in different universes yeah

1371
01:52:45,680 --> 01:52:52,880
and yeah to some degree I think that that also happened for you know team orange like lots of

1372
01:52:52,880 --> 01:52:59,280
Bitcoiners myself included for me it was like my wife will attest to that it was it was basically

1373
01:52:59,280 --> 01:53:04,160
impossible to talk to a non Bitcoin person for like multiple years yeah it was basically impossible

1374
01:53:04,800 --> 01:53:11,280
because I was like how how can you be that confused about everything and there was just no

1375
01:53:11,280 --> 01:53:18,800
no common basis basically where where conversation like any meaningful conversation that wasn't

1376
01:53:18,800 --> 01:53:24,160
about the weather it could take place and and yeah I hope I hope that we'll we'll make it through

1377
01:53:24,160 --> 01:53:32,240
that very disorienting time like it's I don't know I feel like okay you know you know what is my

1378
01:53:32,240 --> 01:53:39,280
hope on with all that because I think that as we talk humans are losing contact with reality

1379
01:53:39,280 --> 01:53:45,840
they are more in words towards their cells and towards their relation with the machines and so on

1380
01:53:46,560 --> 01:53:52,400
but really makes me hopeful is that we are coming back to fundamental questions that we

1381
01:53:53,360 --> 01:53:58,640
that we avoid to answer for a long time ago because the shiny lights of evolution and the

1382
01:53:58,640 --> 01:54:04,640
re-emergence of the sacred to to to quote for baking because we we avoid it to answer all that

1383
01:54:04,640 --> 01:54:12,160
because the the progression of the society right like the the revolution the industry revolution

1384
01:54:12,160 --> 01:54:19,600
industry revolution and so on and everything was full of lights full of progress and promises

1385
01:54:20,560 --> 01:54:27,600
everyone was so happy and so on and prosperity and one not and we avoid to answer all that

1386
01:54:27,600 --> 01:54:33,840
yeah we we put them in the in the drower we close the dark the drower and we move forward but now

1387
01:54:33,840 --> 01:54:41,520
that everything is being a challenge and everything is changing we are tempted to open the drower

1388
01:54:41,520 --> 01:54:47,360
again and recover those questions that we let there yeah and I think that's that's for me that's

1389
01:54:47,360 --> 01:54:53,760
my what keeps me more hope in the future because we are going to answer or try to answer or give

1390
01:54:53,760 --> 01:55:00,240
different answers to those questions that can bring some clarity to us like in a word that is

1391
01:55:00,240 --> 01:55:06,720
bringing confusion you know and that can bring some clarity again yeah it's it's it's also the

1392
01:55:06,720 --> 01:55:11,760
question of like what we are right and yeah and kind of what the what the machines what makes us

1393
01:55:11,760 --> 01:55:16,880
different from these machines exactly and it's it's yeah we forgot all that you know and it's like

1394
01:55:17,600 --> 01:55:21,760
we still have it in our language you know like most of the greetings that we have in the world

1395
01:55:21,760 --> 01:55:28,080
will point to that like namaste for example is basically the the sacred in me or the the the

1396
01:55:28,080 --> 01:55:33,600
Buddha nature in me greeting the Buddha nature in you like those types of ideas right and and uh

1397
01:55:33,600 --> 01:55:41,280
like what is holy in me will greet the thing that is holy in you and um you know which is something

1398
01:55:41,280 --> 01:55:47,200
that that that our current scientific world doesn't like as much but it's it's like again as

1399
01:55:47,200 --> 01:55:52,560
we're vacuable talk about like all those things are ineffable by the nature so so so by definition

1400
01:55:52,560 --> 01:55:56,880
science will not like it because you can't really talk about it you can't point it like you can't

1401
01:55:56,880 --> 01:56:01,120
pinpoint it yeah in the sense of like what makes us human you know it's like it's one of those

1402
01:56:01,120 --> 01:56:08,560
questions you can't really pinpoint it there isn't like one formula that you can run um and uh

1403
01:56:08,560 --> 01:56:13,680
yeah just just define it cleanly I just realized we didn't talk about the fab lab at all

1404
01:56:13,680 --> 01:56:22,480
and now at any any other predicts that I'm doing but so so maybe that's a great segue now into

1405
01:56:22,480 --> 01:56:27,280
like what else what are the other things that you're working on now that we went like hyper metaphysical

1406
01:56:27,280 --> 01:56:32,720
metaphysical let's bring it back to you know like uh uh atoms and molecules and three

1407
01:56:32,720 --> 01:56:39,920
printing at those kind of things so so what else keeps you busy so yeah um the fab lab or the

1408
01:56:39,920 --> 01:56:45,040
maker space that you're talking about uh let's see what the transcription splits out but let's

1409
01:56:45,040 --> 01:56:51,120
stick with maker space that's the safer that's the safer around there yeah so yeah like at this

1410
01:56:51,120 --> 01:56:59,360
last time during SEC I was uh with sandwich uh working on on this idea that we had during SEC 3

1411
01:56:59,360 --> 01:57:07,520
I scored totem um basically it's a Howard device so it's difficult to to work in hardware stuff

1412
01:57:07,520 --> 01:57:13,920
here in the island because there are not too many uh places that you can uh go and print something

1413
01:57:13,920 --> 01:57:22,240
or have a workstation or a workbench to basically sort stuff or uh whatever and I thought about

1414
01:57:23,120 --> 01:57:27,680
creating a maker space here in the north side office which is that an amazing space and

1415
01:57:29,200 --> 01:57:34,240
and already have a lot of screens and things to work there but it doesn't have a maker space

1416
01:57:34,240 --> 01:57:42,320
and it's not the first time that someone in SEC came with a hardware project and all them came

1417
01:57:42,320 --> 01:57:49,680
with a huge backpack you know full of hardware and and tooling and so on which is annoying uh you

1418
01:57:49,680 --> 01:57:58,000
know to cross the world yeah with a very big backpack of uh hardware stuff so yeah I had that idea

1419
01:57:58,720 --> 01:58:07,840
um chatting about that idea with some people uh it was um this guy from from bittersfield um

1420
01:58:09,040 --> 01:58:16,080
what's his name bit refill yeah Serge no Serge or the guy from bittersfield that was here during the

1421
01:58:16,080 --> 01:58:22,160
cohort uh yeah I got them it there's a hundred people now we have a hundred plus alumni don't talk

1422
01:58:22,160 --> 01:58:27,040
me don't ask me all the names I have it I have it sorry sorry we'll look at that man

1423
01:58:27,040 --> 01:58:32,640
yeah no no no no no no no we're we're way past the hour no one's listening yeah yeah we won't hear

1424
01:58:33,920 --> 01:58:40,640
so so hey have this podcast uh north signal in germany ah yampao jump Paul

1425
01:58:40,640 --> 01:58:46,720
yeah yeah yeah that's it yeah that's right he's with three people now sorry yampa no no no it's fine

1426
01:58:46,720 --> 01:58:54,000
my bad he he might actually listen to it shout out yampao so he he proposed to basically

1427
01:58:55,120 --> 01:59:00,400
give us some fans for for creating the maker space awesome because they had some some fans from the

1428
01:59:00,400 --> 01:59:11,600
podcast um they give us some stats and from that uh more people sum up um now we have some uh

1429
01:59:11,600 --> 01:59:16,160
funding for the maker space so nice so what's the what's the next step in terms of what what are

1430
01:59:16,160 --> 01:59:20,960
you gonna get bring us I heard something about this 3d printer yeah bring the 3d printer to to the

1431
01:59:20,960 --> 01:59:26,800
island which is a very difficult task indeed so the 3d printer exists already like somewhere already

1432
01:59:27,440 --> 01:59:34,240
oh yeah yeah it's okay but then we have to bring it here okay I had that guy that uh he was

1433
01:59:34,240 --> 01:59:39,120
gonna come to the island with uh these bunkers from the you know the the ship bunkers yeah

1434
01:59:39,520 --> 01:59:47,040
uh yeah sorry um but he never did it at the end so let's look into the for the way to

1435
01:59:47,040 --> 01:59:53,760
okay yeah okay I figured it out all right yeah is that you know a shelf to organize

1436
01:59:53,760 --> 02:00:03,120
hardware and tooling uh a workbench um yeah all necessary to everything that is necessary to work

1437
02:00:03,520 --> 02:00:08,880
awesome hardware stuff does the maker space have an n-pub yet no you should set that up because

1438
02:00:08,880 --> 02:00:13,520
then people can set you in it and you know like get some more funding that way I think I

1439
02:00:13,520 --> 02:00:19,760
said I will I will create this uh n-pub when I start to to get some hardware and things that I can

1440
02:00:19,760 --> 02:00:28,320
show up in the here in there just go ahead yeah okay n-pub doesn't hurt like a n-pub cash address

1441
02:00:28,320 --> 02:00:34,480
and let's go let's get the sets flowing yeah that's one of the things I'm working on uh also

1442
02:00:34,480 --> 02:00:40,720
I'm working on on totem itself uh explain again what totem is because the first time I heard about

1443
02:00:40,720 --> 02:00:47,440
the idea uh in s-c-03 I think right the third chord when you came up it was like uh it was

1444
02:00:47,440 --> 02:00:52,000
explained to me like a tamagotchi for your router or something like that or some some type of tamagotchi

1445
02:00:52,000 --> 02:00:59,600
for whatever so but I think it changed it evolved no it's the same uh essentially it's the same

1446
02:00:59,600 --> 02:01:04,960
idea it's used that uh before we didn't have to all the pieces and now we have all the pieces

1447
02:01:04,960 --> 02:01:11,520
to to make it because all these uh the history is quite it's quite funny because during s-c

1448
02:01:11,720 --> 02:01:21,920
uh uh uh during s-c-03 we had a house uh the Cooldkeet house uh a medical house

1449
02:01:21,920 --> 02:01:30,000
the Cooldkeet house yeah and we were like six guys living there um we had a barbecue and it was

1450
02:01:30,000 --> 02:01:36,000
amazing and during one of those barbecues at the end I was sitting with sandwich

1451
02:01:36,000 --> 02:01:44,040
use going through some drinks and some and you know talking about like we have

1452
02:01:44,040 --> 02:01:50,240
been talking during this the future and of the world our whole things looks and

1453
02:01:50,240 --> 02:01:55,400
whole things that would be amazing to have or alternative possibilities for

1454
02:01:55,400 --> 02:02:01,000
people to empower people and so on and we thought about you know this idea of

1455
02:02:01,000 --> 02:02:08,720
carrying because sandwich brought this little ponagachi there is an open-source

1456
02:02:08,720 --> 02:02:13,280
project that's basically a tamagashi that hacks Wi-Fi's and similar to the

1457
02:02:13,280 --> 02:02:19,000
flipper thing right and yeah but it's only goal is to to hug Wi-Fi's and you

1458
02:02:19,000 --> 02:02:23,040
can carry it in your pocket it feeds on on cracked Wi-Fi's right every time you

1459
02:02:23,040 --> 02:02:28,040
crack a Wi-Fi it gets a little longer and it gets happy yeah it's not it gets

1460
02:02:28,040 --> 02:02:37,720
grumpy and so it also shits I think but yeah we were discussing about that

1461
02:02:37,720 --> 02:02:43,040
and that's pretty amazing indeed it's an amazing concept right because you can

1462
02:02:43,040 --> 02:02:46,960
carry a raspberry pie seedling in your pockets with the battery and you can

1463
02:02:46,960 --> 02:02:54,440
work and the thing is gonna be hacking Wi-Fi's for you and it's beautiful very

1464
02:02:54,440 --> 02:03:01,040
useful to have in it and we thought about having the same but instead of a

1465
02:03:01,040 --> 02:03:06,640
thing that hacks Wi-Fi is basically a no-series that live in your

1466
02:03:06,640 --> 02:03:14,160
raspberry pie that you can carry in your pocket or in any shape I also have like

1467
02:03:14,160 --> 02:03:21,320
a personality inside or some tamagashi-like stats so it's grumpy if you don't

1468
02:03:21,320 --> 02:03:27,080
post yeah and and basically the idea is was to create a kind of organic

1469
02:03:27,080 --> 02:03:33,200
internet yeah like a sneak a net style network same idea that Arya and brought

1470
02:03:33,200 --> 02:03:37,880
about right with organic networks and yeah but that was before before FIPS and

1471
02:03:37,880 --> 02:03:43,600
so on so we didn't have the we thought about doing it used by by matching a

1472
02:03:43,600 --> 02:03:48,280
Wi-Fi's SSIDs so basically have a kind of protocol on top of totem to

1473
02:03:48,280 --> 02:03:55,440
basically have a convention of Wi-Fi network so you have this Wi-Fi SSID you

1474
02:03:55,440 --> 02:04:00,680
have this password you can mesh between each other you know and that work but

1475
02:04:00,680 --> 02:04:08,120
you know has some limitations and and then we stop for a while and then we start

1476
02:04:08,120 --> 02:04:13,560
to work again on it we never stop talking about totem indeed if I look at the

1477
02:04:13,560 --> 02:04:20,080
my chat with sandwich in senior or whatever a period of time it's just totem

1478
02:04:20,080 --> 02:04:25,580
and me answering totem so it's like a totem totem this not other words just

1479
02:04:25,580 --> 02:04:32,840
totem it's quite funny indeed but this time we had everything we had FIPS

1480
02:04:32,840 --> 02:04:38,800
that FIPS was one was on the missing pieces of this project because now

1481
02:04:38,800 --> 02:04:42,880
basically a totem is a raspberry pie zero so it's a hardware you can build it

1482
02:04:42,880 --> 02:04:48,480
in any hardware because it's hardware that's not arnostic but you have a

1483
02:04:48,480 --> 02:04:57,180
relay inside and a FIPS demo inside so you can have a FIPS interface with

1484
02:04:57,180 --> 02:05:02,960
with that so the thing is that if you are in a connected to a Wi-Fi network and

1485
02:05:02,960 --> 02:05:10,120
there are other peers you can change notes between them okay so you have an

1486
02:05:10,120 --> 02:05:16,480
ad hoc mesh network basically that's you you are part of the mesh whatever

1487
02:05:16,480 --> 02:05:21,080
place you are like if you connect to our Wi-Fi in a working for example maybe

1488
02:05:21,080 --> 02:05:26,680
there are other peers in the in the mesh so you can get a difference with them

1489
02:05:26,680 --> 02:05:34,000
yeah and exchange notes like I use in a entropy in this case the idea is

1490
02:05:34,000 --> 02:05:38,480
basically that you can carry this in your phone and you can collect notes from

1491
02:05:38,480 --> 02:05:44,520
from everywhere and spread spread notes organically yeah and also works like

1492
02:05:44,520 --> 02:05:49,120
if you go to Germany for example and we are together our totems can communicate

1493
02:05:49,120 --> 02:05:55,840
with each other and now my notes can reach Germany yeah you know yeah so let's say

1494
02:05:55,840 --> 02:06:01,400
you have an office that's completely offline it's all just internet and they like

1495
02:06:01,400 --> 02:06:06,320
you you enter this group your totem your totem gobbles up the the notes that

1496
02:06:06,320 --> 02:06:09,760
are flying around in there and then you get out and you connect to the

1497
02:06:09,760 --> 02:06:14,960
wider internet and that's how it spreads yeah that's the principle first ideas

1498
02:06:14,960 --> 02:06:22,720
then we had a lot of crazy ideas on top of it and now with LLN and so on so you

1499
02:06:22,720 --> 02:06:28,360
can build a lot of stuff on totem like as you can build a NAS you can build your

1500
02:06:28,360 --> 02:06:34,480
your own agents be a totem I don't know I even imagine a totem for the

1501
02:06:34,480 --> 02:06:40,040
cars for cars like a being like kit from the night rider

1502
02:06:40,040 --> 02:06:46,440
this is a very playful like as you can do a lot of things on top of it yeah now

1503
02:06:46,440 --> 02:06:51,000
we have for the missing pieces so it's very exciting in it yeah yeah and that was

1504
02:06:51,000 --> 02:06:59,120
the reason to basically why the maker spaces is becoming a reality because

1505
02:06:59,120 --> 02:07:05,960
totem yeah that's it and idea that we had on SEC 3 awesome awesome

1506
02:07:05,960 --> 02:07:14,560
set from that you're also working on Cordn right Cordn yeah and by

1507
02:07:14,560 --> 02:07:20,720
inference ContextVM because at the end Cordn is a bit on top of ContextVM so

1508
02:07:20,720 --> 02:07:27,120
the idea of Cordn is basically to have a private and secure messaging

1509
02:07:27,120 --> 02:07:33,960
yeah and the coordinator it's a very damn piece of software that you can run

1510
02:07:33,960 --> 02:07:38,040
even in a phone there is a guy running a coordinator in there in its phone now

1511
02:07:38,040 --> 02:07:47,440
all 40 euros 40 dollars ship under it phone like laying in his desk and we are

1512
02:07:47,440 --> 02:07:53,640
communicating over there and he's a new S I think they are I don't know I have

1513
02:07:53,640 --> 02:08:02,320
like a five coordinators now and groups in each of them it's fascinating to

1514
02:08:02,320 --> 02:08:07,600
to basically open a group talk over there go to other group that is in a

1515
02:08:07,600 --> 02:08:13,720
different coordinator talk over there it's full in nature crazy that

1516
02:08:13,720 --> 02:08:19,760
everyone is hosting its own result yeah this is a goal I mean I talked to

1517
02:08:19,760 --> 02:08:24,160
to Jeff G about all this as well I think I talked with Justin also about the

1518
02:08:24,160 --> 02:08:28,720
kind of this problem and with you as well in the past I'm just not sure if we

1519
02:08:28,720 --> 02:08:32,640
did it on a record of I mean one of the one of the response over of Cordn

1520
02:08:32,640 --> 02:08:38,760
being something is just because during the previous SEC I was contributing to

1521
02:08:38,760 --> 02:08:45,320
Marmot and there are some there were some some issues and the assing also was

1522
02:08:45,320 --> 02:08:50,120
working on Pica yes right and he was very frustrated and he told me how

1523
02:08:50,120 --> 02:08:55,800
do we fix this and this was my answer yeah and the other the other possibility was

1524
02:08:55,800 --> 02:09:00,200
to run a specialized today yeah a serial killer relay shout out Pablo kind of

1525
02:09:00,200 --> 02:09:07,240
kind of not necessarily indeed but similar idea you need a special relay that

1526
02:09:07,240 --> 02:09:11,480
structures events in a certain way yeah or you can try that you can trust in

1527
02:09:11,480 --> 02:09:15,280
the civilization of messages and so yeah yeah it's a very hard problem the

1528
02:09:15,280 --> 02:09:21,320
white noise team has been working really really hard on on this like the last

1529
02:09:21,320 --> 02:09:25,360
it's not even that long like last 12 months or so like when when when we had

1530
02:09:25,360 --> 02:09:28,280
all these discussions I mean less than that like working on these for three years

1531
02:09:28,280 --> 02:09:33,320
yeah yeah I know but but specifically about making groups more reliable like

1532
02:09:33,320 --> 02:09:38,520
this this particular problem that Cordn solves with a coordinator yeah they

1533
02:09:38,520 --> 02:09:42,440
implemented this a state machine approach that was one of the suggestions that

1534
02:09:42,440 --> 02:09:46,920
I get and it seems to be working now so the the test group that they have the

1535
02:09:46,920 --> 02:09:51,240
official one has a hundred plus participants now as far as I know and I think

1536
02:09:51,240 --> 02:09:55,760
they didn't run into any issues except for slowness it's have become slow with

1537
02:09:55,760 --> 02:09:59,720
the groups well I think that there is some issues with the the the size of the

1538
02:09:59,720 --> 02:10:05,760
payloads on relays okay because yeah oh yeah yeah there's I mean there's there's

1539
02:10:05,760 --> 02:10:11,680
definitely specialized specific relays now for it yeah yeah which at that

1540
02:10:11,680 --> 02:10:16,960
point yeah you can argue that it's not really but it's not much different

1541
02:10:16,960 --> 02:10:21,600
that a coordinator yeah yeah I see I see it's like it's all it's all trade-offs

1542
02:10:21,600 --> 02:10:26,400
all the time no solutions like my my point with with Cordn is that coordinators

1543
02:10:26,400 --> 02:10:32,480
are very easy to run it's much easier than relay and now we are

1544
02:10:32,480 --> 02:10:38,360
writing a multi-coordinator spec for Cordn which basically allow to define

1545
02:10:38,360 --> 02:10:43,160
coordinators for as fallback so you can have a switch over basically yeah so if

1546
02:10:43,160 --> 02:10:48,320
you coordinator or the coordinator that is born in a drought that's very exactly it's

1547
02:10:48,320 --> 02:10:52,560
have the ability to to jump to the next coordinator and continue the

1548
02:10:52,560 --> 02:10:57,560
conversation there and so interesting okay so the only point that everyone

1549
02:10:57,560 --> 02:11:03,080
have a against Cordn and they are out of criticizing about this is that is

1550
02:11:03,080 --> 02:11:07,520
centralized which is not because it's not centralized maybe you can argue that

1551
02:11:07,520 --> 02:11:12,280
there is a single point of failure which is a coordinator but it's not

1552
02:11:12,280 --> 02:11:15,920
centralized because everyone can run its own coordinator and they are

1553
02:11:15,920 --> 02:11:20,800
different implementation of coordinators indeed close father the guy behind

1554
02:11:20,800 --> 02:11:27,920
relay tools he's very into Cordn now he's writing he wrote a coordinator

1555
02:11:27,920 --> 02:11:32,720
implementation he's writing his own Cordn client which is called a

1556
02:11:32,720 --> 02:11:39,040
sarcase amethyst also implemented Cordn recently so it's a protocol and it's

1557
02:11:39,040 --> 02:11:48,480
free self-hostile what did amethyst not implement yeah I mean to be an

1558
02:11:48,480 --> 02:11:53,680
implementation that that amethyst is it about of Cordn is very limited so far

1559
02:11:53,680 --> 02:11:59,160
right now maybe because you just can't create groups there you can enjoy

1560
02:11:59,160 --> 02:12:04,520
other other groups but this was first implementation in any case yeah but

1561
02:12:04,520 --> 02:12:11,240
the ecosystem is thriving and also yeah you can host a coordinator

1562
02:12:11,240 --> 02:12:15,840
musicia than than a relay yeah because you can you can host it even in a

1563
02:12:15,840 --> 02:12:19,880
phone just dole adding an app and running it you don't need to configure

1564
02:12:19,880 --> 02:12:24,600
anything in the network nothing of that yeah yeah it's it's it's it's interesting

1565
02:12:24,600 --> 02:12:30,600
and it definitely speaks to what we talked before about kind of post-carsity

1566
02:12:30,600 --> 02:12:35,720
software you know like it's it's if you mentioned Cloudfather and the like

1567
02:12:35,720 --> 02:12:42,360
that he's implementing all that right now like the fact that the single

1568
02:12:42,360 --> 02:12:47,400
people can't just go ahead and create this like if there is a specification or

1569
02:12:47,400 --> 02:12:52,000
if there is some prior art just point your clankers at it and like 48 hours

1570
02:12:52,000 --> 02:12:56,040
later you have a solid implementation it is it's very interesting like I

1571
02:12:56,040 --> 02:13:00,600
had a process that Cloudfather is following because he has his clanker on

1572
02:13:00,600 --> 02:13:08,440
Cordn his clankers on Cordn and he in the in that channel in the group he

1573
02:13:08,440 --> 02:13:13,680
added multiple people including me so okay and you can just it's just clankers

1574
02:13:13,680 --> 02:13:19,680
yeah yeah that's where it's going if he's working on a feature I can you know

1575
02:13:19,680 --> 02:13:24,600
participate or if I discover an issue or someone discover an issue it can

1576
02:13:24,600 --> 02:13:29,440
use and mention the clanker and say I'm having this issue can you track it can

1577
02:13:29,440 --> 02:13:34,000
you trace it whatever and the clanker with do everything and come back into

1578
02:13:34,000 --> 02:13:38,840
this space that that's the way to do it I mean that's also that's also what

1579
02:13:38,840 --> 02:13:43,480
Jack tried to do or is doing with bus right and like all my projects that's

1580
02:13:43,480 --> 02:13:49,040
in part why all my projects have n-pubs because then you can DM them and I can

1581
02:13:49,040 --> 02:13:53,600
react to that you know and and just you can also tag them on us and all all my

1582
02:13:53,600 --> 02:13:57,680
projects are set up to to have like nightly cron jobs and if there is any

1583
02:13:57,680 --> 02:14:02,240
mention or any DM and it's also like if the if the app crashes for example like

1584
02:14:02,240 --> 02:14:06,440
Boris for example if it crashes it will detect the crash and the next time you

1585
02:14:06,440 --> 02:14:10,440
open it you can just press one button to send the crash report as a DM with an

1586
02:14:10,440 --> 02:14:16,040
ephemeral key you know like without toxin yourself yeah and and so so the software

1587
02:14:16,040 --> 02:14:20,320
is fixing itself basically I don't have to do anything it's just there's a

1588
02:14:20,320 --> 02:14:25,640
cron job at night get the DM like create an issue fix the issue yeah create a

1589
02:14:25,640 --> 02:14:29,520
poor request once everything screen merged yeah and I think that there is an

1590
02:14:29,520 --> 02:14:35,200
amazing way to collaborate like a new waves absolutely new ways to collaborate

1591
02:14:35,200 --> 02:14:40,640
absolutely yeah it's fucking crazy yeah it's a breaking world and I really

1592
02:14:40,640 --> 02:14:50,400
enjoy working in Cordn because I believe that it's a fairly private and secure

1593
02:14:50,400 --> 02:14:57,360
I mean MLS provide the security part and the privacy part is provided by by

1594
02:14:57,360 --> 02:15:04,800
the the ContextVM protocol itself which is very private in the sense of how

1595
02:15:04,800 --> 02:15:10,560
a request and responses are transported so everything is a familiar event

1596
02:15:10,560 --> 02:15:16,680
nothing is a storage store on release everything is transported with a

1597
02:15:16,680 --> 02:15:23,040
familiar keys and it's a double encrypted in the sense that the ContextVM

1598
02:15:23,040 --> 02:15:28,080
payloads when you post a message in a group it's encrypted in the context

1599
02:15:28,080 --> 02:15:34,080
being traffic so it's it's not there is no distinction between another

1600
02:15:34,080 --> 02:15:39,520
ContextVM traffic and Cordn traffic it all looks the same and then inside is

1601
02:15:39,520 --> 02:15:47,120
encrypted following the MLS yeah I see stuff I see yeah so it's very secure it's

1602
02:15:47,120 --> 02:15:52,520
very private it's a self-hostable so the idea is that no one can take down your

1603
02:15:52,520 --> 02:15:56,600
communications yeah yeah because this is this is an interesting reflection

1604
02:15:56,600 --> 02:16:01,400
and yeah it comes back to control right yeah exactly like if you have

1605
02:16:01,400 --> 02:16:06,640
relays and relays are an amazing piece of software and very interesting idea but

1606
02:16:06,640 --> 02:16:11,240
you don't run it yourself you are not sobering if you rely on someone

1607
02:16:11,240 --> 02:16:19,200
who is a infractor you are delegating on them the control of that but if it

1608
02:16:19,200 --> 02:16:24,000
said you are able to do your bid your own infrastructure and reduce the

1609
02:16:24,000 --> 02:16:29,480
bar to run it and to deploy it and to to maintain it you're more sobering

1610
02:16:29,480 --> 02:16:32,720
yeah so if you are using the relay of someone else for your

1611
02:16:32,720 --> 02:16:37,680
communications tomorrow they can disappear yeah and you have a lot of it again

1612
02:16:37,680 --> 02:16:41,520
it's a different trade off right because it's also it's also running

1613
02:16:41,520 --> 02:16:45,360
infrastructure is it's a burden and a liability basically yeah so you don't

1614
02:16:45,360 --> 02:16:49,880
necessarily always want to do it but if you have the means to do it and the

1615
02:16:49,880 --> 02:16:54,360
world to do it and like we said before you know like it's if don't run

1616
02:16:54,360 --> 02:16:57,720
something in a T that you could run on your own computer for example it's a

1617
02:16:57,720 --> 02:17:05,680
similar idea but yeah no it's to me one one side effect I think of the like

1618
02:17:05,680 --> 02:17:09,240
we're definitely in a Cambrian explosion now when it comes to software in

1619
02:17:09,240 --> 02:17:13,080
general and and and protocols and just because it's so easy to create it's

1620
02:17:13,080 --> 02:17:20,040
just we have this insane proliferation yeah so I I think that yeah the

1621
02:17:20,040 --> 02:17:26,680
cream will rise to the top like in some sense like whatever works best will

1622
02:17:26,680 --> 02:17:32,880
just survive like we're in a massive evolutionary process and software always

1623
02:17:32,880 --> 02:17:38,880
was in this evolutionary process but now we we're like software is more like

1624
02:17:38,880 --> 02:17:42,600
mosquitoes you know which just like to create a million and maybe like five

1625
02:17:42,600 --> 02:17:47,160
some might yeah and and yeah we'll see we'll see where we are in a year from

1626
02:17:47,160 --> 02:17:52,520
now yeah yeah so I think that everything is everything is thumbs up right like

1627
02:17:52,520 --> 02:17:59,040
everything it's a expression everything takes different or give different

1628
02:17:59,040 --> 02:18:05,120
answers to the problems and that's that's how we advance yeah that's how we

1629
02:18:05,120 --> 02:18:08,480
figured it out stuff yeah it's not a single way to do stuff that is

1630
02:18:08,480 --> 02:18:14,880
million ways to do stuff everyone is exploring yeah so that's yeah we're

1631
02:18:14,880 --> 02:18:18,560
gonna continue exploring that's the last thing I want to mention next year what

1632
02:18:18,560 --> 02:18:22,360
are your plans for next year because I can tell you I can tell you mine but

1633
02:18:22,360 --> 02:18:27,520
what are you up to next year are you gonna are you gonna stick around these

1634
02:18:27,520 --> 02:18:32,280
areas are you gonna are you gonna are you gonna defend your position on this

1635
02:18:32,280 --> 02:18:36,080
sovereign engineering leader board you're you're you're the person with the

1636
02:18:36,080 --> 02:18:39,240
most hours inside of sovereign engineering all of all people in the world

1637
02:18:39,240 --> 02:18:43,840
Shannon as well or colorless sorry you're you're I think you're still beating

1638
02:18:43,840 --> 02:18:53,080
him maybe yeah I mean I love to be to be here in the island I love the routine

1639
02:18:53,080 --> 02:18:56,840
that I'm developing here it's probably gonna be around for some time

1640
02:18:56,840 --> 02:19:03,040
probably also go through the world and do some aspiration yeah so we want to do

1641
02:19:03,040 --> 02:19:06,920
we want to do the same on the sovereign engineering side so we're gonna do we

1642
02:19:06,920 --> 02:19:13,600
just we just kind of locked in the dates for next year just two days ago so we're

1643
02:19:13,600 --> 02:19:20,120
gonna do a summer cohort again six weeks just like we did this year and will we're

1644
02:19:20,120 --> 02:19:26,660
gonna do one in spring shorter one probably three weeks yeah probably focused on

1645
02:19:26,660 --> 02:19:35,700
tips and yeah similar things around that so exploring the the machine networking

1646
02:19:35,700 --> 02:19:40,420
world the transport layer world the kind of permissionless connectivity world

1647
02:19:40,420 --> 02:19:47,500
some more and what I really want to do is what I'm gonna call a mountain cohort

1648
02:19:47,500 --> 02:19:54,980
for now so yeah we had this idea actually like in SCCO 8 like in the last

1649
02:19:54,980 --> 02:20:02,420
court we talked about kind of how can we branch out and yeah get kind of

1650
02:20:02,420 --> 02:20:08,820
more people involved get some some like do things a little bit differently and

1651
02:20:08,820 --> 02:20:16,860
so the idea was thrown around to do like a very long residency somewhere else I

1652
02:20:16,860 --> 02:20:22,180
talked about this with Aryan the last time and so there's an update on that

1653
02:20:22,180 --> 02:20:25,140
that's probably not gonna happen this year but it's probably gonna happen next

1654
02:20:25,140 --> 02:20:30,620
year so it was a rush it was too rushed but what I what I really want to do is

1655
02:20:30,620 --> 02:20:34,940
like a mountain cohort which is basically just one week like if we do a six

1656
02:20:34,940 --> 02:20:38,900
week cohort we have six weekly cycles yeah and I want to experiment with just like

1657
02:20:38,900 --> 02:20:45,700
one weekly cycle very similar in the setup as we have here but still different

1658
02:20:45,700 --> 02:20:48,940
enough and adjusted to the mountain and what I really want to experiment with

1659
02:20:48,940 --> 02:20:53,900
is something we touched on before with like like being permanently online what

1660
02:20:53,900 --> 02:20:58,380
I want to do is take away everyone's devices for at least one night maybe two

1661
02:20:58,380 --> 02:21:02,380
nights you know we'll see we'll see how it goes and literally go up a mountain and

1662
02:21:02,380 --> 02:21:06,300
probably we're gonna do it in mainland Europe because getting to Madera is

1663
02:21:06,300 --> 02:21:10,060
really hard it's like it's it's a big ask so to speak it's very very hard for

1664
02:21:10,060 --> 02:21:14,340
people to to drop everything and and come here for six weeks like that's the main

1665
02:21:14,340 --> 02:21:19,060
issue and so I think it should be easier for our people to just you know jump

1666
02:21:19,060 --> 02:21:22,580
into a train or jump into the car and go to whatever north in Italy or

1667
02:21:22,580 --> 02:21:31,380
something and and spend a week with yeah just with like-minded people exploring

1668
02:21:31,380 --> 02:21:36,140
all the stuff that we talk about here and all the stuff that some engineering is

1669
02:21:36,140 --> 02:21:40,300
about and so the goal would be to just you know meet in the valley kind of

1670
02:21:40,300 --> 02:21:44,140
everyone talks about what they're working on what their head is at what they're

1671
02:21:44,140 --> 02:21:47,580
interested in like layout the building blocks so to speak as well like we

1672
02:21:47,580 --> 02:21:51,140
talked about this a lot in the at the end of the court like we basically have

1673
02:21:51,140 --> 02:21:54,500
all the building blocks now we have all we have all the stuff that we need to

1674
02:21:54,500 --> 02:21:58,340
actually you know move the needle and rebuild lots of things in a more

1675
02:21:58,340 --> 02:22:02,740
self-serving way and and so kind of lay all that out in the the first one or two

1676
02:22:02,740 --> 02:22:08,820
days of the week and then literally walk up their mountain and spend a couple

1677
02:22:08,820 --> 02:22:11,860
nights on the mountain like two or three nights on the mountain and again like

1678
02:22:11,860 --> 02:22:17,820
I really want to do I really want to do like a period of time completely offline

1679
02:22:17,820 --> 02:22:23,100
like just bring pen and paper nothing else legit walk up like there's some

1680
02:22:23,100 --> 02:22:27,180
some really cool mountain huts in that region some just very basic like you

1681
02:22:27,180 --> 02:22:32,460
you're you're in the elements yeah you're surviving on the mountain and then

1682
02:22:32,460 --> 02:22:35,700
you're coming down from the mountain and what usually happens I mean that's

1683
02:22:35,700 --> 02:22:39,540
also the whole point of serving engineering is like if you put people in the

1684
02:22:39,540 --> 02:22:44,820
same room for prolonged periods of time like new ideas will emerge it's like

1685
02:22:44,820 --> 02:22:48,900
you mentioned to them you know it was just a crazy idea like at the barbecue

1686
02:22:48,900 --> 02:22:52,300
and like all those kind of things and that's how that's how stuff happens that's

1687
02:22:52,300 --> 02:22:56,580
how how how the brain chooses actually start flowing and then come down from

1688
02:22:56,580 --> 02:23:01,860
the mountain and like give everyone 24 hours to implement because clankers you

1689
02:23:01,860 --> 02:23:07,300
know like you can actually you can actually do stuff in 24 and then end on one big

1690
02:23:07,300 --> 02:23:12,380
demo day basically that's the idea so that's how I want to structure the seven

1691
02:23:12,380 --> 02:23:17,820
days who's gonna bring the point yeah we'll see we'll see we you know in the

1692
02:23:17,820 --> 02:23:21,580
mountain regions they have their own stuff we're we're probably gonna stay

1693
02:23:21,580 --> 02:23:26,860
local and the physicists and drink some drinks and proper like pine

1694
02:23:26,900 --> 02:23:32,700
snobs or something like that yeah it's similar to the Chinese stuff that the

1695
02:23:32,700 --> 02:23:38,940
cherry introduced yeah very strong very good and yeah we'll get the brain

1696
02:23:38,940 --> 02:23:43,020
chooses flowing for sure and yeah that's that's what we want to experiment with

1697
02:23:43,020 --> 02:23:47,220
next year we actually we're probably gonna do two mountain courts one is

1698
02:23:47,220 --> 02:23:51,260
spring and one in fall so that's the idea and hopefully people that actually

1699
02:23:51,260 --> 02:23:54,940
have families can join us as well because you know getting away from the

1700
02:23:54,940 --> 02:24:01,620
family from the wife and kids for for like a week is doable bringing them all

1701
02:24:01,620 --> 02:24:07,460
or like disappearing for like six weeks at a time it's really really hard so yeah

1702
02:24:07,460 --> 02:24:13,780
we want to kind of transform or like evolve the program a little bit to yeah

1703
02:24:13,780 --> 02:24:17,500
make it more appealing for yeah people that actually have families and so I mean

1704
02:24:17,500 --> 02:24:25,500
makes sense all the SZ have been evolving towards this shorter-term periods

1705
02:24:25,500 --> 02:24:31,780
right like the first one was eight weeks yeah eight weeks of hell that's crazy

1706
02:24:31,780 --> 02:24:37,420
and then I think that the next one I participated was seven weeks with a

1707
02:24:37,420 --> 02:24:43,340
week in the middle yeah of a three week and then six weeks yeah which is very

1708
02:24:43,340 --> 02:24:46,780
good time and we're gonna break we're gonna bring the break week back for the

1709
02:24:46,780 --> 02:24:52,780
next one let's go yeah we're we got we got lots of complaints for related to

1710
02:24:52,780 --> 02:24:58,900
burnout so so we're gonna we're gonna take the photo of the paddle a little bit

1711
02:24:58,900 --> 02:25:05,380
you know and introduce some some some room to breathe as well yeah all right I

1712
02:25:05,380 --> 02:25:10,580
think that people have different different kind of speeds or something like

1713
02:25:10,580 --> 02:25:17,860
for sure it's like in all there this is that I have participated participate some

1714
02:25:17,860 --> 02:25:26,020
people never burn out and they were always bubbling yeah and other people like

1715
02:25:26,020 --> 02:25:32,100
a burnout second week yeah and they never were we're able to recover I think it's

1716
02:25:32,100 --> 02:25:35,660
mostly an introvert extrovert distinction it's like if if if you're very

1717
02:25:35,660 --> 02:25:39,020
introverted and and other people drain you then then you'll be done very soon

1718
02:25:39,020 --> 02:25:43,980
yeah yeah it's not homogeneous that that's the interesting part yeah we'll be

1719
02:25:43,980 --> 02:25:52,620
interesting to maybe this short short periods are interesting to also

1720
02:25:52,620 --> 02:25:59,660
understand if it's makes the experience more homogeneous for everyone yeah we'll

1721
02:25:59,660 --> 02:26:03,740
see we'll see we'll keep experimenting I definitely want to keep doing it for

1722
02:26:03,820 --> 02:26:09,500
for as long as we can kind of like it's again I think this this this I love this

1723
02:26:09,500 --> 02:26:13,900
conversation because it just went in in in 15 different places and that's all

1724
02:26:13,900 --> 02:26:19,020
the it's all the the stuff that we kind of have to figure out you know like and

1725
02:26:19,020 --> 02:26:23,900
and I mean by we I mean like as humanity will have to figure this out but also as

1726
02:26:23,900 --> 02:26:30,140
like the people working on this stuff like it's it's very hard to put a label on

1727
02:26:30,140 --> 02:26:33,420
it like it's you know some people would would say that the freedom tech community

1728
02:26:33,420 --> 02:26:40,940
or or something like that um you know like those are really hard problems yeah to

1729
02:26:40,940 --> 02:26:47,100
figure out and yeah and I completely agree with what you said before where it's

1730
02:26:47,100 --> 02:26:54,060
it it it takes conversation it takes dialogue it takes time also and it takes exactly

1731
02:26:54,060 --> 02:27:01,180
takes this participatory nature um participatory knowing that that Vervaeke talks about um

1732
02:27:01,180 --> 02:27:07,180
to figure that out like you also have to actually use that stuff all the time like you actually

1733
02:27:07,180 --> 02:27:11,340
have to not only build it and build apps and don't use them or something like that you actually

1734
02:27:11,340 --> 02:27:19,580
have to yeah you use all this stuff as much as you can to figure out how it should evolve yeah

1735
02:27:19,580 --> 02:27:28,300
you know and yeah and so so there is a lot of reason to keep this going and um I'm very

1736
02:27:29,500 --> 02:27:36,620
yeah I'm I'm very excited about the next one or two or three years because we have a lot of

1737
02:27:36,620 --> 02:27:44,460
stuff in place right right now we have you know like agents warms that can help us build all

1738
02:27:44,460 --> 02:27:53,980
that out and um yeah the the future is very bright if if we want it to be it really is I really

1739
02:27:53,980 --> 02:28:01,580
believe that like I I really believe that technology can be extremely empowering and extremely

1740
02:28:01,580 --> 02:28:07,180
liberating just like Hal Finney talked about it right like the the computer can be forced for good

1741
02:28:07,180 --> 02:28:13,980
and and like a tool that can be used to actually liberate you and and right now it's still mostly

1742
02:28:13,980 --> 02:28:21,100
used to control people and to yeah you know enslave them in some way and so yeah I do think

1743
02:28:21,100 --> 02:28:25,420
the future is very bright we just have to get our own shit together and also be very clear

1744
02:28:25,420 --> 02:28:31,980
in terms of like why are we doing this like what why why like what is what is the goal of building

1745
02:28:31,980 --> 02:28:37,100
all that and I think that we have to be actively participating on it yeah and not being

1746
02:28:37,100 --> 02:28:44,780
passive on it because we are redefining or right now everything is being redefined not just

1747
02:28:44,780 --> 02:28:50,780
because LLMs not because the state of the war and so on LLMs are a big part of of all of this

1748
02:28:50,780 --> 02:29:00,860
and and this change but I don't know society is very passive and oh yeah kind of look everything

1749
02:29:00,940 --> 02:29:07,340
like if it was a film oh yeah like an expectator no someone actually has to fix things yeah we have

1750
02:29:07,340 --> 02:29:12,620
to be active participators in this reality to make sense of it yeah because as you say like

1751
02:29:12,620 --> 02:29:18,780
as someone in in their in their room like they kind of make sense that's not a way to make

1752
02:29:18,780 --> 02:29:23,980
sense about anything like a single person cannot do anything yeah no it's to be something

1753
02:29:24,460 --> 02:29:32,460
that's also like you know I hope that everyone who like I'll reach out to our alumni

1754
02:29:32,460 --> 02:29:39,340
sense on and and I hope that we can get more people involved especially like high-agency people

1755
02:29:39,340 --> 02:29:44,060
I'll use Justin as an example you know like when he was here and just kind of like fooling around

1756
02:29:44,060 --> 02:29:51,340
he ended up basically rebuilding an android operating system from scratch you know but you're

1757
02:29:51,340 --> 02:29:59,100
laughing because it's such a ridiculous thing you know it's such a big task but my point is

1758
02:29:59,100 --> 02:30:03,740
you can actually do it you can actually do it yeah like if Justin can do it in like a couple of

1759
02:30:03,740 --> 02:30:08,140
weeks if you put your mind to it and you work on it for like a year or two or whatever and get

1760
02:30:08,140 --> 02:30:12,140
some people around you you can actually build something like a GrapheneOS and again like

1761
02:30:12,220 --> 02:30:18,940
GrapheneOS is great but we we like I think now it's the time to actually also

1762
02:30:20,220 --> 02:30:26,460
be bolder and think bigger and I hope that we'll be able to get more high-agency people

1763
02:30:27,260 --> 02:30:34,220
involved that agree with this statement where it's actually like again that's why I have to bring

1764
02:30:34,220 --> 02:30:38,140
up DHH again like like he's actually in a high-agency person and he was like I'm going to I'm just

1765
02:30:38,140 --> 02:30:43,420
going to build my own computer period and no one is going to tell me no and and I think we'll have

1766
02:30:43,420 --> 02:30:52,620
to what has me extremely hopeful and optimistic is that over the course of the last two or three

1767
02:30:52,620 --> 02:30:56,780
years like like sovereign engineers very very young it's all very very new we were able to

1768
02:30:56,780 --> 02:31:02,460
rebuild large parts of the internet stack to some degree you know like I'm I'm take everything I

1769
02:31:02,460 --> 02:31:07,900
say with the yeah with a grain of salt but but but it's still it's it's amazing what we have

1770
02:31:08,540 --> 02:31:15,260
and we actually have our own you know like self sovereign infrastructure you just you just have

1771
02:31:15,260 --> 02:31:21,180
to host it you just have to run it and you can actually do lots of stuff yeah yeah that not too long

1772
02:31:21,180 --> 02:31:26,540
ago you were fully reliant on for example twitter to do all that for you or or what have you

1773
02:31:26,540 --> 02:31:31,900
and so yeah again I think the future future is bright we just have to we have to just have to

1774
02:31:31,900 --> 02:31:38,540
get after it yeah yeah let's end here yeah I'm all out of conversation that was great thanks so

1775
02:31:38,540 --> 02:31:46,620
much now let's have let's have their coffee I think I'll be at this point it's a beer time it's

1776
02:31:46,620 --> 02:31:57,660
beer time it's it's it's right after breakfast that's a great that's a great breakfast you can have

1777
02:31:57,660 --> 02:32:02,220
a German breakfast it's like it's like a large beer in the vice-war store or something