1
00:00:00,000 --> 00:00:07,360
Okay, good.

2
00:00:07,360 --> 00:00:11,320
So I've been listening to some of your past conversations.

3
00:00:11,320 --> 00:00:16,440
I didn't even know that you were talking about this stuff publicly already.

4
00:00:16,440 --> 00:00:18,240
Yeah, that's a twister.

5
00:00:18,240 --> 00:00:20,560
Yeah, I would say that looks a lot more twister.

6
00:00:20,560 --> 00:00:24,440
There's rain on the ocean and there's like a...

7
00:00:24,440 --> 00:00:25,440
How do you call it?

8
00:00:25,440 --> 00:00:26,440
The water hose, basically?

9
00:00:26,440 --> 00:00:29,160
Yeah, I think it's a water hose when we're over the water hose.

10
00:00:29,440 --> 00:00:36,120
So it's sucking up water so that it can rain down on us later.

11
00:00:36,120 --> 00:00:38,480
I've always thought of seeing those on land.

12
00:00:38,480 --> 00:00:39,640
Yeah, but that looks wild.

13
00:00:39,640 --> 00:00:43,360
It's very, very far away, so it's very, very big, actually.

14
00:00:43,360 --> 00:00:44,840
Yeah, yeah.

15
00:00:44,840 --> 00:00:45,880
That looks really cool.

16
00:00:45,880 --> 00:00:46,400
It's nuts.

17
00:00:46,400 --> 00:00:48,520
Yeah, let me try to...

18
00:00:48,520 --> 00:00:51,960
Maybe I can snap a picture that looks like something.

19
00:00:51,960 --> 00:00:53,320
It definitely doesn't look like a cloud.

20
00:00:53,320 --> 00:00:56,760
We can post it later.

21
00:00:56,760 --> 00:00:58,360
And it's transforming.

22
00:00:59,240 --> 00:01:01,800
Looks like it's going out now, it's going up off the water.

23
00:01:01,800 --> 00:01:07,320
Yeah, the weather is something I didn't factor into the equation of, you know,

24
00:01:07,320 --> 00:01:10,840
just taking developers for a walk and talking with them for an hour.

25
00:01:10,840 --> 00:01:13,480
I was like, oh, it's always going to be sunny and perfect,

26
00:01:13,480 --> 00:01:18,840
but we had hurricanes, basically, and rain, lots of rain.

27
00:01:18,840 --> 00:01:21,480
So I didn't think, I didn't think of this through.

28
00:01:21,480 --> 00:01:26,680
Just being outside in general.

29
00:01:26,680 --> 00:01:33,240
But yeah, it's the first kind of conversation where I prepared a little bit,

30
00:01:33,240 --> 00:01:38,120
so I listened to some of your past conversations where you're trying to

31
00:01:38,120 --> 00:01:40,840
explain Noster to the podcasting to point out people.

32
00:01:40,840 --> 00:01:42,680
That's probably good because I don't remember this.

33
00:01:42,680 --> 00:01:44,920
What I talked about a lot of those conversations.

34
00:01:44,920 --> 00:01:51,800
It was really enlightening in some sense because I've been so embedded on the Noster

35
00:01:51,800 --> 00:01:56,440
side that I don't even think about it anymore in some sense.

36
00:01:57,000 --> 00:01:59,880
How it is different from other things.

37
00:01:59,880 --> 00:02:01,240
You get used to it fast.

38
00:02:01,240 --> 00:02:05,560
Yeah, and we mentioned this yesterday, I think, very briefly.

39
00:02:05,560 --> 00:02:06,520
Or was it two days ago?

40
00:02:06,520 --> 00:02:09,000
I think it was two days ago where it's just like, okay,

41
00:02:09,000 --> 00:02:13,400
there's a small little difference, but it actually makes a huge difference.

42
00:02:13,400 --> 00:02:14,760
It has a huge effect.

43
00:02:14,760 --> 00:02:18,200
And the small little difference is basically that the data is signed.

44
00:02:18,200 --> 00:02:22,120
And so that moves the power from the servers,

45
00:02:22,120 --> 00:02:24,920
which usually holds the power, to the key holders.

46
00:02:24,920 --> 00:02:28,280
And that's also why the stuff can be in 100 different places.

47
00:02:28,280 --> 00:02:31,160
Because you have assurances that it's the same thing.

48
00:02:31,800 --> 00:02:36,280
Yeah, it opens up a weird, it actually moves the data away from the service,

49
00:02:36,280 --> 00:02:39,000
which doesn't sound like much.

50
00:02:39,000 --> 00:02:40,760
But then it lets you...

51
00:02:40,760 --> 00:02:42,280
It sends the data free in some sense.

52
00:02:42,280 --> 00:02:45,640
Yeah, which then brings up the most really difficult question,

53
00:02:45,640 --> 00:02:46,600
and really interesting questions.

54
00:02:46,600 --> 00:02:48,280
Do you need the service anymore?

55
00:02:48,280 --> 00:02:49,800
Which is like something we've never...

56
00:02:49,800 --> 00:02:52,040
Like literally since the beginning, the internet's always been...

57
00:02:52,040 --> 00:02:54,360
What's the point of the service in some sense?

58
00:02:54,360 --> 00:02:59,000
Because historically, because I still think we need services in some sense.

59
00:02:59,000 --> 00:03:00,920
And we also, we still need servers.

60
00:03:00,920 --> 00:03:04,120
Like there was this whole discussion, like our relay still servers,

61
00:03:04,120 --> 00:03:04,920
and those kind of things.

62
00:03:04,920 --> 00:03:07,320
And you were like, yeah, kind of, but also not in some sense.

63
00:03:07,320 --> 00:03:08,680
I mean, it was really interesting.

64
00:03:08,680 --> 00:03:12,280
It's always convenient to run code faster on someone else's machine.

65
00:03:12,280 --> 00:03:12,760
Yeah.

66
00:03:12,760 --> 00:03:13,880
So like that's a service.

67
00:03:14,600 --> 00:03:16,120
You will have specialisation.

68
00:03:16,120 --> 00:03:18,280
A server is basically a specialized computer,

69
00:03:18,280 --> 00:03:20,120
and the relay is also a specialized computer.

70
00:03:20,120 --> 00:03:21,000
Yeah, yeah.

71
00:03:21,000 --> 00:03:23,080
But they are different in nature a little bit.

72
00:03:23,160 --> 00:03:26,040
And it all relates to cryptographic signatures and identities.

73
00:03:26,040 --> 00:03:29,160
It's like, basically, Pablo and I talked about this a lot,

74
00:03:29,160 --> 00:03:31,240
but in a kind of a different context.

75
00:03:31,240 --> 00:03:34,280
And the context is like, you don't care.

76
00:03:34,280 --> 00:03:37,800
Like in the nostril world, you don't have to care where the stuff comes from.

77
00:03:37,800 --> 00:03:39,080
And you mentioned this briefly as well,

78
00:03:39,080 --> 00:03:40,680
where it's like, it can come from a USB stick.

79
00:03:40,680 --> 00:03:41,640
It doesn't matter.

80
00:03:41,640 --> 00:03:42,680
You know, it can come from...

81
00:03:43,880 --> 00:03:45,640
I don't care where the data comes from,

82
00:03:45,640 --> 00:03:49,320
as long as it lands on my client or lands on my machine,

83
00:03:49,320 --> 00:03:52,280
and I can check the signatures, then we're fine.

84
00:03:52,280 --> 00:03:56,200
Then I know I don't care about the source in some sense.

85
00:03:56,200 --> 00:03:57,400
And it means the same thing.

86
00:03:57,400 --> 00:03:58,360
It means the same thing.

87
00:03:58,360 --> 00:04:00,600
It's like you said.

88
00:04:03,800 --> 00:04:05,240
That's why you can set the data free.

89
00:04:05,240 --> 00:04:07,480
That's the mechanism that sets the data free in some sense.

90
00:04:08,200 --> 00:04:09,640
But you still need...

91
00:04:12,120 --> 00:04:13,640
I mean, we see this already.

92
00:04:13,640 --> 00:04:14,920
We still have...

93
00:04:14,920 --> 00:04:17,000
Relay still have centralising effects, of course.

94
00:04:17,000 --> 00:04:20,520
It's still convenient to have stuff in one place

95
00:04:20,520 --> 00:04:22,360
for indexing and search and other things.

96
00:04:22,360 --> 00:04:23,560
Really indexing and search,

97
00:04:23,560 --> 00:04:27,560
but also just because latency is a nice thing to have.

98
00:04:27,560 --> 00:04:29,160
Like low latency is a nice thing to have.

99
00:04:29,160 --> 00:04:30,760
Because if you have to wait, you know,

100
00:04:31,720 --> 00:04:34,360
if we were synchronizing large amounts of events every week,

101
00:04:34,360 --> 00:04:36,920
it would be inconvenient to only get an update

102
00:04:36,920 --> 00:04:39,640
from people you follow, like every week with all their notes.

103
00:04:39,640 --> 00:04:42,120
Yeah, and we see the specialization in a really,

104
00:04:42,120 --> 00:04:43,960
relay world as well right now, right?

105
00:04:43,960 --> 00:04:46,440
I mean, we have different kinds of relays and so on.

106
00:04:46,440 --> 00:04:47,560
What's your take on all that?

107
00:04:47,560 --> 00:04:48,840
How do you think about this?

108
00:04:49,480 --> 00:04:51,640
This evolution where we have like...

109
00:04:51,640 --> 00:04:52,680
I mean, if you open anethys,

110
00:04:52,680 --> 00:04:56,040
you have like six different ways of setting a relay.

111
00:04:56,040 --> 00:04:57,560
Yeah, the configuration page.

112
00:04:57,560 --> 00:04:59,320
I don't know anybody who actually understands that.

113
00:05:02,120 --> 00:05:03,640
Yeah, the special is really...

114
00:05:03,640 --> 00:05:05,560
I mean, it's the natural way of...

115
00:05:05,560 --> 00:05:07,320
The natural way things are going to go

116
00:05:07,320 --> 00:05:09,800
because you want to make everything...

117
00:05:09,800 --> 00:05:11,080
It starts out as a dumb relay,

118
00:05:11,080 --> 00:05:14,200
and then eventually you want to start...

119
00:05:14,200 --> 00:05:15,880
If nothing else, you want to start like

120
00:05:15,880 --> 00:05:17,640
categorizing the data, like putting stuff...

121
00:05:17,640 --> 00:05:19,720
Like, the inbox and the box relay is a good example of that.

122
00:05:19,720 --> 00:05:21,800
Like, what goes in the inbox relay is stuff

123
00:05:21,800 --> 00:05:24,440
that other people want me to care about.

124
00:05:24,440 --> 00:05:26,920
What goes in my out box relay is what I want them to see.

125
00:05:27,720 --> 00:05:30,760
So that's already like two specialized categories right there.

126
00:05:32,440 --> 00:05:34,360
Another one would be your draft,

127
00:05:34,360 --> 00:05:36,280
your private things, and so on.

128
00:05:36,280 --> 00:05:37,800
I'm still not sold on that.

129
00:05:37,800 --> 00:05:38,520
I don't know about that.

130
00:05:40,520 --> 00:05:41,400
It's like the DMs.

131
00:05:41,400 --> 00:05:43,400
Like, it's everything under one key,

132
00:05:43,400 --> 00:05:46,280
and then it's a lot of...

133
00:05:46,360 --> 00:05:47,880
Because I'm kind of in the mindset of...

134
00:05:51,240 --> 00:05:53,640
I've been thinking recently about the encryption...

135
00:05:55,800 --> 00:05:58,680
Encryption and decryption functions in Noster.

136
00:05:59,480 --> 00:06:01,080
And the signing functions by extension.

137
00:06:01,080 --> 00:06:02,840
Kind of like the most expensive thing we have.

138
00:06:04,040 --> 00:06:06,920
And the payoff doesn't seem worth it in my mind.

139
00:06:06,920 --> 00:06:07,720
To be...

140
00:06:07,720 --> 00:06:10,280
To have a draft system where you're constantly

141
00:06:10,280 --> 00:06:12,440
doing an encryption and signing.

142
00:06:12,440 --> 00:06:13,000
Yeah.

143
00:06:13,000 --> 00:06:14,760
Just to like save something.

144
00:06:15,400 --> 00:06:17,560
When we have local storage, we have personal servers.

145
00:06:20,200 --> 00:06:22,600
I'm just thinking about it in the context of the remote signers.

146
00:06:22,600 --> 00:06:25,080
Trying to make remote signers actually viable,

147
00:06:25,080 --> 00:06:27,960
or offline signers, or hardware signers, anything like that.

148
00:06:27,960 --> 00:06:31,720
The issue with the local storage is it doesn't follow you around.

149
00:06:31,720 --> 00:06:35,080
So if I open a different device and I want to continue working on a draft,

150
00:06:35,080 --> 00:06:35,720
what do I do?

151
00:06:37,480 --> 00:06:40,840
It depends on how you manage your local storage.

152
00:06:40,840 --> 00:06:43,640
Because local storage can be something like the browser local storage,

153
00:06:43,640 --> 00:06:45,080
which doesn't follow you.

154
00:06:45,080 --> 00:06:48,280
Or a documents folder that synchronize with iCloud.

155
00:06:48,280 --> 00:06:50,520
Oh yeah, I mean, you're a sick thing guy as well.

156
00:06:51,480 --> 00:06:53,560
I mean, I could make this work.

157
00:06:55,240 --> 00:06:56,840
I think that's a big ask.

158
00:06:57,880 --> 00:06:59,400
Do the synchronization yourself.

159
00:07:01,480 --> 00:07:04,200
I mean, to me, that's part of the magic of Noster.

160
00:07:04,200 --> 00:07:06,360
That all the stuff follows you around.

161
00:07:07,320 --> 00:07:09,400
And in the case of Nip 60 and those kind of things,

162
00:07:09,400 --> 00:07:11,960
including your money and everything else.

163
00:07:12,040 --> 00:07:15,800
I think it is a very powerful thing to have,

164
00:07:16,520 --> 00:07:21,400
that you don't need to think about anything in some sense.

165
00:07:21,400 --> 00:07:27,320
You just log in with your Noster identity and all the stuff is there.

166
00:07:27,320 --> 00:07:30,360
I think that is a massive, massive killer feature.

167
00:07:30,360 --> 00:07:36,840
I think every step towards destroying some of that is in some sense a step in the right,

168
00:07:36,840 --> 00:07:37,720
in the wrong direction.

169
00:07:38,840 --> 00:07:39,640
That's how I see it.

170
00:07:39,640 --> 00:07:41,480
But maybe, you know, like, I don't know.

171
00:07:41,480 --> 00:07:44,360
The drafts, I guess the drafts are useful.

172
00:07:44,360 --> 00:07:48,680
Because I have, I have used them like the article drafts popping up on various devices.

173
00:07:49,400 --> 00:07:57,720
But I guess they're just, they're not implemented in any good way right now.

174
00:07:59,240 --> 00:08:04,600
I guess it's my complaint with a lot of a lot of existing Noster implementations and features

175
00:08:04,600 --> 00:08:08,760
and clients is it's implemented in such a way that the user doesn't see it.

176
00:08:10,280 --> 00:08:11,800
That's why I vote Noster.

177
00:08:13,720 --> 00:08:18,200
I can't trust the drafts because I can't see that it actually saved it.

178
00:08:18,200 --> 00:08:20,040
It just says, I did something.

179
00:08:21,400 --> 00:08:23,800
I don't know, did it do something and get to the relay?

180
00:08:24,600 --> 00:08:25,960
I can't trust the system yet.

181
00:08:28,600 --> 00:08:34,120
You spoke about this in the past, kind of, you know, how UIs and systems in general

182
00:08:34,120 --> 00:08:36,040
like should never ever lie to you.

183
00:08:36,040 --> 00:08:39,080
And if they don't lie to you for a long time,

184
00:08:39,080 --> 00:08:41,400
that's what builds up the trust in the system in some sense.

185
00:08:42,040 --> 00:08:42,920
Is that part of it?

186
00:08:42,920 --> 00:08:47,960
Like, do you want to, do you want to, like, you're very adamant about

187
00:08:49,160 --> 00:08:53,560
showing the users what happens underneath the hood in some sense.

188
00:08:55,080 --> 00:08:55,560
Yeah.

189
00:08:55,560 --> 00:08:57,480
Do you think it's important for most users?

190
00:08:57,480 --> 00:09:00,040
Or is it more personal preference of viewers?

191
00:09:00,040 --> 00:09:02,040
Or how do you think about it?

192
00:09:02,040 --> 00:09:07,400
I think it's necessary for all applications.

193
00:09:08,280 --> 00:09:13,400
It's just how you show it can change based on, you know,

194
00:09:13,400 --> 00:09:17,240
how you might say, know me friendly or how slick you want the UI to look.

195
00:09:17,880 --> 00:09:22,440
But I haven't been doing web development very long.

196
00:09:22,440 --> 00:09:25,960
But the little I've learned so far, you know,

197
00:09:25,960 --> 00:09:30,040
because I came from the applications I was working on before Noster,

198
00:09:30,040 --> 00:09:37,320
were like internal company applications for managing like financial stuff,

199
00:09:37,320 --> 00:09:38,840
like internal company financial stuff.

200
00:09:38,840 --> 00:09:42,600
So it was like the auditors and the dashboards and things.

201
00:09:43,240 --> 00:09:44,360
Yeah, dashboards and stuff.

202
00:09:44,360 --> 00:09:46,760
People like looking to like, you know, what are the customers,

203
00:09:46,760 --> 00:09:49,240
OS, what money do we owe them and stuff like that.

204
00:09:49,240 --> 00:09:55,560
And so it kind of developed a way of like you never,

205
00:09:55,560 --> 00:09:57,480
even if you have to do another HTTP request,

206
00:09:57,480 --> 00:10:01,880
even if you do another API request, you never show cash data,

207
00:10:01,880 --> 00:10:03,320
you never show inaccurate data.

208
00:10:04,360 --> 00:10:07,320
And if it's stuck, you show something.

209
00:10:07,320 --> 00:10:08,680
It's like, it's not, it's loading.

210
00:10:08,680 --> 00:10:11,320
It's taking longer because the users would like,

211
00:10:11,320 --> 00:10:14,040
they're not technical users and they need to know what the like,

212
00:10:15,000 --> 00:10:17,400
they trust the system entirely because it's their job to do that.

213
00:10:17,400 --> 00:10:17,960
Yeah.

214
00:10:17,960 --> 00:10:20,200
So if it ever showed like inaccurate data,

215
00:10:20,200 --> 00:10:23,240
then they're just going to look at that and they're not going to wait for the refresh.

216
00:10:23,240 --> 00:10:25,400
Yeah, and it's going to be super expensive for the company

217
00:10:25,400 --> 00:10:28,520
because they're going to make a mistake because they work on outdated data, right?

218
00:10:28,520 --> 00:10:29,000
Yeah.

219
00:10:29,000 --> 00:10:30,760
Like it's going to be very expensive down the line.

220
00:10:30,760 --> 00:10:33,160
So you never want to be in this situation, basically.

221
00:10:33,160 --> 00:10:33,240
Yeah.

222
00:10:33,240 --> 00:10:34,840
It's interesting on the narcissist side as well,

223
00:10:34,840 --> 00:10:38,920
because you know, we have big clients to do caching, prime.

224
00:10:40,760 --> 00:10:44,840
Which I, again, like I'm a prime fan, like don't get me wrong.

225
00:10:44,840 --> 00:10:47,080
I think they're doing a lot of things right.

226
00:10:47,080 --> 00:10:52,040
But you land in these weird situations where just you move away from the

227
00:10:52,120 --> 00:10:56,600
noster native way of doing things and you just land in weird states.

228
00:10:56,600 --> 00:10:59,400
And like sometimes posting takes forever for some reason,

229
00:10:59,400 --> 00:11:02,040
because it goes through their servers and so on.

230
00:11:02,040 --> 00:11:08,040
And what I like about noster proper is like it's so insanely snappy.

231
00:11:08,040 --> 00:11:10,600
Like you are never in an outdated state in some sense.

232
00:11:10,600 --> 00:11:12,920
You know, like you don't, you never have to wait in some sense.

233
00:11:13,480 --> 00:11:13,960
That's it.

234
00:11:13,960 --> 00:11:14,920
It's very immediate.

235
00:11:14,920 --> 00:11:15,240
Yeah.

236
00:11:15,240 --> 00:11:15,720
That's what I mean.

237
00:11:15,720 --> 00:11:16,840
It's very raw and immediate.

238
00:11:16,840 --> 00:11:18,280
Well, that's the nice thing about it is like,

239
00:11:19,240 --> 00:11:21,400
because you can progressively load data.

240
00:11:21,400 --> 00:11:22,920
The events come in like a stream.

241
00:11:22,920 --> 00:11:27,400
So you can start showing the user things working immediately

242
00:11:27,400 --> 00:11:29,400
instead of like in the tradition.

243
00:11:29,400 --> 00:11:31,240
It's so awesome when you're developing it as well.

244
00:11:31,240 --> 00:11:31,720
It's right.

245
00:11:31,720 --> 00:11:36,120
It's like you, it goes back to like just showing the user something's happening.

246
00:11:36,120 --> 00:11:38,600
So the user's like, like the connection's working.

247
00:11:38,600 --> 00:11:39,400
There's data coming in.

248
00:11:39,400 --> 00:11:40,840
It's not complete yet, but it's coming in.

249
00:11:40,840 --> 00:11:41,320
Yeah.

250
00:11:41,320 --> 00:11:44,040
And that just builds more trust with users.

251
00:11:44,040 --> 00:11:47,560
So that you don't have, because if it just sits there and loads forever,

252
00:11:47,560 --> 00:11:52,680
and then eventually shows everything, the user for like the five seconds,

253
00:11:52,680 --> 00:11:53,800
wherever long it's taking to load.

254
00:11:53,800 --> 00:11:55,080
The user doesn't know if they're offline.

255
00:11:55,080 --> 00:11:56,680
Is this thing actually working or not?

256
00:11:56,680 --> 00:11:57,400
Yeah, exactly.

257
00:11:59,000 --> 00:12:03,160
It's actually the thing I do, whether I, if I want to check whether I'm online or not,

258
00:12:03,160 --> 00:12:07,480
like if my Wi-Fi is working and this stuff, I open my nostril client and refresh it.

259
00:12:07,480 --> 00:12:11,800
I'll know immediately, because it just, it just, you know, it's so immediate.

260
00:12:11,800 --> 00:12:14,360
It's the quickest way I know of, basically.

261
00:12:14,360 --> 00:12:17,160
Like everything else I'll have to, I would have to, you know,

262
00:12:17,160 --> 00:12:20,040
like do a bogus search or something or whatever.

263
00:12:20,040 --> 00:12:21,240
Or, yeah, yeah.

264
00:12:21,240 --> 00:12:26,200
The funny thing too is like what we were talking about earlier is like getting used to it,

265
00:12:26,200 --> 00:12:27,400
how quickly you get used to it.

266
00:12:27,400 --> 00:12:32,280
Because I'd been building and using it for a year, as most of us have.

267
00:12:32,280 --> 00:12:35,720
And it's like, then you go back to some of the traditional social media,

268
00:12:36,360 --> 00:12:39,000
and it actually takes a long time for that stuff to load.

269
00:12:39,560 --> 00:12:40,920
Even on like my fast connection.

270
00:12:41,560 --> 00:12:45,960
Like, I mean, whatever you think traditional, like Facebook, Reddit, Twitter,

271
00:12:45,960 --> 00:12:49,000
stuff like that, like it gets stuck loading.

272
00:12:49,000 --> 00:12:50,760
And you're just sitting there and no data showing.

273
00:12:50,760 --> 00:12:52,520
And you don't know if it's online or not.

274
00:12:52,520 --> 00:12:52,920
Yeah.

275
00:12:52,920 --> 00:12:53,880
It's like the nostrils.

276
00:12:55,960 --> 00:13:02,760
I sometimes, you know, like venture out into the dark corners of the legacy internet as well.

277
00:13:02,760 --> 00:13:07,800
And what's so weird to me is that you actually, you never know what it's really doing.

278
00:13:07,800 --> 00:13:12,440
Because it's like every time I refresh something, it's a new feed.

279
00:13:12,440 --> 00:13:18,200
And the algorithm tries to, you know, shove something else down my throat.

280
00:13:18,200 --> 00:13:19,560
I'm really allergic to that now.

281
00:13:19,560 --> 00:13:22,360
Where it's just like, on the nostrils, I know where I am.

282
00:13:22,760 --> 00:13:24,040
It's like, I know how it works.

283
00:13:24,040 --> 00:13:30,280
I, again, like the, in some sense, how should I put it?

284
00:13:30,280 --> 00:13:32,280
It's, I think it relates to what you said before.

285
00:13:32,280 --> 00:13:36,680
Like the system should be transparent and not lying to you.

286
00:13:37,320 --> 00:13:40,440
And if someone tries to shove something down your throat,

287
00:13:40,440 --> 00:13:43,560
because they want to sell you something, both of these are not given.

288
00:13:43,560 --> 00:13:45,880
Like they're not going to be very transparent about it.

289
00:13:45,880 --> 00:13:50,440
Like with the algorithm changing the feed and a lot of all the social clients is like,

290
00:13:50,440 --> 00:13:54,360
another lie is like, you remember a video you saw five minutes ago.

291
00:13:54,360 --> 00:13:55,320
Oh, you can't find it again.

292
00:13:55,320 --> 00:13:56,200
You can't find it.

293
00:13:56,200 --> 00:13:56,840
It's gone.

294
00:13:56,840 --> 00:13:57,160
Yeah.

295
00:13:57,160 --> 00:13:57,560
And it's like...

296
00:13:57,560 --> 00:13:59,720
Because you were refreshed or the thing refreshed automatically.

297
00:13:59,720 --> 00:13:59,960
Yeah.

298
00:13:59,960 --> 00:14:01,640
It's like, did they actually post that?

299
00:14:01,640 --> 00:14:03,960
And you just imagined it or was it actually there?

300
00:14:03,960 --> 00:14:04,760
You can't find it.

301
00:14:04,760 --> 00:14:05,240
Yeah.

302
00:14:05,240 --> 00:14:07,480
And so it's those little things that build up where it's just...

303
00:14:08,280 --> 00:14:08,760
Yeah, it's fresh.

304
00:14:08,760 --> 00:14:09,640
The thing didn't work.

305
00:14:12,040 --> 00:14:13,720
The step before frustration.

306
00:14:15,000 --> 00:14:16,200
The step people don't notice.

307
00:14:16,200 --> 00:14:19,800
Like a good example of the step people don't notice is like,

308
00:14:19,800 --> 00:14:25,160
you click on a button and it just takes two seconds to do before it changes state.

309
00:14:25,960 --> 00:14:27,240
Like it's not frustrating.

310
00:14:27,240 --> 00:14:29,000
So you're not sure if you actually clicked it on?

311
00:14:29,000 --> 00:14:29,560
Yeah, you're not sure.

312
00:14:29,560 --> 00:14:31,880
For those two seconds, you're guessing yourself.

313
00:14:31,880 --> 00:14:32,840
And it's not frustration.

314
00:14:32,840 --> 00:14:34,520
It's just something else.

315
00:14:34,520 --> 00:14:35,000
Yeah.

316
00:14:35,000 --> 00:14:37,400
It's like lag in video games.

317
00:14:37,400 --> 00:14:38,200
Yeah, exactly.

318
00:14:38,200 --> 00:14:40,040
It builds the frustration.

319
00:14:40,040 --> 00:14:41,000
You have it a few times.

320
00:14:41,000 --> 00:14:42,440
It's like, you can deal with it.

321
00:14:42,440 --> 00:14:45,560
But if you keep lagging, it's like, that's really just rage going.

322
00:14:45,560 --> 00:14:46,040
Yeah.

323
00:14:46,040 --> 00:14:46,600
Yeah.

324
00:14:46,600 --> 00:14:47,240
Yeah.

325
00:14:47,240 --> 00:14:47,560
Yeah.

326
00:14:47,560 --> 00:14:49,720
No, I think all these are the issues are super interesting.

327
00:14:49,720 --> 00:14:51,800
I use no struggle a lot for that reason.

328
00:14:51,800 --> 00:14:54,760
Like when I want to know kind of...

329
00:14:55,800 --> 00:15:00,440
I mean, to me, it's part client, part developer tool in some sense.

330
00:15:00,440 --> 00:15:00,920
Yeah.

331
00:15:00,920 --> 00:15:04,440
Like if I need to sign a ROI event, it's very useful.

332
00:15:04,440 --> 00:15:06,760
If I really want to see what's happening under the hood,

333
00:15:06,760 --> 00:15:07,880
it's very useful.

334
00:15:07,880 --> 00:15:10,440
If I want to clean up some of my things, it's very useful.

335
00:15:11,480 --> 00:15:13,480
It's a very useful client in general.

336
00:15:13,480 --> 00:15:15,800
It has a very opinionated UI.

337
00:15:17,160 --> 00:15:17,800
I did.

338
00:15:17,800 --> 00:15:19,720
I put like hardly any work into that.

339
00:15:21,080 --> 00:15:23,080
It's funny because I saw it grow all the time,

340
00:15:23,080 --> 00:15:25,400
you know, like you add things and other things and other things.

341
00:15:25,400 --> 00:15:25,720
Oh, yeah.

342
00:15:25,720 --> 00:15:26,200
Yeah.

343
00:15:26,200 --> 00:15:28,280
And now it's like...

344
00:15:29,240 --> 00:15:31,320
It's the amethyst of web clients in some sense.

345
00:15:31,320 --> 00:15:32,600
No offense, you know, but it's like...

346
00:15:32,600 --> 00:15:33,320
It's a little take.

347
00:15:33,320 --> 00:15:34,680
It's really, it's really little take.

348
00:15:34,680 --> 00:15:36,040
It can do anything now, basically.

349
00:15:36,040 --> 00:15:37,080
Yeah, like that's...

350
00:15:37,080 --> 00:15:39,720
I keep taking forever and make updates to it.

351
00:15:39,720 --> 00:15:42,600
But that's the next thing I'm trying to work on and...

352
00:15:42,600 --> 00:15:45,240
What do you have in the pipeline on the nostril side?

353
00:15:45,240 --> 00:15:46,440
What do you want to build next?

354
00:15:48,280 --> 00:15:50,200
Half the views, these don't need to be removed,

355
00:15:50,200 --> 00:15:52,600
or they need to be moved to a place that's easier to find.

356
00:15:52,600 --> 00:15:54,600
So a lot of, like, there's a lot of experimental stuff in there

357
00:15:54,600 --> 00:15:56,680
that doesn't work anymore and isn't useful.

358
00:15:56,680 --> 00:15:58,040
How do you use it most of?

359
00:15:58,600 --> 00:16:02,040
Like what's your main use case of the client that you build yourself?

360
00:16:02,040 --> 00:16:05,400
The social sides, like the main timeline is useful.

361
00:16:06,120 --> 00:16:09,080
Notifications, but then that view still sucks.

362
00:16:09,800 --> 00:16:10,440
It needs to be rebuilt.

363
00:16:10,440 --> 00:16:11,320
You think so, why?

364
00:16:11,320 --> 00:16:13,480
I think it's actually not terrible at all.

365
00:16:13,480 --> 00:16:16,360
I don't think notifications work as a timeline in nostril.

366
00:16:17,160 --> 00:16:22,760
Because when I go to what I want to know what I should care about,

367
00:16:22,760 --> 00:16:25,800
what people are tagging me on, what people are mentioning on,

368
00:16:25,800 --> 00:16:27,240
I'm not looking for a timeline.

369
00:16:27,240 --> 00:16:29,320
I'm not looking for the order of things that happened.

370
00:16:29,320 --> 00:16:32,040
I'm looking for the people that tagged me, the context they tagged me in.

371
00:16:32,040 --> 00:16:34,360
You kind of look for the most important stuff, in some sense.

372
00:16:34,360 --> 00:16:36,760
And I have to filter, I have to dig through a timeline.

373
00:16:36,760 --> 00:16:39,480
Like the order of the things came in doesn't really matter to me.

374
00:16:40,200 --> 00:16:42,200
Like, and also like the zaps and the reactions.

375
00:16:42,200 --> 00:16:44,600
Like, I don't care what time those came in.

376
00:16:44,600 --> 00:16:45,560
I just kid it came in.

377
00:16:45,560 --> 00:16:47,560
Yeah, I like that it can be filtered.

378
00:16:47,560 --> 00:16:48,440
That's how I use it.

379
00:16:48,440 --> 00:16:53,320
So I filter it only to revise and quote mentions and those kind of things.

380
00:16:53,320 --> 00:16:56,040
Maybe like, now I've gotten the vibe-coding stuff.

381
00:16:56,040 --> 00:16:57,720
Like, I'm trying to rebuild that cleaner.

382
00:16:57,720 --> 00:17:02,360
Next, get the AI to suggest some of the layouts.

383
00:17:02,360 --> 00:17:04,920
But my idea is to turn it into kind of a dashboard, where it's like,

384
00:17:05,560 --> 00:17:09,800
you have the in-box style still for like comments, replies,

385
00:17:09,800 --> 00:17:12,440
for people like replying to you or getting your attention.

386
00:17:13,000 --> 00:17:16,040
And then you have other dashboards for things like zaps and reactions.

387
00:17:16,040 --> 00:17:19,720
And where the data is more summarized, because that's what you want to see.

388
00:17:19,720 --> 00:17:19,960
Yeah.

389
00:17:20,760 --> 00:17:22,440
Yeah, no, I think that's right.

390
00:17:22,440 --> 00:17:23,000
I think that's right.

391
00:17:23,000 --> 00:17:25,480
I mean, I use the combination of all kinds of things.

392
00:17:25,480 --> 00:17:29,800
In Amethyst, they have a, I'm not even sure who built it.

393
00:17:29,800 --> 00:17:31,400
Maybe it was Vitor himself.

394
00:17:31,400 --> 00:17:34,680
But Amethyst has a bunch of contributors that just add stuff in Vitor.

395
00:17:34,680 --> 00:17:35,960
It's just like, yeah, yeah, let's add it.

396
00:17:35,960 --> 00:17:36,360
Approved.

397
00:17:36,360 --> 00:17:37,640
Approved, yeah, yeah.

398
00:17:37,640 --> 00:17:38,600
Auto-Eck.

399
00:17:40,280 --> 00:17:49,160
But they have a nice visualization of kind of, they draw a graph for the last week of,

400
00:17:49,160 --> 00:17:53,000
you know, the zaps and other reactions and that's really neat.

401
00:17:53,000 --> 00:17:54,760
I glance at it from time to time.

402
00:17:54,760 --> 00:17:55,720
It's quite useful.

403
00:17:55,720 --> 00:17:59,480
It's like, it's more useful than I thought it would be in some sense.

404
00:17:59,480 --> 00:18:00,600
Well, that's what I want to get to.

405
00:18:00,600 --> 00:18:03,400
That's what I want to get to, because like, I found that like,

406
00:18:03,400 --> 00:18:07,160
if you're building dashboards, you can't put the graphs in first.

407
00:18:07,160 --> 00:18:08,600
Because then you don't know what you're visualizing.

408
00:18:08,600 --> 00:18:11,640
You got to put the data there and I got to figure out how to massage the data into like,

409
00:18:12,440 --> 00:18:13,480
digestible format.

410
00:18:13,480 --> 00:18:14,680
And then you put the pretty graphs.

411
00:18:14,680 --> 00:18:14,920
Yeah.

412
00:18:14,920 --> 00:18:16,120
Because I want the, I want the pretty graphs.

413
00:18:16,120 --> 00:18:20,040
I want, I want something that like, shows me, even when like I click on a user,

414
00:18:20,040 --> 00:18:23,560
I want to see like a simple graph like activity over time in the last week.

415
00:18:23,560 --> 00:18:24,600
So like, have they been active?

416
00:18:24,600 --> 00:18:25,400
When are they active?

417
00:18:26,120 --> 00:18:27,080
Stuff like that.

418
00:18:27,080 --> 00:18:27,400
Yeah.

419
00:18:27,480 --> 00:18:32,040
I think you, but again, I think it's, it's, I think you're on to something there,

420
00:18:32,040 --> 00:18:36,920
because the reason, the reason why I mentioned the notification

421
00:18:37,720 --> 00:18:42,840
type of no struggle, I like it is because I have it set to, I think what you want to

422
00:18:42,840 --> 00:18:44,120
build as the next iteration.

423
00:18:44,120 --> 00:18:47,240
I only have it set like, if someone mentions me, if someone replies to me,

424
00:18:47,240 --> 00:18:51,960
if someone quotes something that I put out in the past and so on,

425
00:18:51,960 --> 00:18:53,640
then that's all I filter for.

426
00:18:53,640 --> 00:18:54,440
So that's all I see.

427
00:18:54,520 --> 00:18:55,960
So I hide all reactions.

428
00:18:55,960 --> 00:18:57,560
I hide all steps and so on and so forth.

429
00:18:57,560 --> 00:19:01,640
That's what, that's the only, that's kind of what time I saw before.

430
00:19:01,640 --> 00:19:05,640
Kind of almost like a, like a DM, but for public dimensions, like a DM,

431
00:19:06,440 --> 00:19:07,640
because I ignore all the amps.

432
00:19:07,640 --> 00:19:10,600
It's just like, it's, it's, it's, it's beyond repair.

433
00:19:10,600 --> 00:19:12,280
My DMs are beyond repair.

434
00:19:12,280 --> 00:19:18,040
So I stopped looking a long time ago, but, but, uh, you know, I, I liked, uh,

435
00:19:18,840 --> 00:19:25,800
I'm also, I'm a way bigger fan of public conversations than DMs in some sense.

436
00:19:25,800 --> 00:19:28,040
Because most of the time when people reach out to me,

437
00:19:28,040 --> 00:19:29,640
I'm not even the best person to answer it.

438
00:19:29,640 --> 00:19:31,640
Yeah, you just want to tag someone else into the same thing.

439
00:19:31,640 --> 00:19:33,400
Exactly, just to ask the question kind of publicly,

440
00:19:33,400 --> 00:19:35,640
because most of the time it's questions and stuff and so on.

441
00:19:35,640 --> 00:19:36,840
And it's not even private stuff.

442
00:19:36,840 --> 00:19:39,880
I mean, you know, like I buy, I, I do care about

443
00:19:39,880 --> 00:19:43,720
encrypted communication and private communication and so on for private stuff.

444
00:19:43,720 --> 00:19:47,160
But most of the things we are dealing with, like we are building in public anyway.

445
00:19:47,240 --> 00:19:48,920
We're also discussing in public.

446
00:19:48,920 --> 00:19:52,200
We're also like, you know, we're having the conversation where things should go

447
00:19:52,200 --> 00:19:52,760
with public.

448
00:19:52,760 --> 00:19:55,880
We're working, we're, we're using and living on a protocol.

449
00:19:55,880 --> 00:19:56,360
Exactly.

450
00:19:56,360 --> 00:19:57,560
No, no, has always no.

451
00:19:57,560 --> 00:20:00,760
So if you have a question for another person, just ask it and tag the person.

452
00:20:00,760 --> 00:20:05,640
And chances are either they will reply or someone else who is more qualified will reply.

453
00:20:05,640 --> 00:20:07,320
And both of those are great outcomes.

454
00:20:07,320 --> 00:20:07,640
Yeah.

455
00:20:07,640 --> 00:20:12,280
And if you just use DMs, then you, you block yourself off from the, off from the great outcomes.

456
00:20:12,280 --> 00:20:14,120
What do you think about just removing DMs?

457
00:20:14,120 --> 00:20:18,040
Because I've been thinking about this for the, I guess, the last, last few weeks.

458
00:20:18,040 --> 00:20:18,520
Yeah.

459
00:20:18,520 --> 00:20:20,760
Has been looking into like NIP 17 more finally.

460
00:20:21,400 --> 00:20:25,720
And I mean, it, it, it relates to what it's not about the draft.

461
00:20:25,720 --> 00:20:28,200
It's like, it's, it's very expensive operations.

462
00:20:28,200 --> 00:20:30,840
I'm not sure if the M's truly should live on Noster in some sense.

463
00:20:30,840 --> 00:20:32,280
I mean, I guess they have their use.

464
00:20:32,280 --> 00:20:36,200
I was thinking is like, what if we just remove them and you go, kind of go with,

465
00:20:37,160 --> 00:20:41,720
I think what like Skypad at one point or other platforms have where they had

466
00:20:41,800 --> 00:20:45,400
support for just like, here's my link to talk to me on this platform.

467
00:20:45,400 --> 00:20:45,880
Yeah.

468
00:20:45,880 --> 00:20:47,000
Like here's my simplex link.

469
00:20:47,000 --> 00:20:47,960
Here's my telegram link.

470
00:20:47,960 --> 00:20:48,920
Here's my signal link.

471
00:20:48,920 --> 00:20:49,400
Yeah.

472
00:20:49,400 --> 00:20:51,720
I'm just kind of promoting that more on the user profile.

473
00:20:51,720 --> 00:20:53,080
No, no one stops.

474
00:20:53,080 --> 00:20:57,640
No one stops simplex and those kinds of solutions to use Noster as like a,

475
00:20:58,200 --> 00:21:00,680
the substrate for lookups, the 10 shades, basically.

476
00:21:00,680 --> 00:21:01,320
That's true too.

477
00:21:01,320 --> 00:21:01,640
Yeah.

478
00:21:01,640 --> 00:21:08,200
And that's, that's what, where I think that might just happen all the time,

479
00:21:08,280 --> 00:21:13,480
where it's, I want to initiate a conversation with someone who is on Noster.

480
00:21:13,480 --> 00:21:14,760
And I want to use this channel.

481
00:21:14,760 --> 00:21:17,640
And like the, you know, you can't do the handshake this way.

482
00:21:17,640 --> 00:21:23,400
It's very much like the Bitcoin wall that's integrating Noster lookups to send money to.

483
00:21:23,400 --> 00:21:25,800
It's like, I just need your lighting address, basically.

484
00:21:25,800 --> 00:21:26,120
Yeah.

485
00:21:26,120 --> 00:21:28,200
And it's like, and it's very easy to look it up over Noster.

486
00:21:28,200 --> 00:21:31,240
So you have like a sent to address book and so on.

487
00:21:31,240 --> 00:21:35,160
That would, that would might be the best solution for, because a lot of people use

488
00:21:35,160 --> 00:21:39,480
simplex and then the signal and, I mean, it always changes every single time one of them gets

489
00:21:39,480 --> 00:21:43,080
from one thing, but, you know, there's always one that a lot of people are using.

490
00:21:43,080 --> 00:21:43,480
Yeah.

491
00:21:43,480 --> 00:21:45,800
It wouldn't be very difficult to add like a simple link.

492
00:21:45,800 --> 00:21:46,360
Yeah.

493
00:21:46,360 --> 00:21:48,120
Contact information basically.

494
00:21:48,120 --> 00:21:53,720
Getting the M's right is really hard because, because people, like one on one,

495
00:21:53,720 --> 00:21:54,760
the M's are one thing.

496
00:21:54,760 --> 00:21:59,960
That's already hard, but people really love group chats and getting group chats right

497
00:21:59,960 --> 00:22:00,920
is really, really hard.

498
00:22:00,920 --> 00:22:04,440
I mean, there's like five different people working on it on the Noster side right now.

499
00:22:04,520 --> 00:22:08,120
And we'll see what kind of protocol, what kind of approach wins out.

500
00:22:08,600 --> 00:22:12,920
And the thing about the group chat part is that how private is a group chat, really?

501
00:22:12,920 --> 00:22:15,880
Like, you have to differentiate between a group chat between three people,

502
00:22:15,880 --> 00:22:21,880
which is reasonably private, probably, or like 500 people, where there's the likelihood of being

503
00:22:21,880 --> 00:22:26,520
a mole in there or someone who just streams the whole chat to Twitch or takes screenshots every

504
00:22:26,520 --> 00:22:27,240
five minutes.

505
00:22:27,240 --> 00:22:28,920
It's like basically a hundred percent.

506
00:22:28,920 --> 00:22:31,000
It's like it works too in the future.

507
00:22:31,000 --> 00:22:32,280
So why encrypt all the things?

508
00:22:32,280 --> 00:22:32,920
You know what I mean?

509
00:22:32,920 --> 00:22:39,400
It's like, why share secrets between 500 people and do all the fancy cryptography stuff that's

510
00:22:39,400 --> 00:22:44,760
like super, super, super expensive and hard to do if the group chat size is like 500 people.

511
00:22:44,760 --> 00:22:50,120
Like, it's don't assume privacy if you're talking in a room of 500 people.

512
00:22:50,760 --> 00:22:56,920
Yeah, it's a weird, and the funny part is you wouldn't assume that if you were standing

513
00:22:56,920 --> 00:23:00,600
next within a group of 500 people in the real world, you wouldn't assume that at all.

514
00:23:01,080 --> 00:23:03,880
But something about the technology hides that.

515
00:23:03,880 --> 00:23:10,280
Yeah, no, especially I think the engineering mindset wants to get it perfect, you know?

516
00:23:10,280 --> 00:23:10,680
Yeah, yeah.

517
00:23:10,680 --> 00:23:13,240
And it's like, it's a complete waste of time.

518
00:23:13,240 --> 00:23:17,160
It's like, you might as well switch off encryption if the group chat is above like,

519
00:23:17,160 --> 00:23:18,440
you know, 20 people or whatever.

520
00:23:18,440 --> 00:23:21,480
It's like, at a certain point, it's useless.

521
00:23:21,480 --> 00:23:25,240
You'll either have a mole or an asshole in there.

522
00:23:25,240 --> 00:23:26,360
It's like, and it doesn't matter.

523
00:23:26,360 --> 00:23:27,800
Like, someone's going to leak all the things.

524
00:23:27,800 --> 00:23:29,960
Someone's just going to back it up, archive it.

525
00:23:30,040 --> 00:23:31,080
Yeah, exactly.

526
00:23:31,080 --> 00:23:35,400
I mean, it's like, I mean, me and my friends, like, there's like three of us.

527
00:23:35,400 --> 00:23:38,520
We have a small Discord server from way back in the day.

528
00:23:38,520 --> 00:23:43,320
And I have all of the, all of the chats downloaded this text file stored somewhere.

529
00:23:43,320 --> 00:23:43,560
Yeah.

530
00:23:43,560 --> 00:23:45,880
Because I was planning on dumping them into some AI program.

531
00:23:45,880 --> 00:23:48,040
With sync thing across like 15 different devices.

532
00:23:48,040 --> 00:23:49,480
Yeah, so like, that's just three of us.

533
00:23:49,480 --> 00:23:52,120
And that's, that data is already like stored for everywhere.

534
00:23:52,120 --> 00:23:53,640
Going to be leaked at some point.

535
00:23:53,640 --> 00:23:54,360
Probably.

536
00:23:54,360 --> 00:23:57,960
So, yeah, it's kind of that point to exactly, but.

537
00:23:58,760 --> 00:24:00,200
Yeah, I don't know.

538
00:24:00,200 --> 00:24:01,720
It's a really hard question.

539
00:24:01,720 --> 00:24:07,960
It's like, I, because, I mean, what attracted me to, to Nostra and still does is, is,

540
00:24:07,960 --> 00:24:11,480
because it's raw, it's immediate, and it's very, very simple.

541
00:24:11,480 --> 00:24:15,720
And, and to your point, you know, like being cryptographic operations,

542
00:24:15,720 --> 00:24:17,720
being expensive and all those kinds of things.

543
00:24:17,720 --> 00:24:21,240
Having encrypted group chats on Nostra directly is not very simple.

544
00:24:21,240 --> 00:24:23,160
Like there's some, the simplicity stops.

545
00:24:23,160 --> 00:24:25,480
Because you lose, you don't have consistent state,

546
00:24:25,480 --> 00:24:26,920
which is the first problem you get over.

547
00:24:27,640 --> 00:24:29,880
And like public groups, the NIP 29 kind of.

548
00:24:29,880 --> 00:24:33,000
And with Nostra proper, you never will have consistent state.

549
00:24:33,000 --> 00:24:36,040
That's kind of the, the thing, the thing to understand.

550
00:24:36,040 --> 00:24:38,600
Like if you actually use Nostra properly, how it was designed.

551
00:24:38,600 --> 00:24:39,160
Well, if you agree.

552
00:24:39,160 --> 00:24:40,680
You cannot guarantee consistent state.

553
00:24:40,680 --> 00:24:43,400
You don't have a server that can guarantee consistent state.

554
00:24:43,400 --> 00:24:46,120
If you agree on a server, you can negotiate like consistent state.

555
00:24:46,120 --> 00:24:49,000
But yeah, again, if you're not, if you're trying to.

556
00:24:49,000 --> 00:24:52,200
There's very few situations where it makes sense to agree on a server.

557
00:24:52,200 --> 00:24:53,000
Like on a relay.

558
00:24:53,880 --> 00:24:57,560
Yeah, but then again, it's like an anti-pattern, right?

559
00:24:57,560 --> 00:25:00,520
Like the beauty of Nostra and the simplicity comes from it.

560
00:25:00,520 --> 00:25:03,560
That there is no global and you don't have data consistency guarantees

561
00:25:03,560 --> 00:25:04,440
and all those kinds of things.

562
00:25:04,440 --> 00:25:07,320
Like it strikes a different tradeoff balance.

563
00:25:07,320 --> 00:25:10,280
And trying to shoehorn it into a use case

564
00:25:10,280 --> 00:25:12,120
where you need perfect data consistency.

565
00:25:12,120 --> 00:25:14,120
It's kind of an anti-pattern in a Nostra world.

566
00:25:14,120 --> 00:25:17,560
There's one, there's one use case where I think it does make sense to like

567
00:25:17,560 --> 00:25:20,600
agree on a relay or a server or whatnot.

568
00:25:20,600 --> 00:25:24,920
And that is like community infrastructure.

569
00:25:24,920 --> 00:25:26,200
Oh, yeah, okay, fair enough.

570
00:25:26,200 --> 00:25:27,960
Like a lot of what we discord in.

571
00:25:27,960 --> 00:25:29,640
It's like the self-hosted idea basically.

572
00:25:29,640 --> 00:25:30,440
Yes, yeah, yeah.

573
00:25:30,440 --> 00:25:32,200
Like what people use matrix servers for now.

574
00:25:32,200 --> 00:25:32,600
Yeah.

575
00:25:32,600 --> 00:25:33,880
They all agree on the same.

576
00:25:33,880 --> 00:25:35,080
And someone runs the server.

577
00:25:35,080 --> 00:25:35,960
The server's maintained.

578
00:25:35,960 --> 00:25:36,680
Server has features.

579
00:25:36,680 --> 00:25:39,400
Yeah, and you basically invite only whitelist the people.

580
00:25:39,400 --> 00:25:40,440
Yeah, stuff like that.

581
00:25:40,440 --> 00:25:41,320
That makes a lot of sense.

582
00:25:41,320 --> 00:25:44,200
But in other cases, if I'm just talking to strangers on the internet,

583
00:25:44,200 --> 00:25:46,760
like it's impossible to think.

584
00:25:46,760 --> 00:25:47,560
Yeah.

585
00:25:47,560 --> 00:25:47,960
Yeah.

586
00:25:47,960 --> 00:25:51,560
I mean, we're in the process of figuring all that out.

587
00:25:51,560 --> 00:25:53,000
You just mentioned before that, you know,

588
00:25:53,000 --> 00:25:55,000
byte coding is going to make the notification

589
00:25:55,000 --> 00:25:56,680
if you have no struggle great again.

590
00:25:58,360 --> 00:26:01,400
So what's your experience being the last couple of weeks

591
00:26:01,400 --> 00:26:03,800
getting into the byte coding world?

592
00:26:03,800 --> 00:26:05,000
Do you have high hopes for it?

593
00:26:06,120 --> 00:26:08,600
Like the opinions are very split online,

594
00:26:08,600 --> 00:26:10,520
whether it's useful or not.

595
00:26:10,520 --> 00:26:14,120
Or like, you know, will we produce AI slot forever?

596
00:26:14,760 --> 00:26:16,200
People have to argue online.

597
00:26:18,920 --> 00:26:20,280
Ah, yeah.

598
00:26:20,280 --> 00:26:22,280
And it's going to work.

599
00:26:22,280 --> 00:26:23,000
It's going to work.

600
00:26:23,000 --> 00:26:24,840
I mean, we suspected that from the beginning.

601
00:26:24,840 --> 00:26:27,880
And like I said, the last few weeks is a lot of good evidence

602
00:26:27,880 --> 00:26:29,240
that it definitely does work.

603
00:26:30,840 --> 00:26:32,040
Whether it's maintainable though,

604
00:26:32,040 --> 00:26:35,640
like that's the interesting, the last interesting question.

605
00:26:35,640 --> 00:26:39,000
Because, I mean, there's a reason why Pablo loves it so much, right?

606
00:26:39,000 --> 00:26:40,600
He can just shit out in that game.

607
00:26:40,600 --> 00:26:42,760
But Amanda did not maintain it anymore.

608
00:26:42,760 --> 00:26:44,280
Now he can do it on steroids.

609
00:26:44,280 --> 00:26:46,280
Well, that's the interesting part is like,

610
00:26:47,240 --> 00:26:51,160
what does it, because the code's definitely not maintainable.

611
00:26:51,160 --> 00:26:51,640
It produces.

612
00:26:52,600 --> 00:26:55,960
And if it keeps producing, it produces really,

613
00:26:56,680 --> 00:27:00,280
in the best of worlds, it produces code that is somewhat maintainable.

614
00:27:00,840 --> 00:27:01,160
Yeah.

615
00:27:01,720 --> 00:27:06,520
But does it really matter anymore if you can just regenerate some of the code?

616
00:27:07,800 --> 00:27:11,720
And basically, if you go to more of a modular approach,

617
00:27:12,600 --> 00:27:13,240
and you can just...

618
00:27:13,240 --> 00:27:13,960
Yeah, fair enough.

619
00:27:13,960 --> 00:27:14,600
Okay.

620
00:27:14,680 --> 00:27:16,600
Generate, you can just, like something doesn't work.

621
00:27:16,600 --> 00:27:17,640
Trusting talks about this as well.

622
00:27:17,640 --> 00:27:19,640
Like small, specialized things, basically.

623
00:27:20,440 --> 00:27:21,240
It pushes things.

624
00:27:21,240 --> 00:27:23,000
It either pushes things in two directions I see.

625
00:27:23,000 --> 00:27:25,160
It's like, it's small, specialized things.

626
00:27:26,200 --> 00:27:29,960
Where like, your UI components are built independently

627
00:27:29,960 --> 00:27:34,040
and plug in in easy ways to your, primarily,

628
00:27:34,760 --> 00:27:38,120
your application logic and stuff works in a very modular way.

629
00:27:38,840 --> 00:27:41,320
Or it kind of goes in the other way where,

630
00:27:41,400 --> 00:27:44,600
or just both, where it just embraces wholeheartedly,

631
00:27:44,600 --> 00:27:46,920
like, a framework that does everything, like next to it.

632
00:27:46,920 --> 00:27:47,560
Yes.

633
00:27:47,560 --> 00:27:47,800
Yeah.

634
00:27:48,600 --> 00:27:48,920
Yeah.

635
00:27:50,200 --> 00:27:51,640
It seems to do both of those really well.

636
00:27:51,640 --> 00:27:52,760
But the interesting thing is,

637
00:27:53,560 --> 00:27:55,880
it doesn't produce maintainable code in either situation.

638
00:27:57,320 --> 00:27:58,040
Or what, you know,

639
00:27:59,720 --> 00:28:02,040
from the developer's opinion, perfect code.

640
00:28:03,240 --> 00:28:03,640
But then...

641
00:28:04,200 --> 00:28:05,960
Don't you think that's fixable with,

642
00:28:05,960 --> 00:28:08,440
basically, larger models and larger context windows?

643
00:28:09,400 --> 00:28:12,040
Yeah, and then we'll just try to do more complicated things.

644
00:28:12,040 --> 00:28:12,920
And it's not fixed a lot more.

645
00:28:12,920 --> 00:28:13,400
Fair enough.

646
00:28:13,400 --> 00:28:13,880
Yeah, yeah.

647
00:28:13,880 --> 00:28:14,120
Okay.

648
00:28:14,120 --> 00:28:14,680
That's the...

649
00:28:14,680 --> 00:28:17,960
So yeah, I mean, it's like building a broader highway.

650
00:28:17,960 --> 00:28:18,680
Yeah.

651
00:28:18,680 --> 00:28:19,560
People more people.

652
00:28:19,560 --> 00:28:20,120
More parts of it.

653
00:28:20,120 --> 00:28:20,600
Yeah.

654
00:28:20,600 --> 00:28:22,600
Like, it nails buttons now.

655
00:28:22,600 --> 00:28:23,720
You want UI components.

656
00:28:23,720 --> 00:28:25,240
It can build you, all the UI components.

657
00:28:25,240 --> 00:28:25,800
Yeah.

658
00:28:25,800 --> 00:28:27,320
But now we're trying to do more.

659
00:28:27,320 --> 00:28:28,280
And so, I, well,

660
00:28:29,480 --> 00:28:32,600
I get it's a lot like Bitcoin and all the other emerging technologies where

661
00:28:34,280 --> 00:28:36,280
it's never, it never works.

662
00:28:36,840 --> 00:28:40,280
We never think it works because we're always looking at like what we're working on right now.

663
00:28:40,280 --> 00:28:40,600
Yeah.

664
00:28:40,600 --> 00:28:43,800
And if you turn around and look back, it works amazing.

665
00:28:43,800 --> 00:28:44,440
Yeah.

666
00:28:44,440 --> 00:28:45,800
For everything it's already fixed.

667
00:28:45,800 --> 00:28:47,880
That's the funny thing about humanity.

668
00:28:47,880 --> 00:28:52,760
Like, we only call the things technology that just don't work right yet.

669
00:28:52,760 --> 00:28:55,960
You know, like chairs, we stopped calling chairs technology.

670
00:28:55,960 --> 00:28:56,920
Or doors.

671
00:28:56,920 --> 00:28:58,600
They're just chairs and doors.

672
00:28:58,600 --> 00:29:00,200
Yeah, all that stuff.

673
00:29:00,200 --> 00:29:01,160
That just works.

674
00:29:01,160 --> 00:29:01,800
It just works.

675
00:29:01,800 --> 00:29:03,160
It's not technology anymore.

676
00:29:03,160 --> 00:29:05,320
But everything that's kind of half broken.

677
00:29:05,320 --> 00:29:07,320
Or just, you know, fails sometimes.

678
00:29:07,320 --> 00:29:08,920
It's still technology.

679
00:29:10,200 --> 00:29:10,520
Yeah.

680
00:29:10,520 --> 00:29:11,640
That's an interesting point.

681
00:29:11,640 --> 00:29:15,480
That point like computers have been technology for.

682
00:29:15,480 --> 00:29:16,040
For a long time.

683
00:29:16,040 --> 00:29:16,680
For a long time.

684
00:29:16,680 --> 00:29:17,480
Yeah.

685
00:29:17,480 --> 00:29:19,960
I think it's because computers are not a static thing, right?

686
00:29:19,960 --> 00:29:22,760
It's like they're always evolving with us in some sense.

687
00:29:22,760 --> 00:29:24,760
Like our modern computers are very different.

688
00:29:24,760 --> 00:29:29,400
And like the, you know, big machines that fill the whole buildings in the 70s.

689
00:29:29,400 --> 00:29:32,200
You know, it's like, it's, it's very, very different.

690
00:29:33,000 --> 00:29:35,560
Yeah, they're not, they're not a machine anymore.

691
00:29:35,560 --> 00:29:36,280
It's something else.

692
00:29:36,280 --> 00:29:36,760
Yeah.

693
00:29:36,760 --> 00:29:39,560
Cars are kind of, you know, a car is a car.

694
00:29:39,560 --> 00:29:43,480
But it's still, they, they, they, they, modern cars still are.

695
00:29:43,480 --> 00:29:46,360
Yeah, like a slice of technology.

696
00:29:46,360 --> 00:29:46,680
Yeah.

697
00:29:46,680 --> 00:29:47,720
Because it doesn't work well.

698
00:29:47,720 --> 00:29:48,520
Yeah, exactly.

699
00:29:48,520 --> 00:29:50,120
It just breaks down sometimes.

700
00:29:50,120 --> 00:29:51,080
But yeah, a tractor.

701
00:29:51,080 --> 00:29:53,400
We don't think of a tractor as technology anymore.

702
00:29:53,400 --> 00:29:55,000
It's a, it's a piece of equipment now.

703
00:29:55,000 --> 00:29:55,320
Yeah.

704
00:29:55,320 --> 00:29:55,960
Or like books.

705
00:29:55,960 --> 00:29:59,400
I mentioned books a lot, you know, where it's like books are absolutely technology.

706
00:29:59,400 --> 00:30:03,320
Books are, are human technology and are a psycho technology and so on.

707
00:30:03,320 --> 00:30:06,040
And we don't view books as technology, like at all.

708
00:30:06,680 --> 00:30:07,080
Yeah.

709
00:30:07,080 --> 00:30:07,480
Yeah.

710
00:30:07,480 --> 00:30:10,840
But we, like there was a time where it didn't have books, you know, like it's just,

711
00:30:10,840 --> 00:30:11,800
it's just people forget.

712
00:30:12,680 --> 00:30:14,120
It's actually a human invention.

713
00:30:14,840 --> 00:30:15,000
Yeah.

714
00:30:17,000 --> 00:30:20,360
But yeah, no, I mean, for me, some of this not just blew my mind.

715
00:30:20,360 --> 00:30:23,320
And I, to me, it's out of the question that it will

716
00:30:24,680 --> 00:30:26,040
accelerate things massively.

717
00:30:26,040 --> 00:30:27,240
Good and bad in some sense.

718
00:30:27,320 --> 00:30:30,040
Like, like in the sense of, I think it will accelerate

719
00:30:30,760 --> 00:30:34,200
what we are building massively, but I think it will also accelerate what,

720
00:30:34,200 --> 00:30:38,360
I don't know, the world coin people and this kind of people are building, you know, like it's,

721
00:30:38,360 --> 00:30:41,240
it's just open AI and, and

722
00:30:41,240 --> 00:30:43,400
It'll take longer on their front, I think, because

723
00:30:44,200 --> 00:30:45,320
I mean, they have infinite money.

724
00:30:45,320 --> 00:30:46,120
That's the difference.

725
00:30:46,120 --> 00:30:46,920
We have limited money.

726
00:30:46,920 --> 00:30:50,600
Yeah, but they have infinite money, but they have the problem of being greedy,

727
00:30:50,600 --> 00:30:53,400
which is, which really inhibits their ability to work together.

728
00:30:54,520 --> 00:30:55,720
Like the cooperation.

729
00:30:55,720 --> 00:30:55,880
Yeah.

730
00:30:55,880 --> 00:30:56,680
Almost non-existent.

731
00:30:56,680 --> 00:30:57,400
Oh, that's fair.

732
00:30:57,400 --> 00:30:59,080
It's, it's, it's competitive silos.

733
00:30:59,080 --> 00:31:00,920
It's not synergistic.

734
00:31:00,920 --> 00:31:01,320
Yeah.

735
00:31:01,320 --> 00:31:05,560
Cooperation just because we want to do something working at some point,

736
00:31:06,040 --> 00:31:10,760
but it'll take them longer than we like, it's taking them longer than some of us have expected

737
00:31:10,760 --> 00:31:13,400
to get, uh, CBDCs.

738
00:31:13,400 --> 00:31:13,720
Yeah.

739
00:31:13,720 --> 00:31:14,440
Because we knew they're coming.

740
00:31:14,440 --> 00:31:15,160
They are coming.

741
00:31:15,160 --> 00:31:16,360
Oh, they are absolutely coming.

742
00:31:16,360 --> 00:31:16,520
Yeah.

743
00:31:16,520 --> 00:31:17,240
They can't work.

744
00:31:18,920 --> 00:31:23,080
Every government thing that's working on CBDCs is building it in the silo.

745
00:31:23,080 --> 00:31:24,200
They're not talking to each other.

746
00:31:24,200 --> 00:31:26,840
They're not even helping each other in some sense.

747
00:31:26,840 --> 00:31:27,080
Yeah.

748
00:31:27,080 --> 00:31:33,240
And so it would have come sooner if they'd actually like somehow all agreed on one CBDC to use.

749
00:31:33,880 --> 00:31:36,040
By the way, I was like, oh, it's such a beautiful day.

750
00:31:36,040 --> 00:31:37,560
No wind, no rain.

751
00:31:37,560 --> 00:31:38,680
It's going to be perfect.

752
00:31:38,680 --> 00:31:42,520
And now we're walking next to waterfalls and big waves.

753
00:31:42,520 --> 00:31:44,600
So noisy as fuck again.

754
00:31:46,120 --> 00:31:47,080
But that's fine.

755
00:31:47,080 --> 00:31:53,000
I just, you know, like, I only want the people to listen that are able to suffer

756
00:31:53,000 --> 00:31:57,080
through the bad quality with us.

757
00:31:57,880 --> 00:32:02,040
That's kind of an entry point already, you know, like, or a filter already.

758
00:32:03,240 --> 00:32:06,200
I only want the most interested people to listen.

759
00:32:06,200 --> 00:32:08,520
I don't know the same way I've taken with a lot of podcasts.

760
00:32:08,520 --> 00:32:12,200
Now I've gone through the usual cycle everyone does, which is, uh,

761
00:32:13,160 --> 00:32:15,400
it's probably smart if we turn around, I think.

762
00:32:15,400 --> 00:32:15,480
Yeah.

763
00:32:16,280 --> 00:32:19,400
The fence is falling down, but we can't do it.

764
00:32:19,400 --> 00:32:24,280
But just also the, it's kind of get, it's not necessarily going to get better with the noise.

765
00:32:28,040 --> 00:32:32,040
But yeah, it's a good, it's a good filter to make things hard in some sense.

766
00:32:32,040 --> 00:32:36,120
Like, again, for the, for the cohort, for example, it's like,

767
00:32:36,120 --> 00:32:37,720
you actually have to come to the island.

768
00:32:37,720 --> 00:32:38,200
Yeah.

769
00:32:38,200 --> 00:32:42,600
You have to, you know, like, make time for it.

770
00:32:42,600 --> 00:32:45,720
You have to come here on your own dime, all those kinds of things.

771
00:32:45,720 --> 00:32:49,160
And it filters out, it filters out the slop a bit.

772
00:32:49,960 --> 00:32:53,880
It's like, only people that really want to be here and want to work on this stuff are here.

773
00:32:53,880 --> 00:32:56,520
So I think it's similar on an ostracide, right?

774
00:32:56,520 --> 00:33:01,560
Like, it's like, you have to put in the work to actually use an ostrand, an ostrand, an ostrand, in some sense.

775
00:33:01,560 --> 00:33:04,920
So it's very, we have a, we have some barrier to entry.

776
00:33:04,920 --> 00:33:07,160
And that's not necessarily always a bad thing, I think.

777
00:33:08,120 --> 00:33:08,600
Yeah.

778
00:33:08,600 --> 00:33:15,080
Unfortunately, like, you can't ever know, but it seems to filter out a lot of people who

779
00:33:15,080 --> 00:33:18,440
could thrive, like an ostrand, other stuff like that.

780
00:33:18,520 --> 00:33:23,080
But you can't know, because you have to have some sort of filter.

781
00:33:23,880 --> 00:33:24,680
Like, it's...

782
00:33:24,680 --> 00:33:26,280
What kind of people are you thinking about?

783
00:33:26,920 --> 00:33:30,360
Well, in the case, like, this, like, sovereign engineering, like, coming to an island and all that, like,

784
00:33:30,360 --> 00:33:31,880
it definitely feels like people who can't travel.

785
00:33:31,880 --> 00:33:33,240
Oh, yeah, that's true. But on the ostracide, I mean...

786
00:33:33,240 --> 00:33:34,120
Oh, the ostracide.

787
00:33:39,120 --> 00:33:43,000
There's people who do a lot of interesting stuff in the world.

788
00:33:43,960 --> 00:33:47,800
Activists, people who publish, publish things, artists, stuff like that.

789
00:33:47,800 --> 00:33:53,160
Who just can't quite, they just don't quite know how to use it.

790
00:33:53,160 --> 00:33:55,800
They just can't put the time into using technology.

791
00:33:55,800 --> 00:34:00,440
And a lot of those existing platforms, like, you know, Facebook, log in with your phone number,

792
00:34:01,080 --> 00:34:04,840
like, are low enough, like, it works for them, because whenever they forget their password,

793
00:34:04,840 --> 00:34:07,320
they just, like, reset it with their phone number.

794
00:34:08,920 --> 00:34:11,560
What would be the main benefit for them, if you have them in mind?

795
00:34:12,120 --> 00:34:16,440
Because I saw you, I saw you were following some artists and someone making sketches of

796
00:34:16,440 --> 00:34:17,480
Pedro Natsu or whatever.

797
00:34:17,480 --> 00:34:20,520
Yeah, I'm trying to collect as many artists on those things as I can, because, like,

798
00:34:21,400 --> 00:34:24,200
it's the most... I like looking at that stuff, selfishly.

799
00:34:24,200 --> 00:34:28,200
I want to see more pictures, just because I like zapping that stuff, and...

800
00:34:30,920 --> 00:34:33,080
I think that's the main benefit, actually, of being there,

801
00:34:33,080 --> 00:34:36,760
that you can actually just straight up receive money very easily outright.

802
00:34:36,760 --> 00:34:37,640
Oh, absolutely, yeah.

803
00:34:37,640 --> 00:34:41,560
And artists are usually getting screwed by all the platforms, especially the smaller ones.

804
00:34:42,440 --> 00:34:46,600
So we see this in the value-for-value music world, where people, you know,

805
00:34:46,600 --> 00:34:52,920
switch from Spotify, to, you know, receiving boosts, and streaming sets, and so on, and they,

806
00:34:52,920 --> 00:34:59,480
like, 100 extra revenue from, like, you know, one cent a year to five bucks a week or something.

807
00:35:00,760 --> 00:35:02,760
It's a million extra revenue.

808
00:35:04,040 --> 00:35:08,200
I guess, like, that's kind of the way it feels on us, or it's been up for a while since we've

809
00:35:09,000 --> 00:35:12,600
since we've got it running, it's just... We know we have, like,

810
00:35:12,600 --> 00:35:16,040
it's proof that we have something here that works, and then...

811
00:35:16,040 --> 00:35:18,440
It's still a small seedling, but it actually works.

812
00:35:18,440 --> 00:35:20,280
Yeah, and the next hard part is just, like, how to...

813
00:35:21,640 --> 00:35:23,400
What does it actually take to onboard people?

814
00:35:23,400 --> 00:35:25,880
What does it actually take to make this work?

815
00:35:25,880 --> 00:35:28,360
It's like, we're developing any application or building anything.

816
00:35:28,360 --> 00:35:34,600
It's like, you can get half the thing done in a night, in one good day of work,

817
00:35:34,600 --> 00:35:38,440
and then it takes two, three more weeks to get the other half finished.

818
00:35:38,440 --> 00:35:41,480
Yeah, the last five per cent are the hardest part,

819
00:35:42,200 --> 00:35:45,880
getting all the little details right, and polishing it, and fixing all the edge cases,

820
00:35:45,880 --> 00:35:48,200
and so on. That's kind of, that's where the pain is.

821
00:35:48,760 --> 00:35:49,320
Yeah.

822
00:35:49,320 --> 00:35:52,120
I'm not sure if my coding can fix that yet.

823
00:35:53,800 --> 00:35:55,480
Pablo is trying, Pablo is trying.

824
00:35:56,200 --> 00:36:00,840
It might be able to generate less bugs.

825
00:36:01,560 --> 00:36:05,400
I think it's going to be a different set of bugs, right?

826
00:36:05,400 --> 00:36:06,600
Yeah, yeah.

827
00:36:06,600 --> 00:36:09,880
It's not going to make the same mistakes that human makes, that human makes.

828
00:36:09,880 --> 00:36:12,200
It'll be interesting to see if it, like, because when you fix bugs,

829
00:36:12,200 --> 00:36:17,400
you don't generally write more code, and since it's an LLM and it outputs code,

830
00:36:18,200 --> 00:36:22,680
it will be interesting to see if it can fix bugs by not writing code,

831
00:36:22,680 --> 00:36:25,160
like rewriting it, basically, and keeping it working.

832
00:36:26,120 --> 00:36:31,800
Yeah, by coding agents, they are very good at maximizing the number of lines written

833
00:36:31,800 --> 00:36:33,720
metrics. They're not very good.

834
00:36:33,720 --> 00:36:36,360
Like, not a single commit that was done by my age,

835
00:36:36,360 --> 00:36:39,720
but one of my agents was, like, removing code, or refactoring,

836
00:36:39,720 --> 00:36:42,600
restructuring code, so that it's, like, less code.

837
00:36:42,600 --> 00:36:45,000
That metrics are extremely productive.

838
00:36:45,000 --> 00:36:49,320
In an actual, you know, software development team, those are the best commits,

839
00:36:49,320 --> 00:36:52,520
like, something that yanks out, like, 10,000 lines.

840
00:36:52,520 --> 00:36:54,280
That's what I've been doing, like, that's...

841
00:36:55,160 --> 00:36:59,400
In the last two months of work on NoStrudel is just removing code.

842
00:36:59,400 --> 00:36:59,960
Yeah.

843
00:36:59,960 --> 00:37:03,720
I think I've removed more, like, 20% of the code base has gotten now,

844
00:37:04,760 --> 00:37:05,880
eventually.

845
00:37:05,880 --> 00:37:09,960
And whenever I release it, the next version will have a significantly less code in it.

846
00:37:11,880 --> 00:37:12,840
But yeah, those are the best.

847
00:37:15,320 --> 00:37:17,560
Yeah, we'll see how it develops.

848
00:37:17,560 --> 00:37:21,960
I'm quite optimistic about it, because I think every single

849
00:37:22,840 --> 00:37:25,000
part of the whole thing will get better.

850
00:37:25,000 --> 00:37:29,000
Like, our understanding of where and when to use LLMs will get better.

851
00:37:29,000 --> 00:37:31,080
The models themselves will get better.

852
00:37:31,080 --> 00:37:33,720
The hardware will get better, which means the context winners will get bigger,

853
00:37:33,720 --> 00:37:34,840
all those kind of things.

854
00:37:34,840 --> 00:37:36,520
We'll have more specialized agents.

855
00:37:36,520 --> 00:37:39,240
I mean, that's what Pablo is currently focusing on.

856
00:37:39,240 --> 00:37:42,520
He's, like, building, like, NoStr experts, basically, that are...

857
00:37:42,520 --> 00:37:50,120
I think it's, like, the dynamic, like, what Yuzuz was demoing, and we'll...

858
00:37:50,200 --> 00:37:51,720
Yeah, there's, like, the dynamic...

859
00:37:51,720 --> 00:37:55,960
DVMCP stuff, where it's just, like, you have a marketplace of...

860
00:37:57,400 --> 00:38:00,280
In some sense, a marketplace of agents, but more than that, like,

861
00:38:00,280 --> 00:38:06,040
stuff that agents can access easily, and MCP tools that agents can use.

862
00:38:07,000 --> 00:38:10,760
And if you add, like, because it's all master-based,

863
00:38:10,760 --> 00:38:13,480
those things automatically have identity reputation,

864
00:38:13,480 --> 00:38:17,800
web of trust, so you can kind of filter out what is useful and what isn't,

865
00:38:17,800 --> 00:38:20,760
that you can actually... That's the coolest part.

866
00:38:20,760 --> 00:38:25,000
It's like, you can give agents money so that they can pay their own stuff.

867
00:38:25,000 --> 00:38:26,520
That's, like, so amazing.

868
00:38:26,520 --> 00:38:27,640
That's the coolest part about...

869
00:38:28,760 --> 00:38:32,200
To your point from before, that's something that OpenAI really can't do.

870
00:38:32,200 --> 00:38:33,880
No, they can't, because they won't...

871
00:38:33,880 --> 00:38:38,040
An OpenAI agent, if you're running an OpenAI agent on OpenAI's platform,

872
00:38:39,080 --> 00:38:42,280
it will never automatically, without authentication or anything,

873
00:38:42,280 --> 00:38:43,880
it will never automatically connect to a group.

874
00:38:44,280 --> 00:38:48,680
They can't do it, because they themselves have to take care of the KYC,

875
00:38:48,680 --> 00:38:51,160
of the betting, of what is allowed, what isn't, what is okay,

876
00:38:51,160 --> 00:38:53,480
what isn't, what is trusted, what isn't, and so on.

877
00:38:53,480 --> 00:38:55,640
And Noster pushes all of that to the edges.

878
00:38:55,640 --> 00:38:56,840
Noster doesn't have to do that.

879
00:38:56,840 --> 00:38:58,680
There is no Noster CEO that has to do it.

880
00:38:59,560 --> 00:39:01,000
You as the developer, or you as the user,

881
00:39:01,000 --> 00:39:04,840
you have to figure out, okay, what are my trust-worthy sources,

882
00:39:04,840 --> 00:39:07,240
what is my point of view, and so on.

883
00:39:07,240 --> 00:39:09,240
And it relates back to what we said at the beginning,

884
00:39:09,240 --> 00:39:13,080
where it's like the fact that you give users keys,

885
00:39:13,080 --> 00:39:17,480
and the signing of the data moves the power to the edge of the network,

886
00:39:17,480 --> 00:39:19,560
to the users themselves, the keyholders have the power,

887
00:39:19,560 --> 00:39:21,640
not the servers anymore, the keyholders.

888
00:39:21,640 --> 00:39:23,320
The user is not on the server anymore.

889
00:39:23,320 --> 00:39:24,360
Exactly.

890
00:39:24,360 --> 00:39:27,240
That's why, I've been seeing this interesting,

891
00:39:27,240 --> 00:39:29,800
trying to rewrite, when I write Noster apps,

892
00:39:29,800 --> 00:39:34,120
I'm trying to switch the verbiage, because there is no login button.

893
00:39:34,120 --> 00:39:35,560
Exactly, exactly.

894
00:39:35,560 --> 00:39:37,800
It's also not necessarily user accounts.

895
00:39:38,520 --> 00:39:41,080
I spoke about this in the past, I think with Pablo and some other people,

896
00:39:41,080 --> 00:39:45,720
but it's like, to me, what I call them is cryptographic identities,

897
00:39:45,720 --> 00:39:46,840
but that's a mouthful.

898
00:39:46,840 --> 00:39:52,440
But the difference between a regular platform profile

899
00:39:52,440 --> 00:39:53,960
and the cryptographic identity is like,

900
00:39:53,960 --> 00:39:57,160
you can create a cryptographic identity anywhere.

901
00:39:57,160 --> 00:39:59,320
You can lock yourself in a bathroom and flip a coin,

902
00:39:59,320 --> 00:40:01,240
and you walk out with a cryptographic identity.

903
00:40:01,240 --> 00:40:04,520
So to your point, the identity, the profile, the username,

904
00:40:04,520 --> 00:40:07,080
it's not on the server, it's somewhere else.

905
00:40:07,080 --> 00:40:08,360
And that's what makes it so powerful.

906
00:40:08,520 --> 00:40:11,000
And to the agentics stuff,

907
00:40:11,000 --> 00:40:14,760
and agents have identity too, and need identity too,

908
00:40:14,760 --> 00:40:16,600
and need reputation too, and all those kind of things.

909
00:40:16,600 --> 00:40:17,640
That's the thing they're missing.

910
00:40:17,640 --> 00:40:18,360
That's the thing they're missing on the exit.

911
00:40:18,360 --> 00:40:19,640
That's why, openly, I can't do it.

912
00:40:19,640 --> 00:40:21,320
Because they don't give their agents identities.

913
00:40:21,320 --> 00:40:23,480
Yeah, and every agent can have an NPAP,

914
00:40:23,480 --> 00:40:24,920
but with an NPAP, it also has a wallet.

915
00:40:24,920 --> 00:40:27,160
But if they were doing it right in their own system,

916
00:40:27,160 --> 00:40:29,080
like they would give each agent an email address.

917
00:40:29,640 --> 00:40:31,000
So it could sign up for services.

918
00:40:31,000 --> 00:40:31,960
Sure, but then they're boxed in.

919
00:40:31,960 --> 00:40:34,040
Yeah, it's still a main problem, but like...

920
00:40:34,040 --> 00:40:35,000
Which is not very useful.

921
00:40:35,000 --> 00:40:37,560
But they don't even, they haven't even realized that yet.

922
00:40:38,120 --> 00:40:39,320
Oh yeah, I agree, I agree.

923
00:40:39,320 --> 00:40:42,440
And also most of the AI people haven't realized

924
00:40:42,440 --> 00:40:44,760
that they are going to run into the problem

925
00:40:44,760 --> 00:40:46,360
of web of trust reputation,

926
00:40:46,360 --> 00:40:49,000
which data to trust payments and micro payments.

927
00:40:49,000 --> 00:40:51,720
And you can't do micro payments on Stripe, you just can't.

928
00:40:51,720 --> 00:40:53,800
You have to build up your own credit lines and so on.

929
00:40:53,800 --> 00:40:56,040
You need to sign in to activate your subscription.

930
00:40:56,040 --> 00:40:58,600
Yeah, not only that, first of all, you need to KYC,

931
00:40:58,600 --> 00:41:00,440
but you can't do micro payments on credit drills.

932
00:41:00,440 --> 00:41:02,760
You can't, you just can't, it's not feasible.

933
00:41:02,760 --> 00:41:04,360
It's not economically feasible.

934
00:41:04,360 --> 00:41:07,160
Because I always keep coming back to the same example.

935
00:41:07,240 --> 00:41:10,520
That's why every single New York Times subscription and so on

936
00:41:10,520 --> 00:41:12,920
is like, you know, at least five bucks a month.

937
00:41:12,920 --> 00:41:14,920
Because everything else is not economically feasible.

938
00:41:14,920 --> 00:41:21,000
You cannot settle like a fraction of a penny using Stripe,

939
00:41:21,000 --> 00:41:23,080
you know, like using any credit system.

940
00:41:23,080 --> 00:41:23,800
Too much overhead.

941
00:41:23,800 --> 00:41:25,240
You can't, you have credit risk.

942
00:41:25,240 --> 00:41:27,480
You need to know the identities of the counterparties

943
00:41:27,480 --> 00:41:29,800
because credit can go bad and so on.

944
00:41:29,800 --> 00:41:31,160
You can only do it with money proper,

945
00:41:31,160 --> 00:41:32,840
which means you can only, on the internet,

946
00:41:32,840 --> 00:41:33,960
you can only do it with Bitcoin.

947
00:41:33,960 --> 00:41:35,960
Bitcoin is the only money proper we have online.

948
00:41:36,600 --> 00:41:38,280
It's like, that's, that's the thing.

949
00:41:38,280 --> 00:41:40,920
And that's, that's what they will need to figure out

950
00:41:40,920 --> 00:41:43,960
very, very painfully as they run into these walls.

951
00:41:44,760 --> 00:41:49,800
The meantime, what's cool is with the fact that we can build,

952
00:41:49,800 --> 00:41:52,280
we have an open system, is that we can,

953
00:41:53,400 --> 00:41:57,560
we can just buy their stuff and resell it.

954
00:41:57,560 --> 00:41:58,120
Yeah.

955
00:41:58,120 --> 00:42:01,080
I mean, that's what's happening with Paper Q and some other things.

956
00:42:01,080 --> 00:42:03,640
Like they're building very powerful tools.

957
00:42:03,640 --> 00:42:04,280
Yeah.

958
00:42:04,760 --> 00:42:07,160
Google search for a while was incredibly powerful.

959
00:42:07,720 --> 00:42:12,520
And like, if you don't want to go in the friction,

960
00:42:12,520 --> 00:42:14,040
if you don't, if you're not in their system,

961
00:42:16,920 --> 00:42:19,880
we can, we have a system now where we can just buy it and resell it.

962
00:42:19,880 --> 00:42:21,960
You can basically, exactly.

963
00:42:21,960 --> 00:42:22,840
You can set it free.

964
00:42:22,840 --> 00:42:25,400
You can, you can put it out of the box and set it free.

965
00:42:25,400 --> 00:42:28,280
That's, in some sense, that's what we are doing with our

966
00:42:28,280 --> 00:42:29,400
user-generated data.

967
00:42:29,960 --> 00:42:32,600
Like if you, in some sense, that's what we're doing.

968
00:42:32,680 --> 00:42:35,480
If you, if you, if you're a YouTube and you have all your videos on YouTube,

969
00:42:36,200 --> 00:42:39,080
you can, right now, build a script,

970
00:42:39,080 --> 00:42:42,120
use a little download or whatever to download all your videos,

971
00:42:42,120 --> 00:42:44,520
sign them, upload them to Blossom.

972
00:42:44,520 --> 00:42:47,560
Like you can make them truly yours and you can set them free.

973
00:42:49,720 --> 00:42:51,720
And it's the same thing about the agent stuff.

974
00:42:51,720 --> 00:42:53,800
It's just, it's exactly the same principle.

975
00:42:54,680 --> 00:42:56,520
So that's, I think, where it's heading.

976
00:42:56,520 --> 00:42:57,720
But it's going to take a long time.

977
00:42:57,720 --> 00:43:00,760
Like this is, this is definitely not going to happen overnight.

978
00:43:00,840 --> 00:43:04,440
And it's going to take a lot of, you know, grinding away and engineering.

979
00:43:04,440 --> 00:43:07,320
It's like, to me, and I mentioned this a couple of days ago, I think.

980
00:43:08,520 --> 00:43:14,120
It's, it's very similar to back in the 90s where, like, Linux was

981
00:43:14,120 --> 00:43:15,960
basically still a joke in some sense.

982
00:43:15,960 --> 00:43:22,360
And, and, and Microsoft had like 99.9% market dominance or something ridiculous.

983
00:43:22,360 --> 00:43:24,600
It was like a microsoft, absolutely everything.

984
00:43:25,160 --> 00:43:29,400
And it was just a couple of nerds kind of chipping away at the problem

985
00:43:29,400 --> 00:43:33,640
and building all these little tools and building everything out in the, in the

986
00:43:33,640 --> 00:43:35,400
new slash Linux world.

987
00:43:36,680 --> 00:43:37,960
And look where we are now.

988
00:43:37,960 --> 00:43:39,800
You know, it's only a couple of decades later.

989
00:43:39,800 --> 00:43:40,680
They couldn't compete.

990
00:43:40,680 --> 00:43:42,280
They just too much, too much.

991
00:43:42,280 --> 00:43:46,120
But that's, that's what excites me about like the DVMCP is that we're already seeing

992
00:43:46,920 --> 00:43:47,800
some of it working.

993
00:43:48,360 --> 00:43:48,680
Yeah.

994
00:43:48,680 --> 00:43:50,280
I didn't expect it to work so quickly.

995
00:43:50,280 --> 00:43:55,320
But like the ability for what excites me, I want to look more into is,

996
00:43:56,200 --> 00:44:00,520
because it's, it's interesting to configure your agents to be set up for certain tasks.

997
00:44:00,520 --> 00:44:00,760
Yeah.

998
00:44:01,320 --> 00:44:02,120
Build your own specialist.

999
00:44:02,120 --> 00:44:06,440
It's going to be interesting to see how far an agent can go starting from a blank slate

1000
00:44:07,080 --> 00:44:11,160
and just find a building itself with its own capabilities.

1001
00:44:11,160 --> 00:44:14,520
Because a lot of the like the notion of discovery when it can go out and browse

1002
00:44:14,520 --> 00:44:19,000
noster and make decisions like all Weber Trust, what, what can be trusted, what can't be trusted.

1003
00:44:19,000 --> 00:44:20,280
You give it a perspective like that.

1004
00:44:21,240 --> 00:44:25,160
It could start from a blank slate and you've given a task and it could just

1005
00:44:25,320 --> 00:44:30,760
pull in other tools, maybe in the future, other agents to complete the task and then

1006
00:44:30,760 --> 00:44:32,440
shut down and reset.

1007
00:44:33,160 --> 00:44:37,080
To me, what's magical is, is the life component of it.

1008
00:44:37,080 --> 00:44:40,920
It's like you're working on something and it doesn't work because it's just, it's very new.

1009
00:44:40,920 --> 00:44:41,640
It's a new library.

1010
00:44:41,640 --> 00:44:42,760
It's a new thing, whatever.

1011
00:44:42,760 --> 00:44:47,480
And then someone publishes a snippet and your system can automatically,

1012
00:44:47,480 --> 00:44:49,960
will automatically pick it up and then it works.

1013
00:44:50,760 --> 00:44:53,880
It's like, you know, someone just released a library or a way of doing things and so on.

1014
00:44:53,880 --> 00:44:56,920
And your agent is confused, like, I don't know how to use this thing.

1015
00:44:56,920 --> 00:44:59,960
And then he publishes almost like a little tutorial for the agent,

1016
00:44:59,960 --> 00:45:01,080
which is very easy to do.

1017
00:45:01,080 --> 00:45:05,000
That's the idea of the code snippets and then it magically works.

1018
00:45:05,000 --> 00:45:07,000
That's so mind-blowing and that works already.

1019
00:45:07,000 --> 00:45:09,000
Like, that's what we tried to show off yesterday.

1020
00:45:09,000 --> 00:45:12,520
I'm not sure if we managed yet, but it's very hard to explain all this.

1021
00:45:13,320 --> 00:45:17,720
But you see it, you see it in playing out in slower sense, like existing developer world.

1022
00:45:17,720 --> 00:45:22,680
Like whenever someone comes out with a new UI framework, like the newest hottest thing,

1023
00:45:22,760 --> 00:45:24,440
it takes a while for people to pick it up.

1024
00:45:24,440 --> 00:45:28,680
Because what they're waiting for is the first developers

1025
00:45:28,680 --> 00:45:31,560
to start writing blog posts about it, to start explaining.

1026
00:45:31,560 --> 00:45:33,080
Yeah, you have to understand it first.

1027
00:45:33,880 --> 00:45:36,520
And basically give instructions to the people who come next.

1028
00:45:37,320 --> 00:45:41,320
That's kind of what, yeah, now we're just doing, we're just going to try to do that with AI.

1029
00:45:42,200 --> 00:45:43,240
Just give it instructions.

1030
00:45:44,200 --> 00:45:46,040
Yeah, I mean, that's what vibe coding fixes.

1031
00:45:46,040 --> 00:45:49,560
You don't have to understand what it's doing or what kind of code it produces.

1032
00:45:49,560 --> 00:45:51,480
It's just, ah, it just works and it looks great.

1033
00:45:51,480 --> 00:45:52,040
Fantastic.

1034
00:45:52,600 --> 00:45:54,440
Do you think that's going to be an issue down the line?

1035
00:45:55,320 --> 00:45:59,240
I follow some really, really good coders.

1036
00:45:59,240 --> 00:46:05,720
And they all basically talk about how using these tools, let me give an example.

1037
00:46:05,720 --> 00:46:08,520
It's like, you know, you had all these taxi drivers in New York,

1038
00:46:08,520 --> 00:46:10,760
and they knew the city in and out, inside out.

1039
00:46:10,760 --> 00:46:15,800
And then GPS came along and you outsourced that part of your brain

1040
00:46:15,800 --> 00:46:17,960
that understands the traffic flows of the city,

1041
00:46:17,960 --> 00:46:20,760
that understands like even like where stuff is and so on.

1042
00:46:20,760 --> 00:46:23,720
And now all of that has gone out the window.

1043
00:46:23,720 --> 00:46:26,920
Like, you know, like, I don't know how many decades in after GPS now,

1044
00:46:26,920 --> 00:46:31,240
but it's like most taxi drivers of the world don't understand this stuff anymore

1045
00:46:31,240 --> 00:46:33,240
and just like do whatever the GPS tells them to.

1046
00:46:33,240 --> 00:46:35,640
And they don't have a deep understanding of how the city works,

1047
00:46:35,640 --> 00:46:38,840
where to drive what's what the best routes are and so on.

1048
00:46:38,840 --> 00:46:46,760
And so the people I follow, their main criticism is that you outsourced so many decisions

1049
00:46:46,760 --> 00:46:51,000
and don't really, you even outsource understanding in some sense,

1050
00:46:51,000 --> 00:46:54,680
like you don't even understand yourself anymore what the code base is doing.

1051
00:46:54,680 --> 00:46:58,280
And this will be lead to kind of terrible results down the line.

1052
00:46:58,280 --> 00:47:05,160
Well, you definitely limit how much like, yeah, you outsource your understanding,

1053
00:47:05,160 --> 00:47:07,880
you outsource some of your ability, and you probably lose it.

1054
00:47:07,880 --> 00:47:12,280
So you probably become by like any human measurable terms, a worse coder.

1055
00:47:12,280 --> 00:47:20,040
But like, whether that produces bad, I mean, that's, that's a bad outcome for yourself,

1056
00:47:20,040 --> 00:47:23,000
because now you're doing stuff that you don't quite understand, you're relying on the AI.

1057
00:47:23,800 --> 00:47:28,760
But I don't know if it produces a bad outcome for the software itself.

1058
00:47:28,760 --> 00:47:32,920
Yeah, the software itself or the industry, because I think it's just,

1059
00:47:32,920 --> 00:47:39,400
it might split into two type categories of, it might split into what we see right now,

1060
00:47:39,480 --> 00:47:47,320
and in some sense, like the Linux kernels and the foundational software, the low level stuff is

1061
00:47:48,200 --> 00:47:53,080
written by wizards. Yeah, written by wizards. And, and then a lot of the other stuff is just like,

1062
00:47:53,080 --> 00:47:57,160
like the JavaScript websites, all the crappy apps. Already kind of vibe together,

1063
00:47:57,160 --> 00:48:01,160
which is a copy paste from Stack Overflow. Yeah, the copy paste, half work.

1064
00:48:01,720 --> 00:48:03,560
So there's, I think there's always a level,

1065
00:48:03,800 --> 00:48:09,560
it probably going to end up like that, where there's the low level stuff that the AI ends up

1066
00:48:09,560 --> 00:48:18,200
using is, is curated, made, made with intention. And then everything on top is just kind of

1067
00:48:18,200 --> 00:48:23,960
regenerated, retried, retried, retried, try again, try again, until someone gets something that looks

1068
00:48:23,960 --> 00:48:28,520
right. Yeah, fair enough. I think that's an interesting perspective. You see with like,

1069
00:48:29,480 --> 00:48:33,640
it's just all slower timescale, I think. You see it with like open source, like everything gets

1070
00:48:33,640 --> 00:48:37,320
forward. Your point is that's already happening. Yeah, something gets fork developer goes where

1071
00:48:37,320 --> 00:48:41,240
someone forks in. It's like, well, let me rewrite the UI now because it doesn't look right. Let me

1072
00:48:41,240 --> 00:48:47,400
update it from jQuery to something else. And you're just redoing work that someone else did. And I

1073
00:48:47,400 --> 00:48:55,480
was now it's just become bad work. So yeah, yeah, fair enough, fair enough. Yeah, I'm not sure how

1074
00:48:55,480 --> 00:49:01,400
I think about this. I feel like it's kind of too early to tell. I don't know. It's with the most

1075
00:49:01,400 --> 00:49:09,720
surprising outcome for me was that LLMs, especially if you actually speak to them, not type to them

1076
00:49:09,720 --> 00:49:17,320
or write to them, but speak to them, they make excellent Robodock debuggers, excellent kind of

1077
00:49:18,120 --> 00:49:23,000
brainstorming partners. It's really interesting. I never expect, I would have never expected them.

1078
00:49:23,000 --> 00:49:28,520
The output, the output side is so kind of up in the air. We don't really know how much is

1079
00:49:28,520 --> 00:49:32,280
capable on the output side, but the input side, the fact that we managed to make something that

1080
00:49:33,240 --> 00:49:39,640
kind of appears to understand like human language. Yeah. Like that is absolutely amazing. And that's

1081
00:49:42,120 --> 00:49:47,080
that's a massive because I remember working with some of that, like the early, like, you know,

1082
00:49:47,080 --> 00:49:51,720
like Siri on the original Siri version on the iPhone and all the AI assistants were released

1083
00:49:51,720 --> 00:49:56,520
after that. That's very, very new stuff. I worked with Dragon, naturally speaking,

1084
00:49:56,520 --> 00:50:02,600
which was basically you had to train it yourself by speaking like a book of text into the microphone

1085
00:50:02,600 --> 00:50:07,640
for like a week so that it understands your voice and how you say things. But like even when it

1086
00:50:07,640 --> 00:50:12,360
managed to convert the, even when we managed to convert the audio to text, because we had

1087
00:50:13,240 --> 00:50:16,200
that for a while. It worked for a while, yeah. You could tell it was just running a bunch of

1088
00:50:16,280 --> 00:50:20,280
statements. Yeah, sure. It was like hard coded to like certain tasks. Yeah, a lot of times

1089
00:50:20,280 --> 00:50:24,680
are different beasts. Yeah. And somehow it's like that is fixed almost overnight.

1090
00:50:26,600 --> 00:50:30,920
Because that's what I'm excited as we get like, I guess you could say the next generation of

1091
00:50:30,920 --> 00:50:36,600
models when they release is the whole MCP, like the tool call and capabilities. And we start getting

1092
00:50:36,600 --> 00:50:43,320
small models with that, the lightweight ones, then the ones that realistically can't reason

1093
00:50:43,400 --> 00:50:49,080
very well, can't come up with like, you can, you can use them almost as like a fuzzy layer

1094
00:50:49,080 --> 00:50:53,960
between like human language and calling API. Yeah, I mean, that's, I think that's what also,

1095
00:50:54,520 --> 00:51:01,720
in some sense, as I watch people using all this stuff, like extensively, and I use it myself as

1096
00:51:01,720 --> 00:51:06,760
well, it's like, I see a lot of people, for example, and, and Justin talked about this as well, it's

1097
00:51:06,760 --> 00:51:11,960
like, okay, it's a fantastic interface between you and your terminal, because you cannot remember,

1098
00:51:11,960 --> 00:51:18,920
you cannot remember, like the humans are very bad at remembering every single kind of flag of,

1099
00:51:18,920 --> 00:51:24,600
you know, FFmpeg or Git, or whatever command, you know, it's like, how, I know what I want to do,

1100
00:51:24,600 --> 00:51:30,600
but I can't do the mapping myself of what I want to do to exactly the, what are the right brackets

1101
00:51:30,600 --> 00:51:34,440
and the right flags and the right, whatever. That's what I mean by like, it's like, it's a human

1102
00:51:34,440 --> 00:51:40,280
interface for calling, yeah, it's really good at that. Yeah. And like, that's like,

1103
00:51:41,160 --> 00:51:44,840
I'm excited to get home and start like building some hardware and putting some experiments together

1104
00:51:44,840 --> 00:51:51,400
for this, because like, that is the piece that was missing, um, in like, home assistant style

1105
00:51:51,400 --> 00:51:58,120
AI. Oh, okay. So turn on my light. Yeah. It doesn't know which light. Yeah. If statement can't tell.

1106
00:51:58,120 --> 00:52:02,920
And now you can do it locally. Well, if you know, you can do it without inviting all the spies

1107
00:52:02,920 --> 00:52:08,600
into your living room. If you have two components in like a smart home, if you have a user location,

1108
00:52:08,680 --> 00:52:13,640
which can be detected, like with your own, by like motion or something. And then you have a light

1109
00:52:13,640 --> 00:52:18,920
bulb. Previously, you'd have to write an if statement saying look for a light bulb that's in the user

1110
00:52:18,920 --> 00:52:24,520
location. If they say this coming now, the AI can just kind of, for lack of a better word, just like

1111
00:52:24,520 --> 00:52:29,880
fuzz it, just kind of. Yeah. It won't figure it out on its own. It'll figure it out. Yeah. It will

1112
00:52:29,880 --> 00:52:34,200
figure out what you mean. That's like, that's just what's so fascinating. It's, it's gonna fail.

1113
00:52:34,920 --> 00:52:39,320
But it's like the 90% solution that actually works. Exactly. It doesn't matter that it's like,

1114
00:52:39,320 --> 00:52:45,000
I've just been working with a junior engineer to kind of basically use an LLM for like a

1115
00:52:45,000 --> 00:52:51,160
summarization task, those kind of things. And he was like, ah, yeah. He was like, I think I got

1116
00:52:51,160 --> 00:52:56,040
a working and so on. But it's like, he produces a different output every time I want to fix that

1117
00:52:56,040 --> 00:53:00,280
and so on. It's like, okay, well, yeah, well, that's, that's the whole point. Like you don't try to

1118
00:53:00,360 --> 00:53:04,680
fix that. Like that's gonna happen with LLM's always. And that's the point. That's why they are

1119
00:53:04,680 --> 00:53:08,200
as a good. That's why they work. It's why they work. They work for problems. They're not if

1120
00:53:08,200 --> 00:53:13,640
statements. Yeah. There's there's problems that are like extremely complicated, like parsing a

1121
00:53:13,640 --> 00:53:20,440
command out of human speech that they work for 90% of the time. Yeah, which is, and that's the

1122
00:53:20,440 --> 00:53:24,600
difference between natural language and computer language. Because natural, natural language is

1123
00:53:24,600 --> 00:53:28,920
always, it always leaves this wiggle room for, for interpretation. That's why it's so hard to do.

1124
00:53:28,920 --> 00:53:34,120
And you have to, you need the context to do the, to get the interpretation, right? And so it is

1125
00:53:34,120 --> 00:53:40,280
always, it will never be 100% never, like only computer language and math can do that, basically.

1126
00:53:40,280 --> 00:53:44,840
And that's the main difference. And so what we have now is a great translation layer, basically,

1127
00:53:44,840 --> 00:53:50,280
because what, what LLM's obviously are good at is actually they, they can read the whole man

1128
00:53:50,280 --> 00:53:55,240
page of FM FM packet. They usually are not going to make a mistake with like the bracket is in the

1129
00:53:55,240 --> 00:54:00,680
wrong spot or whatever, you know, like those kinds of things. That's an interesting point. Yeah.

1130
00:54:01,640 --> 00:54:06,600
It's the whole point of natural language is, is that it is fuzzy. Yeah. It is, and it has to be

1131
00:54:06,600 --> 00:54:11,800
because the world is fuzzy. Future is uncertain that you can't have perfect knowledge of the world

1132
00:54:11,800 --> 00:54:16,360
and all those kinds of things. You're just like one, you are part of the universe in the world.

1133
00:54:16,360 --> 00:54:19,880
It's impossible for you to have a perfect section. There's another bit of entropy we added, we

1134
00:54:19,880 --> 00:54:26,360
managed to add to the computer. Like you needed that entropy in the system for it to be able to

1135
00:54:27,880 --> 00:54:33,720
act like something that's in the real world. I think it's a necessary part to have a proper

1136
00:54:33,720 --> 00:54:39,000
interface between humans and computers in some sense. It is the Star Trek kind of, you just

1137
00:54:39,000 --> 00:54:44,760
talk to the computer kind of thing. Should we do, should we do one more, one more round of the,

1138
00:54:45,320 --> 00:54:49,080
before we hit like the waterfalls and the waves? Yeah.

1139
00:54:51,640 --> 00:54:58,120
But yeah, that's how I see it. It's like, and what to me, to me, it's always important to really

1140
00:54:59,320 --> 00:55:03,640
have a hand, get a hands-on feeling for this stuff and really play around with it for a long time

1141
00:55:03,640 --> 00:55:09,560
because most people I think they don't understand. I mean, who understands these things, you know,

1142
00:55:09,560 --> 00:55:14,120
it's like quantum mechanics. It's like, you know, who truly understands this stuff. But

1143
00:55:14,440 --> 00:55:18,840
even the engineers building their own black boxes, no one understands how they actually

1144
00:55:18,840 --> 00:55:24,600
reason how they work. We're very bad at inspecting them. We're very bad at figuring out what they

1145
00:55:24,600 --> 00:55:30,520
do inside, basically. That was my initial thought about it way back from a long time ago, of course,

1146
00:55:30,520 --> 00:55:35,800
when it came out. But it was like, I thought that they would be more useful once we understood

1147
00:55:35,800 --> 00:55:41,080
them more. But the fact that it's the fuzziness of it that actually makes it useful,

1148
00:55:41,800 --> 00:55:50,680
they almost work better when we specialize them. There's people who understand them how they work.

1149
00:55:51,240 --> 00:55:55,960
And they're probably not very complicated. But it's, you know, how it actually produces the

1150
00:55:55,960 --> 00:56:01,800
outcome it produces is, you can't comprehend, you can't realistically.

1151
00:56:03,800 --> 00:56:06,600
But I would disagree that there's people that understand how they work. It's very much like,

1152
00:56:06,600 --> 00:56:13,080
okay, physicists understand in a perfect scenario how particles behave. But no, physicists understand

1153
00:56:13,080 --> 00:56:19,320
the weather of the earth. That's what I mean. It's too much. It's too big. You can't understand it.

1154
00:56:19,320 --> 00:56:22,680
And you can't even build systems to help you understand it, because it's just too much.

1155
00:56:22,680 --> 00:56:26,360
You'll run into combinatorial explosion and basically the computer that's trying to

1156
00:56:26,360 --> 00:56:30,120
compute it's going to collapse into black hole. You don't have enough context. Too many states.

1157
00:56:31,000 --> 00:56:35,480
Yeah. And it's also, I don't think, I also don't think that, I mean, right now,

1158
00:56:35,480 --> 00:56:39,400
like, I don't understand how you work. I don't understand how you think, you know,

1159
00:56:39,400 --> 00:56:44,120
like, I don't understand how my team works. I don't understand how whatever, even my life works,

1160
00:56:44,120 --> 00:56:50,120
you know. That's an interesting, that's like, you have to, you still, like, I don't want to

1161
00:56:50,120 --> 00:56:54,280
call my wife useful, you know, like, that's not what I mean. That's not what I'm trying to say

1162
00:56:54,280 --> 00:56:59,800
at all. But you still find ways to work together. And you still, you know, are able to divvy up

1163
00:56:59,800 --> 00:57:07,320
tasks. You find a common interface. Yeah. And you don't have to understand the other completely

1164
00:57:07,320 --> 00:57:11,160
to, you know, it's the common interface. Like you said, that's what language is. It's a common,

1165
00:57:11,160 --> 00:57:17,160
it's a fuzzy enough common interface that I can say something. And 90% of the time, you'll understand

1166
00:57:17,160 --> 00:57:22,360
what I'm saying. We feel a lot of work. I mean, we have to learn the same language and so on.

1167
00:57:22,840 --> 00:57:28,920
It's learning the same complex. It's incredibly inefficient because I have to process my thoughts,

1168
00:57:28,920 --> 00:57:34,360
which are, you know, you have to turn what's in your head, it's worse. Yeah. That's a compression

1169
00:57:34,360 --> 00:57:38,760
algorithm. It's very bad. It's expensive. It's an expensive algorithm. But it's the only way we

1170
00:57:38,760 --> 00:57:44,040
found to, like, transfer anything. Because if I, if I transfer my thoughts to you, they'd be

1171
00:57:44,040 --> 00:57:47,800
completely incompatible. It's probably gonna, you know, my brain would probably melt.

1172
00:57:48,040 --> 00:57:52,760
Yeah. It's just two incompatible systems. It's like, it's like one computer system trying to

1173
00:57:52,760 --> 00:57:57,640
transfer, you know, it's database to another. It's like, yeah, if the database is lined up

1174
00:57:57,640 --> 00:58:02,200
perfectly, if they're like genetic copies of each other, you can transfer the data. You can move,

1175
00:58:02,200 --> 00:58:06,440
you can't even move data from a WordPress website to a WordPress website. Yeah. It's like,

1176
00:58:06,440 --> 00:58:11,560
almost like genetically different. Yeah, yeah. Everyone has a different sequel version or

1177
00:58:11,560 --> 00:58:15,400
something. And it's just everything breaks. Yeah. Yeah. And so like you need to just come with like,

1178
00:58:15,400 --> 00:58:22,040
it's like processed export language. So the matrix, like, I know Kung Fu, Kung Fu download

1179
00:58:22,040 --> 00:58:28,360
is a lie that will never work. Probably. Probably. Well, there's a lot of stuff that the matrix has

1180
00:58:28,360 --> 00:58:33,880
that aren't lies. That isn't a lie. A matrix is great. It's basically, you know, it's, it's,

1181
00:58:34,440 --> 00:58:42,280
it's Plato's cave just for the new, the new, actually, I watched it with a bunch of friends at the

1182
00:58:42,360 --> 00:58:48,360
Bitcoin having. You never saw it before? Not in its not entirety. Okay, it's kind of, I knew the

1183
00:58:48,360 --> 00:58:53,560
story and I'd seen clips of it, but I watched it like front to back. And that was, that was a

1184
00:58:53,560 --> 00:58:59,320
great movie. It's one of the best movies of my time. There's so much, I have never seen a movie

1185
00:58:59,320 --> 00:59:05,400
with so much, like, in the context of computers, I've never seen something that says so much symbolism.

1186
00:59:06,040 --> 00:59:10,600
Yeah. Like there's, almost every scene, there's something that relates to like,

1187
00:59:10,600 --> 00:59:14,440
computers or social media or something. You're, you're seeing play out right now.

1188
00:59:14,440 --> 00:59:18,440
And it came out in 1999, which is insane.

1189
00:59:18,440 --> 00:59:27,080
In 1999. It was when it came out. So it was filmed, like, you know, 97, 98. It's a, it's a, you know,

1190
00:59:27,080 --> 00:59:34,920
massive, culturally important movie and, and like a classic now for, for reason, you know.

1191
00:59:35,800 --> 00:59:39,240
It's just true. It was also the age difference between us. It's like, for me, it's like, I,

1192
00:59:41,080 --> 00:59:44,440
oh yeah, right. Okay. It's like, there's people that haven't seen Matrix did too.

1193
00:59:44,440 --> 00:59:48,600
I find that there's like a, I have a backlog of movies that people tell me. And like,

1194
00:59:48,600 --> 00:59:51,960
everyone like, have you seen this? Like, it's like a classic movie or something. Yeah.

1195
00:59:51,960 --> 00:59:55,640
I'm like, I try to just remember, and then they say the release date. I'm like, oh yeah,

1196
00:59:55,640 --> 00:59:57,480
I haven't seen that. Yeah. It came out before I was born.

1197
00:59:57,480 --> 01:00:03,320
But I'm in the same situation. And I, I, you know, I, I, I, I try to work through the backlog. And I,

1198
01:00:04,280 --> 01:00:09,000
I'm, I would say like 50% through or so. But it's like, I'm in the same situation,

1199
01:00:09,000 --> 01:00:13,960
which is people that are 10, 20, 30 years older than me. They will mention like, you know,

1200
01:00:13,960 --> 01:00:18,760
Soyland green or something or like sci-fi movies. Yeah, it's fantastic. Yeah.

1201
01:00:18,760 --> 01:00:24,040
It's fantastic. Soyland green is made of people. No, it's, it's great. And like all, you know,

1202
01:00:24,040 --> 01:00:29,160
space, obviously. Yeah. And, and all that. I just very recently watched Close Encounters of the

1203
01:00:29,240 --> 01:00:34,840
Third Kind. And like all these classic science fiction movies, I still have to watch the,

1204
01:00:34,840 --> 01:00:38,840
it's one with Keanu Reeves that I haven't seen yet. What's, because it goes around

1205
01:00:38,840 --> 01:00:43,400
an Oscar from time to time. Johnny, Johnny, mammonic. I never heard that.

1206
01:00:43,400 --> 01:00:49,080
And I just really want to watch it because it has the word mammonic in it. Like,

1207
01:00:50,120 --> 01:00:56,440
sea phrase. Yeah. Yeah. And it's about the data. Like he, he basically, he stores data in his brain,

1208
01:00:56,440 --> 01:01:00,760
like Mary, I propose to what we were talking about. And he's like a data career and the data is

1209
01:01:00,760 --> 01:01:06,520
somehow too important and almost fries his brain or something. Why, why are the stories back in so

1210
01:01:06,520 --> 01:01:12,280
interesting? Yeah, I mean, it was interesting times when, when computer and computing and this

1211
01:01:12,280 --> 01:01:17,560
stuff came around, it was very interesting times. Now it's kind of embedded in society and so on.

1212
01:01:17,560 --> 01:01:22,920
But the internet really was a very new world and a very foreign thing, you know, like the information

1213
01:01:23,000 --> 01:01:27,320
super highway and no, no, no, no one understood it at all. I think we're in a similar spot now.

1214
01:01:28,040 --> 01:01:34,040
Yeah. Especially with other lamps. Like it's a, no, we are, it's a similar shift.

1215
01:01:34,040 --> 01:01:38,680
Do you think, so the two things we're seeing right now, like the, we're sort of like the kind of

1216
01:01:39,480 --> 01:01:44,840
noster starting up the concept of cryptographic key pairs as identities. And then the

1217
01:01:45,160 --> 01:01:53,400
AI and large language models becoming a thing. Because both of those things, like they're there.

1218
01:01:53,400 --> 01:01:58,040
It's happening at the same time. It's a very interesting coincidence because I don't believe in

1219
01:01:58,040 --> 01:02:03,080
coincidence. Well, I couldn't be related in any way because the large language models are

1220
01:02:03,080 --> 01:02:08,840
systematically destroying the centralized providers because they can't do anything against spam.

1221
01:02:08,840 --> 01:02:14,040
Yeah. Because their systems are set up to block spam from a computer that has like always a

1222
01:02:14,120 --> 01:02:20,200
recognizable pattern. But the spam, when you, when you, when one person runs like 10 Instagram

1223
01:02:20,200 --> 01:02:25,720
accounts is that it can't detect that as spam anymore. The spam protection is also not something

1224
01:02:25,720 --> 01:02:30,360
you can, you can perfectly solve like one man's, one man's trash is another man's treasure, you

1225
01:02:30,360 --> 01:02:35,240
know, like that's the problem. And like, and but like this is happening at the same time that we

1226
01:02:35,240 --> 01:02:42,520
have a growing solution for cryptographic identity. So it's like, it's pushing on one side. It's

1227
01:02:42,520 --> 01:02:47,800
interesting. It's a coincidence that makes me wonder why it's happening at the same time.

1228
01:02:49,800 --> 01:02:53,800
I think it was Applebaum or was it the Sange? I'm not sure. But but he's like, for some reason,

1229
01:02:53,800 --> 01:02:58,760
the universe smiles on encryption. Like it doesn't have to work this way. But it does work this way

1230
01:02:58,760 --> 01:03:05,880
that encryption cryptography itself gives like it moves all the power to the individual. And

1231
01:03:05,880 --> 01:03:09,560
it's a very peculiar thing because there's no reason why the universe should work in that way

1232
01:03:09,640 --> 01:03:15,480
or has to work in that way. But it's basically like the US, China and Russia can't conspire

1233
01:03:15,480 --> 01:03:20,440
against you. If you use strong cryptography, there's nothing they can do. Yeah. Yeah. Like to access

1234
01:03:20,440 --> 01:03:29,400
to open the box to access the data. And and and now in the, you know, AI, LLM world and all those

1235
01:03:29,400 --> 01:03:35,720
kinds of things, it's like the Skynet being the most probable outcome, like the Terminator style.

1236
01:03:35,720 --> 01:03:40,040
Skynet, have you seen Terminator? The original Terminator? Yes. Okay. Yes.

1237
01:03:40,040 --> 01:03:44,440
Where robots take over the world, everyone dies and it's like, it's really bad. I think it's very

1238
01:03:44,440 --> 01:03:52,360
unlikely. It's a very unlikely outcome because the closed system can't compete. Yes. Somehow,

1239
01:03:52,360 --> 01:03:59,320
the universe smiles on open source software and open data and cryptographically signed and verified

1240
01:03:59,800 --> 01:04:04,840
data, which is out in the open. Yeah. It does not smile. Like in this scenario, the universe does

1241
01:04:04,840 --> 01:04:11,320
not smile on up on silos, you know, and siloed systems and silo software and so on. It's very

1242
01:04:11,320 --> 01:04:16,440
much like it's probably not. We talked about this a little bit, but it's like, there's a reason why

1243
01:04:17,240 --> 01:04:25,480
totalitarian societies and can't deal with compounding errors. And you need outside force to prop up

1244
01:04:25,480 --> 01:04:30,680
the silo. Yeah. Yeah. And so a jungle or like a market-style system, like a marketplace,

1245
01:04:30,680 --> 01:04:35,800
an open market, which is jungle-like system, doesn't need an outside force. Yeah. So if you

1246
01:04:35,800 --> 01:04:41,160
have an open self-stabilizing system, it's always superior to a close, it has to be enforced from

1247
01:04:41,160 --> 01:04:47,560
the outside system because you need energy, you need energy from the outside to stabilize the system.

1248
01:04:47,720 --> 01:04:56,280
Yeah. You need that. That's the general principle. Yeah. So and this upkeep and this applying force

1249
01:04:56,280 --> 01:05:01,240
on it and so on has a cost. And so it will bankrupt you on a long enough time frame.

1250
01:05:02,760 --> 01:05:08,680
Yeah. Yeah. It's it's. That's how I see it, at least. You know, like maybe I'm also full of

1251
01:05:08,680 --> 01:05:14,840
shit, but it's like history kind of shows. There's something fundamental there because I think if

1252
01:05:14,840 --> 01:05:17,800
you look at a lot of if you look at a lot of systems, if you look at nature and whatnot,

1253
01:05:19,320 --> 01:05:29,560
a surface level understanding, like successful outcomes continue whatever is can be whatever is

1254
01:05:29,560 --> 01:05:37,080
happening. Errors compound. Yeah. And it that seems to be a fundamental thing where in every

1255
01:05:37,080 --> 01:05:43,640
system, you see it a lot in computer systems where the successful operation of a task

1256
01:05:44,600 --> 01:05:52,360
will just produce the outcome. But if you introduce 1% air, it can't correct. Like there's very

1257
01:05:52,360 --> 01:05:57,720
huge styles of systems that have air correction. Yeah. Yeah. And the reason it's so necessary is

1258
01:05:57,720 --> 01:06:02,440
because 1% then 1 more percent, then 1 more percent, then 1 more percent, and 1 more. And

1259
01:06:03,240 --> 01:06:08,760
successes don't correct in the opposite direction in almost all cases. So like the good example of

1260
01:06:08,760 --> 01:06:14,280
this is like, you try to, I mean, everyone has like their parents or their friends computer,

1261
01:06:14,280 --> 01:06:20,120
and you try to explain to them why it's going slow. Like, you know, we probably know why it's

1262
01:06:20,120 --> 01:06:24,840
slow. It's like too much is installed, there's too much extensions, there's a program running,

1263
01:06:24,840 --> 01:06:29,720
100 programs running in the background. But like, it's so many little 1% air,

1264
01:06:30,600 --> 01:06:35,160
they're like all compound, and the system doesn't have any way of correcting that because it's just

1265
01:06:35,160 --> 01:06:40,200
one more little thing, one more little, and so it just naturally basically falls apart and becomes

1266
01:06:40,200 --> 01:06:48,840
slow. That's an interesting point. It's like the, I talked about this a bunch in the Bitcoin

1267
01:06:48,840 --> 01:06:57,320
context, also in the cluster context. The reason why signing data sets the data free is because

1268
01:06:57,320 --> 01:07:03,560
computers are insanely good at creating perfect copies. Yeah. So, and DNA is also good at that.

1269
01:07:03,560 --> 01:07:07,720
That's the other system that's really good at it. But DNA makes certain mistakes. That's why we

1270
01:07:07,720 --> 01:07:12,520
have mutations and so on, like not all mistakes are catastrophic, some are beneficial. But, but

1271
01:07:12,520 --> 01:07:18,120
the thing about binary encoding and about computers is they're perfect copying machines. And

1272
01:07:19,640 --> 01:07:26,120
because we have perfect copying machines, look at that. Transporting something. Well, like

1273
01:07:26,120 --> 01:07:33,320
helicopters and everything. Oh, that's probably, that's probably like something's burning.

1274
01:07:33,960 --> 01:07:37,160
Because that's a, that's a big bucket of water.

1275
01:07:39,800 --> 01:07:44,760
It's like empty though. Yeah, probably. There's no way about the water flies behind a helicopter.

1276
01:07:44,760 --> 01:07:53,240
Yeah, that's true. Well, hopefully it's just like a little exercise thrill or whatever. But the,

1277
01:07:53,240 --> 01:07:59,160
the thing about being able to copy something perfectly is like it has very interesting side

1278
01:07:59,160 --> 01:08:07,640
effects. So, you can't, you can't create perfect copies in the real world. You just can't. It's like,

1279
01:08:07,640 --> 01:08:10,840
it doesn't matter if you look at cloning, it doesn't matter if you're copying it artwork,

1280
01:08:10,840 --> 01:08:15,640
it doesn't matter if you look at mass production, whatever of like, you know, Ikea chairs, who cares.

1281
01:08:15,640 --> 01:08:19,800
It's like, every little thing is going to be a little bit different. You don't have, you only

1282
01:08:19,800 --> 01:08:23,960
have arrow, like perfect copies, you only have an additional world where you're dealing with bits

1283
01:08:23,960 --> 01:08:27,720
of information and you can run arrow correction algorithms and so on and so forth. And so the

1284
01:08:27,800 --> 01:08:32,280
upside is we can't build systems like Bitcoin, we can't build systems like Nostra, the, the, the,

1285
01:08:33,800 --> 01:08:40,440
that's kind of, you know, like, that are insanely resilient because the same thing,

1286
01:08:40,440 --> 01:08:45,880
exactly the same thing is in multiple places at once, basically, because you cannot, you cannot

1287
01:08:45,880 --> 01:08:51,720
differentiate between a JPEG, like two JPEGs that are exactly the same have the same hash and

1288
01:08:51,720 --> 01:08:57,400
everything. What's the original? You can't know. Yeah, we've, we've successfully, as far as

1289
01:08:57,400 --> 01:09:05,400
information goes in that context, we've successfully solved the, um, location problem. Exactly. Exactly.

1290
01:09:05,400 --> 01:09:09,640
That's why I was about to say that next. That's why it becomes non-local. And that's why Bitcoin

1291
01:09:09,640 --> 01:09:13,080
is not local. And that's why Nostra stuff is not local. That's why we are able to also,

1292
01:09:13,080 --> 01:09:16,680
that's why cryptographic identities are non-local. That's why we can remove stuff from the service.

1293
01:09:16,680 --> 01:09:22,200
Yeah. That is what makes stuff non-local. Yeah. That's exactly it. That's exactly it. It is,

1294
01:09:22,200 --> 01:09:27,960
it is because you guarantee a perfect copy and therefore not only it ceases to become a copy in

1295
01:09:27,960 --> 01:09:33,720
that other location. It's not a copy anymore. No, it's the same thing. It's exactly. Exactly. Yeah. Exactly.

1296
01:09:33,720 --> 01:09:37,960
That's where the non-locality comes in. Which is really weird, right? It's, we don't, we don't

1297
01:09:37,960 --> 01:09:42,040
have this concept in a real world. There's no parallel. No, there's no parallel. I mean, only

1298
01:09:42,520 --> 01:09:47,240
quantum stuff. Let's not talk about that. But my point is, it's like because of the perfect

1299
01:09:47,240 --> 01:09:53,000
copies, that's what brings all these beautiful properties to, to, to the systems we care about,

1300
01:09:53,000 --> 01:09:57,080
to Bitcoin, to Nostra, to blossom as well. It's the same thing. It's the same principle. It's like,

1301
01:09:57,080 --> 01:10:02,440
it's exactly the same file. We can guarantee it. And so which one is the original is a meaningless

1302
01:10:02,440 --> 01:10:07,320
question. It's all the same thing. It's all the same thing. They're all the original. Yeah. For all,

1303
01:10:08,200 --> 01:10:15,160
you cannot know in a truly distributed system, you literally cannot know. You absolutely cannot know.

1304
01:10:15,160 --> 01:10:21,480
And so, where I was going with this is like the downside of that. The downside, it relates to what

1305
01:10:21,480 --> 01:10:26,360
you said, the fuzzing of the things and so on. It's like the downside of that with the LLMs and

1306
01:10:26,360 --> 01:10:30,280
all these things is like, you don't necessarily want to have perfect copies all the time. There's

1307
01:10:30,280 --> 01:10:36,760
a reason why we forget things, for example. Yes. Yes. You need very useful to forget things. Perfect

1308
01:10:36,760 --> 01:10:42,760
memory would be a curse. Another blessing. Yeah. You'd fill up too much. You'd have too much

1309
01:10:42,760 --> 01:10:46,840
because information becomes irrelevant. Your context changes, your environment changes,

1310
01:10:46,840 --> 01:10:51,720
and you need to get rid of things. Exactly. Exactly. Yeah. And I think that's also where open

1311
01:10:51,720 --> 01:10:57,560
systems are superior, where you have a certain way of figuring out what's relevant now,

1312
01:10:58,440 --> 01:11:02,520
what's kind of relevant in the channel based on, you know, your point of view,

1313
01:11:02,520 --> 01:11:07,400
based on the most recent stuff that just happened and so on. And that's very hard to do in close

1314
01:11:07,400 --> 01:11:11,240
systems. Well, you usually have to make a decision of what to forget. Exactly. Which is energy.

1315
01:11:11,240 --> 01:11:15,960
Exactly. Yeah. You can't. That's the important point. The point is you cannot centrally control

1316
01:11:15,960 --> 01:11:22,040
that. It has to be an organic emerging thing. You lose, it's a losing game because if you have to

1317
01:11:22,040 --> 01:11:27,720
spend energy to forget, then you've just doubled the amount of energy you need. Forgetting needs

1318
01:11:27,720 --> 01:11:33,560
to be effortless. Which means you can't like, like, every, like, every life form is optimized

1319
01:11:33,560 --> 01:11:39,400
to get it just as much energy as it needs. And if it needs extra and double energy just to

1320
01:11:39,400 --> 01:11:43,800
just make it so steady. Forgetting and dying needs to be effortless. Yes. And it usually is.

1321
01:11:43,800 --> 01:11:48,120
Yeah. But in a centrally controlled system, it is not. It's not. You need to prune the database.

1322
01:11:48,120 --> 01:11:52,600
Exactly. Which is a task you have to run. You have to do all these checks to make sure you're not

1323
01:11:52,600 --> 01:11:56,840
forgetting something that's important. Yeah. And that's why I'm so bullish on

1324
01:11:56,840 --> 01:12:00,600
Nusser. It's a substrate for all this stuff because it happens organically. It naturally forgets

1325
01:12:00,600 --> 01:12:04,840
the relays go down. Exactly. Things get deleted. If you don't care about data anymore, it's just

1326
01:12:04,840 --> 01:12:10,840
like whatever. That's why, like, that is one of the, everyone loves bringing it up around,

1327
01:12:10,840 --> 01:12:18,040
bringing it up every time. But that is one of the reasons that IPFS and those style of systems

1328
01:12:18,680 --> 01:12:25,320
couldn't work is they strive to not forget anything. Yeah. Yeah, yeah, yeah. It had

1329
01:12:25,320 --> 01:12:30,440
IPFS was aiming for the curse of perfect knowledge. Yeah. Yeah. The curse of perfect knowledge.

1330
01:12:32,280 --> 01:12:37,160
Yeah. That's the good way of putting it. Yeah. And they also tried to build a global state and,

1331
01:12:37,720 --> 01:12:41,720
you know, they made some assumptions which I think was very misguided like they. And they made

1332
01:12:42,520 --> 01:12:48,440
yeah, I mean, they launched a massive ship. Good for them. I watched. I watched. I was in the IPFS.

1333
01:12:48,440 --> 01:12:55,720
I was watching and very interesting. I was very enthusiastic about it. And they had like building

1334
01:12:55,720 --> 01:13:01,480
things that were developing clients. They were making new nodes. Shipcoin launched. The amount

1335
01:13:01,480 --> 01:13:06,840
of commits on the project like dropped in a half. Yeah, yeah. They just stopped shipping. Yeah.

1336
01:13:07,800 --> 01:13:12,600
Yeah. I mean, that's the, that's also the thing about money. It's not always a good thing. Like

1337
01:13:12,600 --> 01:13:19,800
it's oftentimes, it oftentimes corrupts. And, you know, like just throwing endless money at

1338
01:13:19,800 --> 01:13:24,520
the problem doesn't necessarily always fix it, which is interesting actually. Like it's, you know,

1339
01:13:24,520 --> 01:13:31,960
you sometimes it even makes stuff worse. So that's like case in point IPFS and quite a few other

1340
01:13:32,040 --> 01:13:38,680
projects as well. But yeah, I mean, IPFS tried to, like it leaned heavily into deduplication,

1341
01:13:38,680 --> 01:13:43,000
for example. That's why that's why I love loss them so much with exactly you need to,

1342
01:13:43,000 --> 01:13:47,160
because then you need to fit the pieces together again. And that's like all of that is computationally

1343
01:13:47,160 --> 01:13:50,760
expensive, actually hard. If you're missing a shard, what do you do, those kind of things. Yeah.

1344
01:13:50,760 --> 01:13:56,920
And like the left side of the brain solution that is both like, it's all to me, it's all kind of

1345
01:13:56,920 --> 01:14:01,400
left side of the brain, Bitcoin, Noster and Blossom as well, where it's just like, okay, we just have

1346
01:14:01,400 --> 01:14:07,560
this file, we only, we, you know, like, that's the file. That's the hash. That's it.

1347
01:14:08,360 --> 01:14:14,200
It's the previous point, like using energy to forget. It's also like, that's why Bitcoin got

1348
01:14:14,200 --> 01:14:21,720
right. So fundamentally right is it minimize the amount of energy needed to verify down to

1349
01:14:21,720 --> 01:14:28,840
something almost nothing measurable in the system. Because if you, that's more energy that needs

1350
01:14:28,840 --> 01:14:34,840
to be put into the system, if you're spending more to verify. You, yeah, you immediately see

1351
01:14:34,840 --> 01:14:40,760
that it, that the new information you get is proper, is valid and proper. You only need to,

1352
01:14:40,760 --> 01:14:48,360
that the block header basically. Yeah, it's, it's the end, it makes it very, very cheap to figure

1353
01:14:48,360 --> 01:14:53,960
out what to dismiss. Yeah. Very cheap, like almost free. The main, the main thing is to fix the

1354
01:14:53,960 --> 01:14:58,840
spam price. Yeah, yeah. And, and, and Bitcoin really has to figure it out because Bitcoin is

1355
01:14:58,840 --> 01:15:05,560
truly about building up something that doesn't forget forever. So you really have to get this

1356
01:15:05,560 --> 01:15:12,280
part right. It's interesting that it's necessary in some way. One piece. Yeah, those are fundamental

1357
01:15:12,280 --> 01:15:17,080
things. And I briefly, and we spoke about this yesterday as well, I briefly spoke with Pablo

1358
01:15:17,080 --> 01:15:21,640
about this. It's like, there are these certain ways that the universe works and you can't cheat

1359
01:15:21,640 --> 01:15:26,520
that. And on the LLM side, for example, and it is about this stuff. It's like, if you want to, if you

1360
01:15:26,520 --> 01:15:32,680
want to, if you want to artificially build something, if something doesn't emerge organically, if you

1361
01:15:32,680 --> 01:15:38,840
want to, like, you know, if you, if you, if you have a structure that needs upkeep, it sucks energy

1362
01:15:38,840 --> 01:15:43,880
out of the whole system. Yeah. So on a long enough time scale, it will just fall in crumble. Yeah,

1363
01:15:43,880 --> 01:15:49,480
because you can't, yeah, but you can't have, you can't have endless energy, energy input forever.

1364
01:15:49,640 --> 01:15:53,880
So, so only things that organically emerge, that's, that's the constant, basically, you know, like,

1365
01:15:53,880 --> 01:15:58,760
that's the thing that can always happen. Spontaneous order, like crystalline structures forming,

1366
01:15:58,760 --> 01:16:08,680
whatever it would have you, you know, those kind of things. And on the kind of information

1367
01:16:08,680 --> 01:16:13,800
processing side, what I think is so interesting, I mean, we went through the whole how our computer

1368
01:16:13,800 --> 01:16:17,640
language is different from that natural language is how it's compute different from like thinking

1369
01:16:17,720 --> 01:16:23,000
in some sense and so on. And now on the LLM side, that being a generalist and being a specialist

1370
01:16:23,000 --> 01:16:30,600
are in tension with each other. You can't be both. And what I see emerging right now is like, if you,

1371
01:16:31,320 --> 01:16:35,160
the specialists still need to communicate, they still need to have a shared language,

1372
01:16:35,160 --> 01:16:40,920
where, of course, you know, like the model context protocol comes in, DVMCP, all these things come in,

1373
01:16:40,920 --> 01:16:48,440
but you need to provide the shared substrate for the marketplaces and for those all of all,

1374
01:16:48,440 --> 01:16:53,160
for those kind of things. And to me, like, noster is absolutely perfect for that. And so I see,

1375
01:16:53,160 --> 01:17:00,200
I see noster as the breeding ground for specialists in some sense. Like, generalists will have as well,

1376
01:17:00,200 --> 01:17:08,680
but, but in terms of, in terms of housing and training specialist system specialist, LLM

1377
01:17:09,080 --> 01:17:13,800
specialist agents, I think noster is perfect. I don't think as of right now, we have a better

1378
01:17:13,800 --> 01:17:21,880
substrate. Yeah, it's a bold claim, I realize, but I believe it's true. I need to think about

1379
01:17:21,880 --> 01:17:29,720
the whole concept of like training the, training the agents, because that's everything I've,

1380
01:17:29,720 --> 01:17:36,040
I've only brushed up against them and like using them and trying to use them like specialized

1381
01:17:36,040 --> 01:17:40,840
tasks. I have not seen what's possible, like, training them and specialized. I know you can do

1382
01:17:40,840 --> 01:17:47,320
it. Do I wonder what, what, what would you want to, what kind of agent would you want to build

1383
01:17:47,320 --> 01:17:52,040
and what do you want to train it on? I don't know yet. That's a thing. Do you have a lot of personal

1384
01:17:52,040 --> 01:17:59,240
data that you want to feed it? You want to build a LLM version of hazard? Well, see, I don't, I think

1385
01:18:00,040 --> 01:18:06,360
like a database use case is the only use case I can think of like a specialized system. But

1386
01:18:06,360 --> 01:18:13,720
that's because we've had them forever. We've had, you know, like SQL databases. We've had those

1387
01:18:13,720 --> 01:18:20,040
for a long time and those are like a server that's been trained to retrieve data and sort it and

1388
01:18:20,040 --> 01:18:25,400
stuff like that, which the LLM, you know, we've seen that with like rag databases and various

1389
01:18:25,400 --> 01:18:30,680
and there's stuff that I haven't looked into yet. It's really good at like data retrieval.

1390
01:18:30,680 --> 01:18:33,720
So you, you would like to have a second brain in some sense?

1391
01:18:34,920 --> 01:18:39,960
Definitely. That's one I try to build. But I wonder if there's like other use cases that we just

1392
01:18:39,960 --> 01:18:46,040
haven't, that aren't possible because they can't be built in a perfect way that the computers require.

1393
01:18:47,320 --> 01:18:53,240
Because like data retrieval is data archiving and data retrieval, if you put enough effort into

1394
01:18:53,240 --> 01:19:00,120
it is a solved problem. You can get it 100% right. Yeah. But there's, what if there's other systems

1395
01:19:00,120 --> 01:19:05,800
like specialized systems, you know, the concept like training and AI that you can't, we can't do

1396
01:19:05,800 --> 01:19:14,040
right now because we can't get them 100% right. I don't know. So it's just gonna be one of your

1397
01:19:14,040 --> 01:19:21,480
next side projects, get home, buy a bunch of graphics cards. I'm gonna see what I can run

1398
01:19:21,480 --> 01:19:28,200
with the hardware I have now. It's amazing what you can run. Like I, I, I keep being blown away,

1399
01:19:28,200 --> 01:19:33,480
but like, you know, my old thinkpad can run the old armour with llama too. And it's very useful.

1400
01:19:33,480 --> 01:19:40,440
It's like completely local open source model. That's the beauty of, that's what I was waiting

1401
01:19:40,440 --> 01:19:46,280
for to get jump into AI was to see that like something running locally can't, you still can't

1402
01:19:46,280 --> 01:19:50,840
make decisions. It can't run a whole pipeline. And, and you can't, it's pretty dumb when you

1403
01:19:50,840 --> 01:19:55,240
interact with it. Yeah. Like like doing tasks. It's smart enough to be useful. Yeah. That's kind

1404
01:19:55,240 --> 01:20:01,880
of interesting, an interesting threshold to cross. Yes, because it definitely wasn't that way

1405
01:20:01,880 --> 01:20:09,080
a year ago. Yeah. I, I, from what I tested, it was slow, didn't understand a single thing,

1406
01:20:09,080 --> 01:20:15,720
merely understood English. And now it's like to the point of being useful and it's, it's improving.

1407
01:20:15,720 --> 01:20:22,680
Yeah. I think that the, the running is like the, the local part is absolutely key. Because on the,

1408
01:20:22,680 --> 01:20:26,920
on the, especially on the desktop, like it's, we discussed this at length the last couple of weeks,

1409
01:20:26,920 --> 01:20:31,960
but it's like the, the, the beauty of noster, the beauty of all these systems is that you don't need

1410
01:20:32,040 --> 01:20:39,480
an API key ever. Well, it's like, and, and so in this world, like if you would only, if, if I mean,

1411
01:20:39,480 --> 01:20:44,040
yes, okay, there's this wrappers and so on, like pay per queue and so on. But, but in the best case,

1412
01:20:44,040 --> 01:20:48,840
it's like, this, all the stuff runs locally on the, on your computer. We have that way back

1413
01:20:49,560 --> 01:20:54,440
a little while ago was like, how else are you going to trust it? Yes. You can't see it's working.

1414
01:20:54,440 --> 01:20:59,640
Like locally, I can see every piece working. If it's running on open AI servers, if it's running

1415
01:20:59,720 --> 01:21:03,960
on someone else's servers, it becomes a trust me of our situation. Like, yeah, you're not in

1416
01:21:03,960 --> 01:21:07,480
charge. And the whole point, the whole point of the world we're trying to build is to put the

1417
01:21:07,480 --> 01:21:11,480
users in charge. Yeah. But like, if I'm going to build a, like a personal assistant or something

1418
01:21:11,480 --> 01:21:16,680
that's just going to voice commands, one stuff when we do stuff, DM people for me, send emails.

1419
01:21:17,720 --> 01:21:23,400
Like, unless I can absolutely see that the system is working and I can trust it almost completely,

1420
01:21:23,960 --> 01:21:28,600
it will never work. I mean, you can contain it yourself. Like it can be just don't have access

1421
01:21:28,680 --> 01:21:33,880
to it and be on your local network, you know, and you, you see everything that comes in and out.

1422
01:21:33,880 --> 01:21:36,600
Yeah. Yeah. But like, that's, that's the important part. You see,

1423
01:21:41,240 --> 01:21:46,440
you see it, there's something there like, if you can see it working, you can trust it a lot more

1424
01:21:46,440 --> 01:21:52,120
than like, talk to see it more. Of course, you trust a lot more. The other, the other way is like,

1425
01:21:52,200 --> 01:21:59,320
you know, Siri or whatever has access to all your DMS, all your emails, all your, your whole phone,

1426
01:21:59,320 --> 01:22:03,720
all your screen at all times and all those kinds of things. And of course, it's sending stuff to

1427
01:22:03,720 --> 01:22:08,600
a server. How can you trust it? It's impossible to trust it. You shouldn't trust it at all. Like,

1428
01:22:08,600 --> 01:22:13,240
you shouldn't trust it by default. Yeah. But in the local first world, it's very different.

1429
01:22:13,240 --> 01:22:16,920
It doesn't let you adapt to it either. Because if you say the wrong thing, Siri, and it calls

1430
01:22:16,920 --> 01:22:21,720
the wrong person, you don't know if it was Siri making a mistake or whether you said the wrong thing.

1431
01:22:22,520 --> 01:22:26,520
But if you see, oh, yeah, you don't know how it works. Sure, sure. Then you can, oh, I mean,

1432
01:22:26,520 --> 01:22:32,360
that's part of trust, right? It's like, you, you, you, if you can't trust the thing, it won't betray

1433
01:22:32,360 --> 01:22:35,960
you. But if you can't trust the thing and it portrays you, you don't even know what happened.

1434
01:22:35,960 --> 01:22:42,760
Yeah. You know what I mean? It's like, it's, it's unknown. Yeah. Yeah, yeah. No, but it's, it's like,

1435
01:22:44,280 --> 01:22:50,600
I, I still, I'm still surprised that other communities didn't stumble into Noster

1436
01:22:50,600 --> 01:22:56,360
and found it useful enough to, to go kind of very hard on it. There is a, I talked with Pablo

1437
01:22:56,360 --> 01:23:00,840
about this like a couple of weeks ago. And there's the local first community, for example, like,

1438
01:23:00,840 --> 01:23:05,000
at the self-hosting community and so on, lots of hacker community and so on. Like,

1439
01:23:05,000 --> 01:23:09,160
Noster is kind of by design local first in some sense, you know, it's like you said with the

1440
01:23:09,160 --> 01:23:14,280
community relay and so on. You can run your own home relay, do everything there. You can use Noster

1441
01:23:14,280 --> 01:23:19,640
as a substrate. And, and it works beautifully. It's very, very useful for this kind of stuff.

1442
01:23:20,280 --> 01:23:25,320
And, and I'm, I'm, because there is all these people that have the same concerns and they,

1443
01:23:25,320 --> 01:23:29,000
they shared the exact concern that you just mentioned with like, I, I never want to send

1444
01:23:29,000 --> 01:23:34,040
stuff to a server I don't control ever and so on. And, and they are not, they, they didn't,

1445
01:23:35,080 --> 01:23:40,680
I don't know. I, I don't know. Like, maybe it's just time. Maybe it's just like you mentioned with it.

1446
01:23:40,680 --> 01:23:45,160
You know, new, new UI framework and someone needs to write a blog post about it that other people

1447
01:23:45,160 --> 01:23:50,680
understand and so on. But, but it's, yeah, I'm, I'm still surprised that we don't have

1448
01:23:51,480 --> 01:23:55,480
more influx from, from other people. I mean, Bitcoin doesn't help. It's like you, you,

1449
01:23:55,480 --> 01:23:59,160
you go on Noster and it's like all Bitcoin people talking to Bitcoin stuff and everyone hates Bitcoin.

1450
01:23:59,160 --> 01:24:01,640
It's like, from the followers to the exact reason.

1451
01:24:04,200 --> 01:24:07,560
I was starting to agree with you like the care, removing the character when it was a mistake,

1452
01:24:07,560 --> 01:24:13,960
because the amount of philosophical posts that I have no interest in is growing. It's like, it's,

1453
01:24:14,840 --> 01:24:20,600
I, I, we should be short. There was the whole point. Yeah. It's like, we have to bring back the

1454
01:24:20,600 --> 01:24:25,240
character limit. Two hundred ten characters. Nothing more. Yeah.

1455
01:24:30,120 --> 01:24:34,040
Just think about it. Yeah. Idea is different ways of browsing. That's the beauty of it.

1456
01:24:34,040 --> 01:24:38,440
I was like, if I want that back, I can just make it. Yeah. You just filter things out. You don't

1457
01:24:39,000 --> 01:24:43,400
see it. I mean, there is, there is the, I think there are some relays that only accept

1458
01:24:44,120 --> 01:24:48,040
events that are hundred, 144 characters or shorter or something like that. Like, because I talked

1459
01:24:48,040 --> 01:24:52,280
with Pablo about it that he configured one of the relays like this and some other people cloned it.

1460
01:24:52,840 --> 01:24:57,160
So, you know, there, there are inklings of the revolution, but it's not,

1461
01:24:58,840 --> 01:25:05,160
we're, we're not even, there's not even dozens of ads as yet. I think most people enjoy the long

1462
01:25:05,160 --> 01:25:10,200
tweets and long, you know, most posts and so on. It's, and again, like I'm, you know,

1463
01:25:10,200 --> 01:25:14,200
I'm a huge fan of blog posts and articles and so on, but it's a different thing. It's a different

1464
01:25:14,200 --> 01:25:19,560
media type in my mind. I think like tweets should be short. Whatever is in a feed should be short.

1465
01:25:19,560 --> 01:25:24,920
I don't want to see like 5,000 word essay in my feet. No, but you're laughing, you're laughing,

1466
01:25:24,920 --> 01:25:29,480
but that's what we built now. Yeah. It's what everyone built. It's like, it really sucks. Yeah,

1467
01:25:29,480 --> 01:25:36,520
it's completely wrong. Like, that's why I think of changing it because I literally don't read

1468
01:25:36,520 --> 01:25:41,800
anything. If it's two paragraphs, you're in a different mode. You're in a different mode. Like,

1469
01:25:41,800 --> 01:25:45,080
if you're, if you're in a feed, you're in a browsing mode. Yeah, you're in a browsing and

1470
01:25:45,080 --> 01:25:51,960
discovery mode. If I find an article and it's both posted by someone that I, that I, that I trust

1471
01:25:51,960 --> 01:25:56,360
with their opinion on long form stuff, you know, like, like this person only shares interesting

1472
01:25:56,360 --> 01:26:03,080
stuff. I am going to put it into my, I'm going to read this later pile. And when I'm in the reading

1473
01:26:03,080 --> 01:26:07,720
mode, then I'm going to read it. And I have my apps and my things and my lists and so on that

1474
01:26:07,720 --> 01:26:12,520
I go through. But you mentioned like scrolling down, like the, the large amounts of text is

1475
01:26:13,160 --> 01:26:19,000
to put it. It's straight in terms. It's like, it's like dumping a large, it's, it's just expensive.

1476
01:26:19,000 --> 01:26:22,600
Yeah. It's like putting a large amount of input into a LLM. Yeah. The amount of

1477
01:26:23,560 --> 01:26:27,320
just the cost and the processing power to reduce that down and understand it.

1478
01:26:27,320 --> 01:26:32,440
It's a, it's a DDoS on your attention, basically. Yeah. Yeah. It's, it's, it was a big mistake. I,

1479
01:26:33,160 --> 01:26:37,880
yeah. I'm, I'm, I'm, I'm great. I'm glad you're converting. You know, it's like, I'm, I'm, if I

1480
01:26:37,880 --> 01:26:42,680
have to convert every single client developer one by one, I'm going to do that. We can enforce it

1481
01:26:42,680 --> 01:26:46,760
for sure. I mean, that's what, that's what the clients do. They hide like the read more by now.

1482
01:26:46,760 --> 01:26:51,240
Yeah. It's too long. You have read more. Yeah. But then that doesn't really even fix the problem

1483
01:26:51,240 --> 01:26:56,440
because it fixes the problem with the scrolling, but then the person who's making that article

1484
01:26:56,440 --> 01:27:01,480
never gets any engagement. Yeah. Because the first sentence that everyone reads doesn't convey any

1485
01:27:01,480 --> 01:27:05,640
meaning. Yeah. And if it would ever catch on that people would game it and would, would make the

1486
01:27:05,640 --> 01:27:10,760
first sentence really like, you know, like, click bait, basically. Yeah. Click bait, video header,

1487
01:27:10,760 --> 01:27:15,960
titles, all that stuff like that. Yeah. Yeah. Yeah. No. Tweets were perfect. It's like,

1488
01:27:16,840 --> 01:27:21,320
I never used them. I don't know. Everyone fucked everything up, man. It's like, because the

1489
01:27:21,320 --> 01:27:25,080
downfall was the Twitter removed the character limit. Like they made it twice as long. And then

1490
01:27:25,080 --> 01:27:31,240
they removed it completely. And yeah, I thought it was always the same. No, no, no. No, it was,

1491
01:27:31,240 --> 01:27:39,320
I think, I think it was 140 or something at first. And then 280, I think. And then they, they removed

1492
01:27:39,320 --> 01:27:44,040
it completely and so on. And it's like, it wasn't an issue. Like it, I really think it wasn't an

1493
01:27:44,040 --> 01:27:47,560
issue. Well, that's, that's where the whole concept of threads came out. And I'm not mistaken,

1494
01:27:47,560 --> 01:27:52,840
because we had to make your first post compelling to the point, almost like a title. And then you

1495
01:27:52,840 --> 01:27:58,200
would go on more. And I think threads, I still think I have this long thing on threads. I, I, you

1496
01:27:58,200 --> 01:28:04,680
know, I don't, I don't, I went on and on about this for many months. I gotta stop. Gave up a little

1497
01:28:04,680 --> 01:28:11,080
bit. But threats were ingenious, because, because you could, you could link them together. You could

1498
01:28:11,080 --> 01:28:16,360
still, it was kind of like a read more thing, right? Yeah. And, but also, like you could react to

1499
01:28:16,360 --> 01:28:21,640
every single paragraph almost, like to every single, you, in a tweet, you could fit two sentences,

1500
01:28:21,640 --> 01:28:27,560
maybe, you know? Yeah. And so, so, so you, you could also kind of see what people found most

1501
01:28:27,560 --> 01:28:34,680
interesting in this, like little wall effects in the thread, you know? And it was, it was very

1502
01:28:34,680 --> 01:28:39,640
good. It was very, very useful. Interesting. And it grew organically. That's the thing, like it grew

1503
01:28:39,640 --> 01:28:43,800
like this over time. And it all had its use. And, and now all of that is destroyed, basically.

1504
01:28:43,800 --> 01:28:48,280
In some way, like, like median and the way they do like, like the highlighter idea, like,

1505
01:28:48,280 --> 01:28:54,360
highlighting, kind of recaptures that, because you can visualize what points do people talk about,

1506
01:28:54,360 --> 01:28:58,680
highlight, share and stuff like that. And threats was a very organic thing. Yes. Around that.

1507
01:28:59,240 --> 01:29:05,320
Because there was some banger statement and that was like reposted and everything and like

1508
01:29:05,400 --> 01:29:09,560
lots of reactions to it and stuff. I want, I want to, one of these days, I'm going to like

1509
01:29:09,560 --> 01:29:14,920
some sort of client like highlight, some sort of client like highlighter because

1510
01:29:16,280 --> 01:29:20,120
the idea is still good. It's just another, it's like another app on my backlog, where it's like,

1511
01:29:20,120 --> 01:29:23,880
I want to see it work. Justin is, Justin is hopefully going to

1512
01:29:25,560 --> 01:29:31,800
build out his Rust multiplatform. And then, you know, we can just like one-shot native apps that

1513
01:29:31,800 --> 01:29:37,000
do exactly that. That would be my goal. That's an interesting point with AI and whatnot,

1514
01:29:37,800 --> 01:29:45,000
is the people who, like the Rust multiplatform idea, the people who build like foundational

1515
01:29:45,800 --> 01:29:52,280
systems, they usually, I mean, I kind of consider myself in this category too,

1516
01:29:52,280 --> 01:29:57,000
because I get so distracted by building a beautiful library, a beautiful toolbox,

1517
01:29:57,720 --> 01:30:02,600
that I never actually go build anything afterwards. When I do it's not as much as I should. It's like

1518
01:30:03,400 --> 01:30:08,760
maybe the equivalent will be like someone who designs an amazing workshop and they get,

1519
01:30:08,760 --> 01:30:12,600
they just enjoy designing a workshop so much that they forget to build anything else.

1520
01:30:14,040 --> 01:30:22,040
But the AI having the setting up your development environment for like two months and never writing

1521
01:30:22,120 --> 01:30:27,400
a line of code. Yeah, but that's the best, like the one shot in like a vibe coding idea is interesting

1522
01:30:27,400 --> 01:30:33,240
because we've kind of, that like does two things, it almost standardizes work. It makes it less

1523
01:30:33,240 --> 01:30:38,680
important whether it's your workshop or someone else's. And then it lowers the amount of effort

1524
01:30:38,680 --> 01:30:45,800
it takes to just use it. So like, it's like Rust multiplatform. He could build it out to be like

1525
01:30:45,800 --> 01:30:50,680
an amazing app framework. And then he doesn't even have to build any apps for it. If it just,

1526
01:30:50,680 --> 01:30:55,560
I could come along and just prompt an app into existence and use the foundation that he lay in.

1527
01:30:55,560 --> 01:31:02,760
Yeah, yeah. I mean, in some sense, and Pablo made this point also, in some sense, that's what high

1528
01:31:02,760 --> 01:31:08,440
level programming languages are. It's like, I mean, everything, you know, boils down to

1529
01:31:08,440 --> 01:31:15,160
to assembler and binary after that, you know, and it's like everything else on top is scaffolding

1530
01:31:15,160 --> 01:31:18,680
for us to have an easier time. I think that's what we were talking about earlier with like

1531
01:31:19,640 --> 01:31:25,480
Bill AI and vibe coding mess stuff up. And I don't think it will because of that reason,

1532
01:31:25,480 --> 01:31:31,080
because we'll still have to build really good solid foundations. Yeah. And then the

1533
01:31:32,760 --> 01:31:36,840
the process of like using those workshops, using those foundations is always going to be like

1534
01:31:36,840 --> 01:31:43,000
almost general game, like, like build a chair and it's done. It's not perfect. It's good enough.

1535
01:31:43,560 --> 01:31:47,640
Like, this actually brings up an interesting point that I was talking to

1536
01:31:49,240 --> 01:31:55,160
as like, soft engineers, we, there's a weird tendency, it might be from the computer to think

1537
01:31:55,800 --> 01:32:05,320
of stuff we build as global solutions. And like, we don't think of most of the time when we're

1538
01:32:05,320 --> 01:32:10,440
talking like apps we build or tools we make. Yeah, that's the current paradigm. Millions of users.

1539
01:32:10,440 --> 01:32:14,520
Yeah, we never think of like a bespoke solution. Like, if you think the traditional,

1540
01:32:14,520 --> 01:32:18,120
this might be due to the fact that you mentioned like perfect copies and all that, but

1541
01:32:21,560 --> 01:32:25,640
you can't like, you don't have the traditional sense of like a woodworking shop like,

1542
01:32:25,640 --> 01:32:31,480
yeah, make one chair. Yeah, you might make them all look like each other. Yeah, or try to. Yeah,

1543
01:32:31,480 --> 01:32:37,080
I think I think you're right. I think we will have more of that very specialized, highly opinionated

1544
01:32:37,160 --> 01:32:43,320
stuff just because it's easy to build. And if just like 3,000 people use it, that's fine.

1545
01:32:43,320 --> 01:32:46,520
Well, you see it with websites all the time. Yeah. Well, like the original websites that they

1546
01:32:46,520 --> 01:32:51,960
were all bespoke, they were all, they were all just as much as needed to have a website for that

1547
01:32:51,960 --> 01:32:59,640
person. Yeah. And then for the last five years, we've been trying to solve like website building

1548
01:32:59,640 --> 01:33:03,800
with like extremely generic frameworks. Yeah, I would even say like 15 years or so.

1549
01:33:04,760 --> 01:33:11,720
Yeah. Before I was in web development. And it, it, it just keeps getting bigger and bigger and

1550
01:33:11,720 --> 01:33:15,800
bigger and bigger. Yeah. But the, like, yeah, I mean, the ultimate outcome of this is like the

1551
01:33:15,800 --> 01:33:20,200
Facebook page where it's just everything looks the same. Yes. And you set the profile fiction.

1552
01:33:20,200 --> 01:33:24,680
It's just a configure a few things in the massive system. Yeah. But that's one thing I'm looking

1553
01:33:24,680 --> 01:33:31,320
forward to a lot with AI models is the ability to build those like one-off websites that are

1554
01:33:31,320 --> 01:33:35,720
just as big as they need to be. Yeah. And also one-off apps and just one-off whatever in some

1555
01:33:35,720 --> 01:33:40,680
sense. I've been doing that a lot. I've built a lot of kind of little scripts and little things

1556
01:33:40,680 --> 01:33:44,520
that I would have never sat down and build otherwise. I'm looking forward to it for like

1557
01:33:44,520 --> 01:33:49,240
friends and whatnot. Because you always have friends who ask you to build apps, build websites

1558
01:33:49,240 --> 01:33:55,160
and all that. Now you can. You can charge them. Oh, yeah, that's true. But there's some fun

1559
01:33:55,240 --> 01:34:02,440
parties like they don't know yet. Well, it still takes expertise. But just like the fact that I

1560
01:34:02,440 --> 01:34:08,680
could build a, because it used to take a lot of work to make a little web app, a website that was

1561
01:34:08,680 --> 01:34:13,880
just what was needed without using a framework because it's easier with framework speed people

1562
01:34:13,880 --> 01:34:19,080
up. But now you can do a generous single page of HTML looks how you want. Yeah. Ship it done.

1563
01:34:20,040 --> 01:34:26,200
Yeah. Yeah. Yeah. We'll see. We'll see. I mean, I like I like your insight about kind of we're

1564
01:34:26,200 --> 01:34:33,160
already in this situation with, you know, the wizards working on whether it's the Unix kernel

1565
01:34:33,160 --> 01:34:37,720
or Bitcoin core or what have you, you know, like you really need to know what you're doing. You

1566
01:34:37,720 --> 01:34:43,960
need to understand the whole thing. You're not just, you know, like copy-pasting a couple lines of

1567
01:34:43,960 --> 01:34:49,160
JavaScript from Stack Overflow and so on. And that's a very interesting perspective of kind of,

1568
01:34:50,120 --> 01:34:54,520
we're already in this world. It's just like, you've seen it with, um, multiplied now. You've

1569
01:34:54,520 --> 01:34:59,960
seen the JavaScript ecosystem. Oh, yeah. Changes for a while. Like early days, it was everyone had

1570
01:34:59,960 --> 01:35:06,600
to build their own framework. It was, it was, it was hard. And then people who, the engineers who

1571
01:35:06,600 --> 01:35:11,640
really thought through things who, I mean, React came around through it. They, they were really

1572
01:35:11,720 --> 01:35:17,880
intentioned about building a foundational level for a UI framework. Yeah. And then after everything,

1573
01:35:17,880 --> 01:35:22,520
then we just kind of shifted layers. Now we don't write a lot of apps don't write. Yeah.

1574
01:35:22,520 --> 01:35:27,160
Pure JavaScript and more, they write on top of a framework. Yeah. I think, I think it's a safe

1575
01:35:27,160 --> 01:35:33,160
prediction to make that we will have frameworks for LLMs just like that. There'll be languages and

1576
01:35:33,160 --> 01:35:37,400
that's, that's what's already being built now. And that's also what we are trying to build on

1577
01:35:37,400 --> 01:35:43,560
an OSTA side with DB MCP and some of the things that Pablo built with the code snippets and kind

1578
01:35:43,560 --> 01:35:54,520
of having it basically have a, have a very tight integration of your IDE, of the agents it has access

1579
01:35:54,520 --> 01:36:04,440
to, of the MCP tools that you grab from Noster and infusing it with kind of your point of view,

1580
01:36:04,520 --> 01:36:09,400
which means also your web of trust is like, because I'm, you know, if, if Pablo publishes

1581
01:36:09,400 --> 01:36:14,680
and signs something, I'm going to grab it and like, let it run wild on my system. Like, I don't

1582
01:36:14,680 --> 01:36:20,040
care. I don't care. That's enough trust for me. I have a dedicated mic computer. So it's fine.

1583
01:36:20,040 --> 01:36:24,760
If it implodes, you know, I don't trust Pablo completely, but, but you got what I mean, you know,

1584
01:36:24,760 --> 01:36:28,760
and that's what we are trying to build out. And I think we'll have a myriad of those kind of,

1585
01:36:28,760 --> 01:36:32,600
let's call this a framework. It is kind of a framework of how to work with LLMs in some sense.

1586
01:36:32,600 --> 01:36:37,080
And the goal on our side is to build more freedom tech. The goal for us is to, you know,

1587
01:36:37,080 --> 01:36:43,080
one shot, Noster clients, one shot. I mean, one guy basically byte coded a very sophisticated

1588
01:36:43,080 --> 01:36:46,520
Bitcoin wallet, you know, where it's like, okay, it uses Bitcoin script and everything is like,

1589
01:36:46,520 --> 01:36:51,400
I'm never going to use it and trust it, but you can do that too. And we saw, you know,

1590
01:36:51,400 --> 01:36:56,200
Kali and some other people like, you know, by code, cash, your clients and all those kind of things.

1591
01:36:56,200 --> 01:37:00,520
So, so that's what we are aiming at, where it's just like, we want to give agents the tools and

1592
01:37:00,520 --> 01:37:04,920
the understanding, like how to do this right. And then they're never going to make these sets

1593
01:37:04,920 --> 01:37:08,920
of mistakes. Again, they're just going to be able to watch on it. Like, that's how you do,

1594
01:37:08,920 --> 01:37:12,760
that's how you fetch data from relays. That's how you build a feed. That's how you get a profile.

1595
01:37:12,760 --> 01:37:17,000
The cool thing is, there's a chance this time, because we've been optimizing that for a while

1596
01:37:17,000 --> 01:37:21,320
on the existing internet. Like, this is how you make API request. This is how you search Google.

1597
01:37:21,320 --> 01:37:26,440
So, but like, the cool part is with now that we're shifting to cryptographic identities,

1598
01:37:26,440 --> 01:37:29,960
because those systems never work because every 10, every five years, we have to rebuild them.

1599
01:37:30,040 --> 01:37:34,360
Yes. But with cryptographic identities, there's probably not going to be a second,

1600
01:37:34,360 --> 01:37:37,800
like, you're not going to have to rebuild it. It's just going to keep working.

1601
01:37:39,000 --> 01:37:41,080
Like, it. How so? What's the main difference?

1602
01:37:43,320 --> 01:37:46,120
The permission's gone. Like, at the core of it, that's what it is. Like,

1603
01:37:47,160 --> 01:37:50,600
it doesn't go away. The API or the thing doesn't go away. Yes.

1604
01:37:50,600 --> 01:37:53,480
Like, for example, like, if you wanted to search the internet back in the,

1605
01:37:53,480 --> 01:37:58,040
it was, you learned how to, you learned how to make your program interact with Google.

1606
01:37:58,040 --> 01:38:01,560
Alta Vista doesn't exist anymore. It was the search engine before Google.

1607
01:38:01,560 --> 01:38:07,240
Yeah, I don't know that. But the Google API endpoint will go away.

1608
01:38:07,240 --> 01:38:11,720
But my exact point was you built any software in the early days for that earlier search engine.

1609
01:38:12,280 --> 01:38:14,840
Google came along, and now your software doesn't work anymore.

1610
01:38:14,840 --> 01:38:18,760
And that's also why the Unix tools don't go away, because they are local first.

1611
01:38:18,760 --> 01:38:21,080
Yes. Yes. They don't break. And they never go away.

1612
01:38:21,080 --> 01:38:24,360
They don't break because of an external dependency. And what we are trying to do,

1613
01:38:24,440 --> 01:38:29,400
is remove all external dependencies. It doesn't mean that everything has to work on your local

1614
01:38:29,400 --> 01:38:33,800
machine. But you don't want to have external dependencies. That's the idea about moving from

1615
01:38:33,800 --> 01:38:37,480
a server is an external dependency. A relay is not. Yes. Yeah.

1616
01:38:37,480 --> 01:38:41,880
Because that's the difference. That's the thing that the podcasting 2.0 people didn't get, for example.

1617
01:38:41,880 --> 01:38:45,880
It's, that's a crucial difference. A server truly is an external dependency.

1618
01:38:45,880 --> 01:38:51,960
You only have one source of truth. It has power over you. A relay is not an external dependency.

1619
01:38:52,280 --> 01:38:56,120
Because it literally can be an internal thing. I have a relay on my phone. I have a relay at home.

1620
01:38:56,120 --> 01:39:00,520
And so I can run the relay. And I, and it's completely the changer, but I don't have to trust it.

1621
01:39:00,520 --> 01:39:03,480
That's why it's not an external dependency. Commodetized.

1622
01:39:04,600 --> 01:39:07,960
And that's the, that's a crucial difference. And what we want to build out now, like with

1623
01:39:07,960 --> 01:39:15,640
also agents and DBMCP and so on, like a DVM for example, is, is, is not an, it's the same thing.

1624
01:39:15,640 --> 01:39:20,280
It's, a DVM is not an API. It's not an external dependency. It's a different thing. It's a different

1625
01:39:20,280 --> 01:39:27,640
set of fits. It's because there can be multiple in some sense. Like they, they, they don't have to

1626
01:39:27,640 --> 01:39:32,440
be built to work like it. They, they all work the same way. So they're just interchangeable.

1627
01:39:32,440 --> 01:39:37,000
That's, it's just obvious to say, but it's just like a vendor on a large marketplace is not an

1628
01:39:37,000 --> 01:39:41,000
external dependency in that sense. It's like, because you can't go to someone else. A restaurant

1629
01:39:41,000 --> 01:39:44,440
is not an external dependency. You're not going to starve to death if it closes. You can go to

1630
01:39:44,440 --> 01:39:50,200
the next one. That's, yeah. Yeah. It's the same principle. But if all the restaurants are

1631
01:39:50,200 --> 01:39:55,080
operated on subscription models, they're very inconvenient. Because now you have to go on,

1632
01:39:55,080 --> 01:39:59,320
you've, you've lost your credits with the first one. If you're completely relying on the next one,

1633
01:39:59,320 --> 01:40:05,000
yeah. Yeah. That's the difference. And the money lets, the money is what lets you do it.

1634
01:40:05,000 --> 01:40:09,000
The money lets with a hop around. Exactly. Exactly. Again, like if you had, if you had

1635
01:40:09,000 --> 01:40:14,360
subscriptions, if you had like sign up to go to a restaurant, then it becomes a dependency.

1636
01:40:14,360 --> 01:40:19,320
That's why money truly is an asset and credit is a liability in that sense. It's like,

1637
01:40:19,640 --> 01:40:23,960
your credit line, if you, if you have like, you know, like a gift card or whatever,

1638
01:40:23,960 --> 01:40:28,760
it's like a gift card is not money. It's like, it doesn't have the same, the same purpose.

1639
01:40:28,760 --> 01:40:33,000
It's like money proper, proper money. You will always be able to go somewhere else and exchange

1640
01:40:34,040 --> 01:40:38,120
for what you need and so on. Like, and that's, that's, that's, it's such a small difference,

1641
01:40:38,120 --> 01:40:42,040
but it's, it's so crucial, right? It's always a small, it's like, it's so crucial. It's,

1642
01:40:42,760 --> 01:40:46,600
it's, and, and it really helped me to, to think through that once more. Like when,

1643
01:40:46,680 --> 01:40:52,440
when I heard you talk about, like trying to explain why it's not so different and why is it,

1644
01:40:52,440 --> 01:40:55,240
that's also why it's different from an RSS feed, for example, those kind of things,

1645
01:40:55,240 --> 01:40:58,280
because there are a SSP, the data in their RSS feed is not signed. And the podcasting

1646
01:40:58,280 --> 01:41:01,320
to the whole people, they run into this issue all the time. Yeah. Where someone will pirate

1647
01:41:01,320 --> 01:41:05,720
something and, and just exchange the value block. Yeah. Then you send money to the wrong people,

1648
01:41:05,720 --> 01:41:10,040
like that, that's not something that easily happens on an other side. Yeah. Yeah. And with

1649
01:41:10,040 --> 01:41:14,440
that, because it really matters where the data lives, you know, like you have a server and it's

1650
01:41:14,440 --> 01:41:17,880
the source of truth, your feed is the source of food. Yeah. And it's not the data itself that

1651
01:41:17,880 --> 01:41:21,640
it's the source of food. And you can't easily change that after the fact. No. Like you,

1652
01:41:21,640 --> 01:41:27,720
you almost need like a consensus level, like everyone to move to sign stuff now. Like it, it,

1653
01:41:29,560 --> 01:41:33,800
yeah, it's hard to, it's hard to modify that system. Yeah. That's, and that's why it's just

1654
01:41:33,800 --> 01:41:38,760
going to take a long time. It's, it's like, you know, moving people from systems like Twitter to

1655
01:41:39,160 --> 01:41:42,600
from systems like YouTube to an officer based solution and those kinds of things,

1656
01:41:42,600 --> 01:41:47,720
it takes a long, long, long time. It's just like it, it took a long time for, you know, like,

1657
01:41:47,720 --> 01:41:52,760
you know, extend the units tools to, to win over Windows. And then like a couple of decades later,

1658
01:41:52,760 --> 01:41:57,320
Windows was forced to basically integrate Linux into the system. It's like, it's really, it's

1659
01:41:57,320 --> 01:42:04,040
really funny. Well, yeah, it does. Yeah. But kind of systematic. You use latest Windows?

1660
01:42:04,760 --> 01:42:10,200
No, they're trying to go extinct with the way they're designed. You have so much ads. It doesn't

1661
01:42:10,200 --> 01:42:15,720
work. They had ads like in the, in the UI directly. The buttons take like, unless you have like a

1662
01:42:15,720 --> 01:42:20,120
ridiculously fast computer, the buttons, there's like a half, there's like a quarter second delay

1663
01:42:20,120 --> 01:42:26,680
on every button. Okay, it has to load the ads. It's back to like Windows XP. It's crazy. Yeah.

1664
01:42:26,680 --> 01:42:33,080
It queries the ad server first on button down. I mean, that's probably what's happening. Yeah,

1665
01:42:33,080 --> 01:42:37,720
no, I haven't used Windows in a long, long, long time. I still have a very old, I don't even know

1666
01:42:37,720 --> 01:42:43,560
what kind of Windows version it is. But I just unearthed my very old gaming PC just to run models

1667
01:42:43,560 --> 01:42:48,440
on it. It all works well enough. So I'm not, I'm not touching it. I'm not upgrading it and so on.

1668
01:42:48,440 --> 01:42:51,720
Yeah, I only see it because my friends have Windows machines.

1669
01:42:53,000 --> 01:42:56,520
No, I'm not hating. You know, it's like, it's like whatever, whatever works. I did a lot of

1670
01:42:56,520 --> 01:43:01,880
photography back in the day and I kind of had to use a Windows machine for all that because

1671
01:43:02,680 --> 01:43:08,040
all the Linux stuff just wasn't there yet. Now it's way better. Now you have like dark room and

1672
01:43:08,040 --> 01:43:13,400
gimp is also like that was back in the day when gimp was not like one window but 27 different

1673
01:43:13,400 --> 01:43:19,240
floating windows. It was interesting times. It's like getting, it's like what we talked about

1674
01:43:19,240 --> 01:43:25,160
at the beginning almost. It's like getting used to the things so quickly. Yeah. Like you just kind of

1675
01:43:25,720 --> 01:43:29,720
we forget where we came from. Yeah, and then you go back and it's like it's such a strange,

1676
01:43:29,720 --> 01:43:33,560
such a strange experience when like, I mean, you go back to like traditional login photos,

1677
01:43:33,560 --> 01:43:38,120
email stuff. You still have to log with email and it's like, yeah, now you have to check your

1678
01:43:38,120 --> 01:43:41,720
email and there's all this antiquated, right? Yeah, if you're used to it enough stuff,

1679
01:43:41,720 --> 01:43:46,280
it feels very antiquated. I never expected to get used to it so fast. Yeah. Yeah. Yeah. Yeah.

1680
01:43:47,800 --> 01:43:53,480
Yeah, yeah. Same for me. It's like, it's, it's, I mean, for me, like paying my breakfast, lunch,

1681
01:43:53,480 --> 01:43:57,560
and dinner using a lightning wallet is a normal thing to do now and I'm completely used to it.

1682
01:43:57,560 --> 01:44:02,120
And it's like, I don't even think about it, you know, and I sometimes use whatever, you know,

1683
01:44:02,120 --> 01:44:06,360
like if I run out of balance with one thing, then I use my nostril client to pay for my coffee

1684
01:44:06,360 --> 01:44:10,760
and those kind of things, you know, and it's like, I, I, it's, it's a non-event for me,

1685
01:44:10,760 --> 01:44:13,880
which is really interesting. Yeah. It's really interesting. Yeah. And then you go back to the

1686
01:44:13,880 --> 01:44:18,120
system and there's just things you've learned. And also, nostril, nostril working, for example,

1687
01:44:18,120 --> 01:44:22,360
nostril is my like, I only, I only have nostril. I don't have anything else anymore. So it's the

1688
01:44:22,360 --> 01:44:27,000
only thing I basically look into and check and so on, which is also really great because I use

1689
01:44:27,000 --> 01:44:31,400
kind of like, again, like I mentioned this many times in the past, I was really addicted to Twitter.

1690
01:44:33,080 --> 01:44:39,400
And it's a way, like, it's a, it's a more healthy relationship with technology, in some sense.

1691
01:44:39,400 --> 01:44:45,480
Nostril allows me to have a way more healthy relationship with like a feed, with, with, with,

1692
01:44:45,480 --> 01:44:50,280
with my friend network, in some sense. It's, it's, it's, for me, it's way, way, way better. And,

1693
01:44:50,280 --> 01:44:56,280
and just it working and it being a normal thing, it's like, it, it feels like yesterday where

1694
01:44:56,360 --> 01:45:01,320
the only client was like, Brannle, or whatever, he, Jeff call it. And like, everything was broken

1695
01:45:01,320 --> 01:45:05,640
and nothing worked. And like, you had some CLI clients and everything was terrible. And it was

1696
01:45:05,640 --> 01:45:09,320
so weird that no one understood anything. It's like, no one understood what this thing was.

1697
01:45:09,320 --> 01:45:14,520
That was only like, oh, yeah, three years ago or so. That's, that's, that's something like that.

1698
01:45:14,520 --> 01:45:20,040
And, and now it's, it's, it looks very snappy. It's very fast. We're like, optimizing things

1699
01:45:20,040 --> 01:45:25,320
left and right. I mean, again, like, I'm, I'm using it as one of my main wallets for everything,

1700
01:45:25,320 --> 01:45:29,240
which is weird. Like I, I owe Pablo money or something or whatever. It's like, for a coffee,

1701
01:45:29,240 --> 01:45:33,320
it's like, I'm just going to set the people. And it's just completely seamless.

1702
01:45:33,320 --> 01:45:37,800
Like it's, the, the cool part is we haven't even gotten to the ideas of like,

1703
01:45:38,600 --> 01:45:42,120
really seamless yet. Because a lot of the lightning apps, a lot of the other stuff take like multiple

1704
01:45:42,120 --> 01:45:45,960
clicks. And you got the right button somewhere in the corner to get it to do that one thing.

1705
01:45:45,960 --> 01:45:52,360
Yeah. And it's still, oh, that would actually be a very nice and natural language use case.

1706
01:45:52,440 --> 01:45:56,600
Like sent Pablo five bucks and it was a very good look up and it just zaps.

1707
01:45:56,600 --> 01:45:58,040
Well, that's what like, that's what.

1708
01:45:58,760 --> 01:46:02,920
Yeah, that's really seamless. Like the, the, the sorry to interrupt, but in terms of the,

1709
01:46:02,920 --> 01:46:08,360
the vibing stuff, for me, the key unlock was actually speaking to the computer. And, and it's so,

1710
01:46:08,360 --> 01:46:12,120
because I use it now for my notes and my to-dos and like, and the things and so on.

1711
01:46:12,120 --> 01:46:17,080
And I just speak into my phone and that's it. That's all I do. Everything else is taking care

1712
01:46:17,080 --> 01:46:21,240
about the computer. That's what I'm so excited about the dumb models because they, I don't need

1713
01:46:21,240 --> 01:46:25,640
the model to think for me for the first use case. I need to build a powerful system. That's my

1714
01:46:25,640 --> 01:46:32,920
home assistant. And I need to just run natural language commands against it. And like, there's

1715
01:46:32,920 --> 01:46:36,280
so much to do with that. One of them could be, ah, yeah, it sent Pablo five bucks for the coffee

1716
01:46:36,280 --> 01:46:41,320
asset. DM somebody something like, the original use case of Siri was that already works. That's

1717
01:46:41,320 --> 01:46:44,520
the amazing thing. People don't know that this already works with the stuff that you built.

1718
01:46:44,520 --> 01:46:50,040
Talk about this, maybe a little bit more. What's it called? The bakery? I just, for lack of a

1719
01:46:50,040 --> 01:46:54,200
better name. It was going to be a notional thing. That's why it's called bakery. I'm thinking now

1720
01:46:54,200 --> 01:46:59,240
it's, it's going to be its own thing. But it's, it's, it's meant to be kind of,

1721
01:47:00,840 --> 01:47:05,320
the, the vague idea right now is kind of that, like a system that you run in one place,

1722
01:47:06,040 --> 01:47:12,920
that you then prompt with AI to do things, a powerful tool for a powerful,

1723
01:47:12,920 --> 01:47:18,600
no-ster-focused tool for some sort of AI toolbox. Yeah, you sent me a DM yesterday just by saying

1724
01:47:19,560 --> 01:47:23,240
sent G-G-I-D-M and say G-M or something like that. Well, that, that's the cool thing is like,

1725
01:47:23,240 --> 01:47:29,480
anyway, the MCP and all that is like, you plug, again, back to like the, the foundation thing

1726
01:47:29,480 --> 01:47:34,120
and the fuzzy code. The MCP is a really dumb developer. Yeah. Clicking around and copying stack

1727
01:47:34,120 --> 01:47:39,160
overflow. And I'm just providing a solid foundation of tools that can use. This is how you do it. This

1728
01:47:39,160 --> 01:47:42,200
is how you're saying that the M, this is how it works. Yeah, like, this is what you do. This is

1729
01:47:42,200 --> 01:47:46,360
what you need. This is how you look up a person. This is what I trust about. This is how you ask

1730
01:47:46,440 --> 01:47:50,920
me to sign something and then. Yeah, that's so cool because it's like, you can do that. And again,

1731
01:47:50,920 --> 01:47:56,680
like this already works. Trust it completely. And, and, and like you amber on your phone will pop up

1732
01:47:56,680 --> 01:48:02,680
and you will just sign, sign it here and there. Yeah. It's, it's so cool. And just, just, just, just to

1733
01:48:02,680 --> 01:48:07,160
reiterate the point, like the, the, what I think is so cool because once you have the, the

1734
01:48:07,160 --> 01:48:11,640
natural language prompt working, the switch to audio is like a non-event. But no, you can't,

1735
01:48:11,720 --> 01:48:15,800
this was my point of view. You can't switch to audio without that insight into the system.

1736
01:48:15,800 --> 01:48:21,800
Yeah. Because if, if I plugged open AI into like telegram to DM my friends, right?

1737
01:48:23,320 --> 01:48:28,520
I like, I can't, there's no way in a, I can know for 100% that I'm going to be able to verify

1738
01:48:28,520 --> 01:48:33,560
that message before it sends. Yeah. But the fact that we have the keys, the fact that we have that

1739
01:48:33,560 --> 01:48:38,920
last. Yeah, you're completely controlled. I can completely control. I can switch out the model

1740
01:48:39,000 --> 01:48:42,920
for a different one every single day for a different flavor of message. And I won't worry at all

1741
01:48:42,920 --> 01:48:47,320
because I can see the message it wants to sign. And if I don't like it, I just don't let it do it.

1742
01:48:47,320 --> 01:48:50,760
Yeah. And, and it's complete trust. And, and just to paint the picture, like, because that's,

1743
01:48:50,760 --> 01:48:53,720
that's my current workflow. It's like, I'll be out and about and I'll be, you know,

1744
01:48:53,720 --> 01:48:58,040
walking from one place to the other or whatever. And I, I remember like, I, I either have an

1745
01:48:58,040 --> 01:49:01,560
inside or remember something I need to do. So I'll, I'll pop out my phone and I'll speak into the

1746
01:49:01,560 --> 01:49:06,440
phone for like 30 seconds. Oh, I just remember that I need to, you know, like, I don't know.

1747
01:49:06,440 --> 01:49:11,960
I, I need to DM hazard that we need to go on a walk or whatever. And for the exact time. And,

1748
01:49:11,960 --> 01:49:16,440
and with that stuff, it's like, it's, it's exactly the same tool chain and everything. And it can

1749
01:49:16,440 --> 01:49:20,520
actually send at the end. Yeah. And it can actually send them money. Again, like, I, oh, I just

1750
01:49:20,520 --> 01:49:24,200
remember that old Pablo with five bucks for the coffee yesterday or whatever. And it can actually

1751
01:49:24,200 --> 01:49:28,920
literally just pick that up. And on, on your, your phone will pop up with amber and it's like,

1752
01:49:28,920 --> 01:49:34,200
okay, you want to sign this, which is the DM. We want to sign this, which sends to five bucks.

1753
01:49:34,200 --> 01:49:37,960
And that's it. And you sign off on it. And that's it. And that's like, the, the truly is a very

1754
01:49:37,960 --> 01:49:42,760
powerful virtual assistant in some sense. It really is as cool as that. The next, I want to

1755
01:49:42,760 --> 01:49:48,280
experiment with this with a few, a few more on those traps. But we're, this is going to make it

1756
01:49:48,280 --> 01:49:54,200
very obvious that we're lacking in the most signing software category. Yeah, because right now,

1757
01:49:54,200 --> 01:49:58,680
like Amber's Android, Android only. But more, no more, more than that, because

1758
01:49:58,680 --> 01:50:05,480
Amber is awesome, but it signs everything. So it means it's too broad to show me exactly what

1759
01:50:05,480 --> 01:50:10,920
it's doing. Even if it like shows a kind one that it wants to sign, I still can't quite see what

1760
01:50:10,920 --> 01:50:17,000
it's doing. But if I could, what if I had five amber apps on my phone? And one was dedicated

1761
01:50:17,000 --> 01:50:22,840
only signing nip 60 wallet stuff. So all it was focused on was signing the events for me and for

1762
01:50:22,840 --> 01:50:27,000
that. And it would show me the details about everything. Like, you want this transaction,

1763
01:50:27,080 --> 01:50:31,320
it's probably going here, the money's coming out of here. And then another one, it's just social.

1764
01:50:32,040 --> 01:50:37,640
Yeah. And having that the smarter, like, we're going to need smarter sign, like,

1765
01:50:37,640 --> 01:50:40,680
it's going to be interesting. I'll be trying to do something like that, right? They're

1766
01:50:40,680 --> 01:50:45,480
extension with like, let's be reasonable. I fully trust it. I'm paranoid, ask me every time those

1767
01:50:45,480 --> 01:50:50,520
kind of levels. And, and, and the other thing is like, it, I think it just has to be like a

1768
01:50:50,520 --> 01:50:55,400
visual indication that, for example, you're sending money now. So you actually want to

1769
01:50:55,400 --> 01:50:59,240
probably sign up on that and double check it. But I think, I think we're on our way of figuring

1770
01:50:59,240 --> 01:51:02,680
all that out. Like, we have the budgets and those kinds of things, because like, if it's like 20

1771
01:51:02,680 --> 01:51:06,120
cents, like, you don't need to. But it's more than like, the thing is like automation doesn't

1772
01:51:06,120 --> 01:51:11,800
solve it. Yeah. Until you, you need to solve the first problem you need to solve is understanding,

1773
01:51:11,800 --> 01:51:15,480
and then you can automate it. Yeah. So you need, you need, you need to wait for the user to see what

1774
01:51:15,480 --> 01:51:20,120
they're signing. Understand it within half a second. Yeah. Look at the screen and be like,

1775
01:51:20,120 --> 01:51:26,520
this is a signature for a long form post. Yeah. And then like, okay, I, I, I, I read, I, you

1776
01:51:26,520 --> 01:51:30,040
browse through it. Okay. It looks good. Yeah. And then after that, you can, you can say auto

1777
01:51:30,040 --> 01:51:33,480
approval. Yeah. And also, exactly. Exactly. And that's, it relates to what you mentioned about

1778
01:51:33,480 --> 01:51:37,480
the trust, because if you build trust, if you actually trust the thing, I'm gonna auto approve.

1779
01:51:37,480 --> 01:51:41,480
And then you just needs to be transparent that you need to, like, I'm gonna, I want to look back,

1780
01:51:41,480 --> 01:51:45,400
and I want to see what the things spent in the last week, or whatever, and so on, and it's like,

1781
01:51:45,400 --> 01:51:49,160
and that's, I think that's how it's going to work. That's how it's going to evolve. But I really,

1782
01:51:49,160 --> 01:51:52,920
I really like what kind of was cooked up in the last couple of weeks, because it, it all

1783
01:51:52,920 --> 01:51:58,040
evolved so quickly. And now, and that's where we at now. We, we actually have human language

1784
01:51:58,040 --> 01:52:02,280
interfaces into Noster, into wallets, it's like, we can sap, we can do things, we can do all kinds

1785
01:52:02,280 --> 01:52:07,800
of things. We can, we can, a month. Yeah. Where? Oh, it's building the month. Open AI, and they're

1786
01:52:07,800 --> 01:52:12,360
still, they're still trying to get AI. You can't, unless you like sign up for like five subscriptions,

1787
01:52:12,360 --> 01:52:16,200
you're not going to get, you're not going to get your open AI model to send an email for you.

1788
01:52:16,200 --> 01:52:22,360
Now, we just have to get better at explaining it, showing it to other people. I think even people

1789
01:52:22,360 --> 01:52:26,680
inside our little bubble will have trouble wrapping their head around it, and what it really is,

1790
01:52:26,680 --> 01:52:32,120
and what it really means. But I'm, I'm really, really kind of optimistic, because

1791
01:52:32,120 --> 01:52:38,440
shipping examples, we already have the marketplaces. And so, so, so it's very much like, in the beginning,

1792
01:52:38,440 --> 01:52:42,920
when Noster was like, like, Fijajab basically launched Noster, no one understood it, but it

1793
01:52:43,000 --> 01:52:47,800
was already life. And it's the same thing now. What we have is already life. And so, you, you,

1794
01:52:47,800 --> 01:52:55,240
you, if someone publishes a code snippet, if someone publishes, you know, DBMCP thing, whatever,

1795
01:52:55,240 --> 01:52:59,960
it will pop up immediately and other people can pick it up. And you can immediately, instantly

1796
01:52:59,960 --> 01:53:04,360
interact with all of it. And also, on, on, you know, on the, on the bakery side of what you

1797
01:53:04,360 --> 01:53:09,560
build and so on, it's like, you, you, you literally can, you know, use one of your sock

1798
01:53:10,120 --> 01:53:14,360
or whatever or punch in your real life. And you can just let it do stuff. And that's so magical,

1799
01:53:14,360 --> 01:53:19,720
right? It's just like, you know, post GM, and it will do it. Yeah, yeah. And you can literally

1800
01:53:19,720 --> 01:53:24,040
talk into your computer and your phone and it's like, post GM, and it will do the things. It's

1801
01:53:24,040 --> 01:53:28,840
really cool. It's going to be, it's going to be incredible for that. Like I said, the,

1802
01:53:30,120 --> 01:53:35,000
the prompting the commands is that it's, I don't know, like, there's so much we've been,

1803
01:53:35,080 --> 01:53:39,400
like I said, it was the promise of like, when I grew up, it was like, promise like,

1804
01:53:39,400 --> 01:53:43,400
Siri came out. Yeah. And that was the promise. You're going to be able to do anything. You're

1805
01:53:43,400 --> 01:53:46,680
not going to be like, it's not, they had a longer promise. You're like, you'll talk to your phone,

1806
01:53:46,680 --> 01:53:50,120
but I don't know. It was like, for me, it was a star trek computer. That's like generational

1807
01:53:50,120 --> 01:53:57,720
differences. But it just hasn't never, it never came. Yeah. And now like, it's something like,

1808
01:53:57,720 --> 01:54:01,560
yeah, now it's here. Systems are open. And now it's, now it's here. It's actually, you can

1809
01:54:01,560 --> 01:54:07,400
actually build it. You can do it. It runs without, it works, which is crazy.

1810
01:54:07,400 --> 01:54:10,280
Yeah, it's really cool. It's really cool. It actually works now.

1811
01:54:10,280 --> 01:54:15,400
All right. Let's have, let's grab some coffee and see if we can work on explaining it to other

1812
01:54:15,400 --> 01:54:20,520
people a little bit more. Maybe write a, buy the blog posts into existence. Yeah. You need to

1813
01:54:20,520 --> 01:54:24,200
write, you need to write one of the blog posts that you talked about in the, you know, JavaScript

1814
01:54:24,200 --> 01:54:29,800
UI framework context is like, you need to sit down, write about this, explain it to other people,

1815
01:54:30,520 --> 01:54:34,280
explain it to our podcasting 2.0 friends, explain it to some of the other

1816
01:54:34,280 --> 01:54:39,080
JavaScript front end developers and some, some tech people that should be interested in that stuff.

1817
01:54:39,080 --> 01:54:43,000
We'll write more for that. Same reason. Just like use obsidian more, get right more notes down,

1818
01:54:43,000 --> 01:54:48,760
write more blog posts because they, yeah. Yeah. We'll get it. It's good to get the information out.

1819
01:54:48,760 --> 01:54:53,800
We'll get there. All right. What do you say? Coffee, orange juice, something like that. I'm cooked.

1820
01:54:53,800 --> 01:54:57,640
The sun's up now. It's going to be hot today. All right. Let's hope that all this worked.

1821
01:54:57,720 --> 01:55:02,520
It's always, with me, it's always, you know, a gamble whether the recording actually worked or not.

1822
01:55:03,720 --> 01:55:04,920
That looks good.