1
00:00:00,000 --> 00:00:05,720
 Let's see if that looks good. That looks good. Seems to be working this thing.

2
00:00:05,720 --> 00:00:14,480
 All right, we're off to the races. Great, fantastic. So again, I'm fine, I'm fine.

3
00:00:14,480 --> 00:00:18,880
 I'm on lots of... Oh yeah, let's go this way first a little bit, because I

4
00:00:18,880 --> 00:00:22,840
 need to refill my water and then we'll just turn around. But like I

5
00:00:22,840 --> 00:00:27,160
 mentioned, I'm on lots of pain medication, so I'll let you do most of the talking.

6
00:00:27,160 --> 00:00:34,320
 Oh, thanks man. And I'll just chill and listen. So yeah, tell me all about it.

7
00:00:34,320 --> 00:00:40,240
 Like, I was quite surprised to see you sign up for Sovereign Engineering and

8
00:00:40,240 --> 00:00:43,160
 actually be here. I mean, I obviously know about you and your work. You've been

9
00:00:43,160 --> 00:00:47,200
 working on ASMap. So maybe, I don't know, you pick and choose. What would be a

10
00:00:47,200 --> 00:00:52,760
 good entry point to start of kind of... Well, why were you surprised? I don't know.

11
00:00:52,760 --> 00:00:55,520
 Lots of Bitcoin people don't care too much about Nostr and the stuff we're

12
00:00:55,520 --> 00:01:00,320
 doing here. And I consider you like very deep into Bitcoin technicalities and

13
00:01:00,320 --> 00:01:05,320
 those kind of things. Oh, I don't know about that. I don't know. How did you end up working on

14
00:01:05,320 --> 00:01:11,060
 ASMap in the first place? Like, how did that go? Yeah, so I was... I mean, I've been

15
00:01:11,060 --> 00:01:13,920
 interested in Bitcoin for a long time and I was always trying to find ways to

16
00:01:13,920 --> 00:01:17,240
 contribute, but I never really... You know, like, I'm a dev, but I've never done C++,

17
00:01:17,240 --> 00:01:21,760
 and I don't really want to either. And I never really found stuff I could

18
00:01:21,760 --> 00:01:28,440
 contribute to in Core that I thought was very meaningful. But I went to a BTC++

19
00:01:28,440 --> 00:01:34,480
 in Berlin in 2023. That was Nix edition, right? And so I had happened to do a fair

20
00:01:34,480 --> 00:01:38,960
 amount. I went down the Nix rabbit hole. You were a Nix guy before that. I was a

21
00:01:38,960 --> 00:01:43,480
 Nix guy. Yeah, I was a Nix guy. And somehow during the COVID lockdowns, I ended up, you

22
00:01:43,480 --> 00:01:47,720
 know, just in my house and learned a ton about Nix. I went way too far into it. Like,

23
00:01:47,720 --> 00:01:53,460
 approached madness and stepped back from the brink, you know? And so, yeah, Lisa did a...

24
00:01:53,460 --> 00:01:58,120
 chose to do a BTC++ on Nix, which I thought was super cool and super random.

25
00:01:58,120 --> 00:02:05,240
 And so I... she invited me to do a talk. So I did a talk in Berlin. And there I met

26
00:02:05,240 --> 00:02:09,840
 Fabian Yar, who was, you know, a longtime contributor in Bitcoin. He's done a ton of

27
00:02:09,840 --> 00:02:15,120
 work on plenty of different things. And we were just chatting, you know, and he was like,

28
00:02:15,120 --> 00:02:18,240
 "Yeah, you know, I have this problem with this project I'm working on." And I'm trying

29
00:02:18,240 --> 00:02:22,000
 to make sure, like... because a lot of people have to, like, run this software together,

30
00:02:22,000 --> 00:02:26,160
 but I need to make sure their dependencies are, like, the same and that they're reproducible,

31
00:02:26,160 --> 00:02:29,440
 so we can be sure about reproducibility. Yeah, you can see that in the GitHub issues,

32
00:02:29,440 --> 00:02:33,360
 right? Like, everyone is running the thing and reporting the result of, like, the runs

33
00:02:33,360 --> 00:02:35,480
 now. Right, right, right. So we do this collaborative

34
00:02:35,480 --> 00:02:41,080
 run thing with ASMap that, you know, we can get into the ASMap process and whatever, where

35
00:02:41,080 --> 00:02:45,080
 we all kind of run ASMap together and see what the output is. And so to do that correctly,

36
00:02:45,080 --> 00:02:49,640
 you have to at least have, like, the same software and the same dependencies and make

37
00:02:49,640 --> 00:02:53,200
 sure we're on the same version, et cetera. And so Nix was a good solution to that, right?

38
00:02:53,200 --> 00:02:57,400
 So after that conversation, I kind of, like, poked at ASMap, tried to understand what it

39
00:02:57,400 --> 00:03:07,720
 was and started, yeah, and started contributing here and there to ASMap. I was kind of, like,

40
00:03:07,720 --> 00:03:11,000
 moving countries and, like, in between 14 different things over the next handful of months,

41
00:03:11,000 --> 00:03:15,840
 but started contributing to the project a little more. And there was a lot to do. You

42
00:03:15,840 --> 00:03:19,800
 know, it was one of those things where it was kind of a big -- so I should maybe take

43
00:03:19,800 --> 00:03:25,000
 a step back here. Like, there's, like, the thing I'm talking about is this program called

44
00:03:25,000 --> 00:03:30,600
 Cartograph which creates ASMaps, right? >> For those people who don't know what ASMaps

45
00:03:30,600 --> 00:03:33,120
 are. >> Yeah, so, I mean, I can give the whole --

46
00:03:33,120 --> 00:03:38,720
 we can get into what ASMap is. It's a cool story and it's been part of Core for a while.

47
00:03:38,720 --> 00:03:46,120
 So yeah, so ASMap is a feature of Bitcoin Core. And it's been included as a feature

48
00:03:46,120 --> 00:03:53,520
 since about Core 20. So it's been some time. And it was started by Gleb Nalmenko. And then

49
00:03:53,520 --> 00:03:57,240
 a lot of people contributed to it. I think SIPA, Virtu, and a handful of other people

50
00:03:57,240 --> 00:04:02,840
 over the years. And where it came from was to realize kind of an observation that Gleb

51
00:04:02,840 --> 00:04:09,200
 had at the time, which was the way, like, a Bitcoin Core node peers with the network

52
00:04:09,200 --> 00:04:14,680
 with other peers, you know, the node, like, keeps track of all its other peers and other

53
00:04:14,680 --> 00:04:18,040
 addresses, right? >> By addresses, you mean IP addresses.

54
00:04:18,040 --> 00:04:22,920
 >> IP addresses, right? Or Tor addresses or other transports. But let's stick to the clear

55
00:04:22,920 --> 00:04:27,720
 net example, right? Because that's where ASMap is relevant. It would be unfortunate if you

56
00:04:27,720 --> 00:04:32,360
 started your Bitcoin Core node and you connected to, like, mostly peers that were in the same

57
00:04:32,360 --> 00:04:38,280
 AWS data center, the same Hetner data center. So, you know, we want this to be decentralized.

58
00:04:38,280 --> 00:04:41,080
 And so, like... >> In part, this would be problematic because

59
00:04:41,080 --> 00:04:43,600
 that would one easy way to sibyl attack someone, right?

60
00:04:43,600 --> 00:04:47,720
 >> Right. So yeah, the problem we're trying to avoid is that your node gets sibyl attacks

61
00:04:47,720 --> 00:04:51,720
 or that you get eclipse from the network, we sometimes say, right? So that means that,

62
00:04:51,720 --> 00:04:55,680
 you know, a malicious attacker manages to control all your peers and feeds you this,

63
00:04:55,680 --> 00:05:00,040
 like, alternate view of the network. And you're no longer on the Bitcoin network.

64
00:05:00,040 --> 00:05:01,010
 What do you think about the network?

65
00:05:01,010 --> 00:05:03,410
 You're on kind of a fake mirror world that your attacker has created for you.

66
00:05:03,410 --> 00:05:07,690
 Yeah, everyone you attack to is a malicious actor and tries to fool you.

67
00:05:07,690 --> 00:05:13,410
 So you have no way of actually knowing about that if every single person you're connected

68
00:05:13,410 --> 00:05:15,010
 to is part of the bad bunch.

69
00:05:15,010 --> 00:05:16,010
 Right, right, right, right.

70
00:05:16,010 --> 00:05:17,010
 That's the issue.

71
00:05:17,010 --> 00:05:21,530
 Yeah, like we want, you know, the peer to peer network should be the wild west.

72
00:05:21,530 --> 00:05:22,730
 Hello!

73
00:05:22,730 --> 00:05:24,090
 You'll have a guest appearance.

74
00:05:24,090 --> 00:05:25,090
 Big hug.

75
00:05:25,090 --> 00:05:26,090
 Say hi.

76
00:05:26,090 --> 00:05:30,730
 Say hi to a friend, hello.

77
00:05:30,730 --> 00:05:31,730
 Good, good, perfect.

78
00:05:31,730 --> 00:05:32,730
 I sleep almost all day.

79
00:05:32,730 --> 00:05:34,530
 I'm super good now.

80
00:05:34,530 --> 00:05:36,090
 Guest appearance to the show by Sherry.

81
00:05:36,090 --> 00:05:40,330
 Yeah, we are recording a recorded dialogue right now.

82
00:05:40,330 --> 00:05:45,090
 So if I would have a third microphone, I'd clip it on to you and you could join.

83
00:05:45,090 --> 00:05:48,010
 But unfortunately, I don't have a third, so we'll have to.

84
00:05:48,010 --> 00:05:49,010
 Oh, okay.

85
00:05:49,010 --> 00:05:50,010
 I didn't know this.

86
00:05:50,010 --> 00:05:51,010
 Okay, wait, wait.

87
00:05:51,010 --> 00:05:52,010
 Yeah, we look like regular people.

88
00:05:52,010 --> 00:05:53,010
 Are you guys going to cut this part?

89
00:05:53,010 --> 00:05:54,010
 No, no.

90
00:05:54,010 --> 00:05:55,010
 There's no cutting.

91
00:05:55,010 --> 00:05:56,010
 There's no edits.

92
00:05:56,010 --> 00:05:57,010
 There's no edits.

93
00:05:57,010 --> 00:05:58,010
 Just roll with it.

94
00:05:58,010 --> 00:05:59,010
 Just roll with it.

95
00:05:59,010 --> 00:06:03,410
 I'm going to listen.

96
00:06:03,410 --> 00:06:05,810
 So we'll continue and we'll just, where do we need to go?

97
00:06:05,810 --> 00:06:08,810
 I might, I might, I might, I can't help you.

98
00:06:08,810 --> 00:06:11,010
 I will find my flatmate.

99
00:06:11,010 --> 00:06:12,010
 Okay.

100
00:06:12,010 --> 00:06:13,010
 Yoro.

101
00:06:13,010 --> 00:06:15,290
 I still have difficulty pronouncing his name.

102
00:06:15,290 --> 00:06:16,290
 Okay.

103
00:06:16,290 --> 00:06:17,290
 Yeah.

104
00:06:17,290 --> 00:06:18,290
 Okay.

105
00:06:18,290 --> 00:06:19,290
 So you went down here.

106
00:06:19,290 --> 00:06:20,290
 You might, no, that's right.

107
00:06:20,290 --> 00:06:21,290
 Yeah.

108
00:06:21,290 --> 00:06:22,290
 How close do you continue going down?

109
00:06:22,290 --> 00:06:23,290
 Yeah, yeah.

110
00:06:23,290 --> 00:06:24,290
 Just go that way.

111
00:06:24,290 --> 00:06:27,370
 No, you go ahead and we stay in the shade a little bit.

112
00:06:27,370 --> 00:06:28,370
 Okay.

113
00:06:28,370 --> 00:06:31,050
 And then, and then we see, because if you get lost again, we can help you.

114
00:06:31,050 --> 00:06:32,050
 Definitely.

115
00:06:32,050 --> 00:06:33,050
 Okay.

116
00:06:33,050 --> 00:06:36,490
 You, you go ahead and try, try, try to find your way and we'll, we'll, we'll catch up

117
00:06:36,490 --> 00:06:37,490
 with you.

118
00:06:37,490 --> 00:06:38,490
 And don't cut it.

119
00:06:38,490 --> 00:06:39,490
 All right.

120
00:06:39,490 --> 00:06:40,490
 No, I won't worry.

121
00:06:40,490 --> 00:06:41,490
 Bye bye.

122
00:06:41,490 --> 00:06:42,490
 No solutions podcasts.

123
00:06:42,490 --> 00:06:43,490
 Fantastic.

124
00:06:43,490 --> 00:06:44,490
 Yeah.

125
00:06:44,490 --> 00:06:50,250
 We were, we were, we were talking about her today already because you know, like the,

126
00:06:50,250 --> 00:06:54,970
 the female participant rate in the sovereign engineering cohorts is steadily increasing.

127
00:06:54,970 --> 00:06:55,970
 Is it?

128
00:06:55,970 --> 00:06:56,970
 Yeah.

129
00:06:56,970 --> 00:06:57,970
 Cool.

130
00:06:57,970 --> 00:06:58,970
 Yeah.

131
00:06:58,970 --> 00:06:59,970
 We went from zero to one big moment.

132
00:06:59,970 --> 00:07:00,970
 There you go.

133
00:07:00,970 --> 00:07:01,970
 Breaking out anytime soon.

134
00:07:01,970 --> 00:07:04,610
 So we are on the beginning of an exponential curve.

135
00:07:04,610 --> 00:07:06,450
 Oh, that's how you got it like that.

136
00:07:06,450 --> 00:07:07,450
 Yeah.

137
00:07:07,450 --> 00:07:08,450
 Yeah.

138
00:07:08,450 --> 00:07:09,450
 Yeah.

139
00:07:09,450 --> 00:07:10,450
 Yeah.

140
00:07:10,450 --> 00:07:11,450
 Let's see.

141
00:07:11,450 --> 00:07:12,450
 Let's see how it goes.

142
00:07:12,450 --> 00:07:13,450
 Easy optimistic math, my friend.

143
00:07:13,450 --> 00:07:14,450
 Very optimistic math.

144
00:07:14,450 --> 00:07:15,450
 Very nice.

145
00:07:15,450 --> 00:07:16,450
 Yeah.

146
00:07:16,450 --> 00:07:19,610
 So we were, we were, um, uh, like the goal of AS map is to, um, protect users from eclipse

147
00:07:19,610 --> 00:07:20,610
 attacks.

148
00:07:20,610 --> 00:07:21,610
 Right.

149
00:07:21,610 --> 00:07:22,610
 Is that fair kind of?

150
00:07:22,610 --> 00:07:23,610
 Yeah, exactly.

151
00:07:23,610 --> 00:07:24,610
 Exactly.

152
00:07:24,610 --> 00:07:25,610
 That's the goal.

153
00:07:25,610 --> 00:07:26,610
 Right.

154
00:07:26,610 --> 00:07:28,290
 Um, so, you know, obviously like the way core is set up is, is meant to help with this.

155
00:07:28,290 --> 00:07:29,290
 Right.

156
00:07:29,290 --> 00:07:32,410
 You know, the way we manage, we learn about a lot of peers and we connect, we learn about

157
00:07:32,410 --> 00:07:33,530
 a lot of peers through other peers.

158
00:07:33,530 --> 00:07:34,530
 Yeah.

159
00:07:34,530 --> 00:07:36,490
 So we have a very large number of addresses on our core nodes.

160
00:07:36,490 --> 00:07:39,170
 I mean, usually, I don't know, between 40, 50,000, something like that.

161
00:07:39,170 --> 00:07:42,810
 I mean, you're actively connected to a much smaller subset at a time and getting information

162
00:07:42,810 --> 00:07:43,810
 from them.

163
00:07:43,810 --> 00:07:47,650
 But it's a hard, just to, just to make it clear how hard that problem is because we still

164
00:07:47,650 --> 00:07:52,850
 have, correct me if I, if I'm wrong, but they're still, you need, you need to see the initial

165
00:07:52,850 --> 00:07:53,850
 connection somehow.

166
00:07:53,850 --> 00:07:54,850
 Right.

167
00:07:54,850 --> 00:07:59,450
 So you need to start from a trusted seat in some sense, and then you branch out from there.

168
00:07:59,450 --> 00:08:03,610
 There is no, there is no other sensible way of, of doing it in a decentralized network

169
00:08:03,610 --> 00:08:05,450
 without an authority, basically.

170
00:08:05,450 --> 00:08:07,330
 So the, the, yeah.

171
00:08:07,330 --> 00:08:10,330
 Or having a very long bootstrap time where you crawl the network without interacting

172
00:08:10,330 --> 00:08:13,290
 with it beyond just like gathering information about peers or something.

173
00:08:13,290 --> 00:08:14,290
 Right.

174
00:08:14,290 --> 00:08:15,290
 Yeah.

175
00:08:15,290 --> 00:08:16,290
 You can do some stuff, but it'll be quite annoying.

176
00:08:16,290 --> 00:08:17,290
 Yeah.

177
00:08:17,290 --> 00:08:21,690
 So the bootstrap seed thing in Core is, I mean, I don't know, it's, yeah, it's the best solution

178
00:08:21,690 --> 00:08:22,690
 we came up with.

179
00:08:22,690 --> 00:08:23,690
 I wouldn't say it's optimal, right?

180
00:08:23,690 --> 00:08:26,290
 We still trust this set of seeds to start our peer bootstrapping.

181
00:08:26,290 --> 00:08:27,290
 Right.

182
00:08:27,290 --> 00:08:28,290
 Yeah.

183
00:08:28,290 --> 00:08:29,290
 So it's a hard problem for sure.

184
00:08:29,290 --> 00:08:32,730
 And this is not, this is not to do with like, you know, Byzantine generals problem or whatever,

185
00:08:32,730 --> 00:08:33,730
 right?

186
00:08:33,730 --> 00:08:35,930
 We're just talking about like connecting to peers and not getting eclipsed.

187
00:08:35,930 --> 00:08:37,170
 That's either two separate kinds of.

188
00:08:37,170 --> 00:08:38,170
 Yeah.

189
00:08:38,170 --> 00:08:39,170
 Yeah.

190
00:08:39,170 --> 00:08:40,170
 It has nothing to do with consensus in that sense.

191
00:08:40,170 --> 00:08:41,170
 Right.

192
00:08:41,170 --> 00:08:42,170
 Nothing like that.

193
00:08:42,170 --> 00:08:43,170
 Nothing to do with that.

194
00:08:43,170 --> 00:08:44,170
 Networking peer to peer.

195
00:08:44,170 --> 00:08:45,170
 Yeah.

196
00:08:45,170 --> 00:08:48,930
 Networking peer is this thing we call your, it's called Aderman in the code.

197
00:08:48,930 --> 00:08:51,450
 It's just like your address manager that stores information about peers.

198
00:08:51,450 --> 00:08:57,450
 And I think it does where, where it kind of buckets your peers into, yeah, little buckets.

199
00:08:57,450 --> 00:08:58,450
 Right.

200
00:08:58,450 --> 00:09:03,570
 And the way it does it is that it takes your, the first like two octets of your IPv4 address,

201
00:09:03,570 --> 00:09:04,570
 let's say.

202
00:09:04,570 --> 00:09:08,370
 So, you know, the two numbers after the dot, the first two numbers.

203
00:09:08,370 --> 00:09:09,370
 And that's a bucket.

204
00:09:09,370 --> 00:09:14,370
 So we say anybody in this network, given your IP address, you get put in the same bucket.

205
00:09:14,370 --> 00:09:19,570
 We'll pretend if you're, you know, 12.2 and everyone with addresses that start like that,

206
00:09:19,570 --> 00:09:20,570
 you put in the same bucket.

207
00:09:20,570 --> 00:09:21,570
 Right.

208
00:09:21,570 --> 00:09:22,570
 Yeah.

209
00:09:22,570 --> 00:09:25,890
 And the reason for that is because if you would be in the same AWS data center without doing

210
00:09:25,890 --> 00:09:30,810
 any additional fancy IP address masking or anything like that, you would have a sign,

211
00:09:30,810 --> 00:09:33,050
 you would automatically be end up in the same bucket.

212
00:09:33,050 --> 00:09:34,050
 Right.

213
00:09:34,050 --> 00:09:35,050
 Yeah.

214
00:09:35,050 --> 00:09:36,050
 Yeah.

215
00:09:36,050 --> 00:09:37,050
 Right.

216
00:09:37,050 --> 00:09:38,650
 And so what's the assumption that, you know, networks are controlled by the same entities.

217
00:09:38,650 --> 00:09:39,650
 Right.

218
00:09:39,650 --> 00:09:43,690
 So, and it's a good heuristic and it works well for most of the internet because that's

219
00:09:43,690 --> 00:09:44,690
 how the internet works.

220
00:09:44,690 --> 00:09:45,690
 Right.

221
00:09:45,690 --> 00:09:46,690
 Yeah.

222
00:09:46,690 --> 00:09:47,690
 And obviously, I mean, some networks are bigger than that.

223
00:09:47,690 --> 00:09:50,130
 Some networks are smaller than that.

224
00:09:50,130 --> 00:09:52,530
 But that's that held true for most of the internet.

225
00:09:52,530 --> 00:09:53,530
 It's perfectly fine heuristic.

226
00:09:53,530 --> 00:09:54,530
 Like it works.

227
00:09:54,530 --> 00:09:55,530
 Right.

228
00:09:55,530 --> 00:09:56,530
 Yeah.

229
00:09:56,530 --> 00:09:59,970
 But, but there's the internet has kind of.

230
00:10:00,021 --> 00:10:06,661
 I've changed a bit over the last 15 years in that, well, so first I should say that

231
00:10:06,661 --> 00:10:14,061
 the way that the internet works is that you have regional registrars that control IP address

232
00:10:14,061 --> 00:10:15,061
 space, right?

233
00:10:15,061 --> 00:10:21,621
 So, Europe has a thing called RIPE, U.S.S. Air-In, Asia is APNIC or whatever, and they

234
00:10:21,621 --> 00:10:26,701
 partition the internet amongst themselves, and then they go to entities that they give

235
00:10:26,701 --> 00:10:31,941
 into like say Cloudflare or Amazon or whatever, and then Amazon goes and gives them to their

236
00:10:31,941 --> 00:10:35,661
 customers or consumers or universities, individuals, whatever, right?

237
00:10:35,661 --> 00:10:41,341
 And these entities that enter into a contractual relationship with their regional registrar

238
00:10:41,341 --> 00:10:43,261
 is called an AS, an autonomous system, right?

239
00:10:43,261 --> 00:10:46,981
 So they're called autonomous systems, so that's why it's called ASMap.

240
00:10:46,981 --> 00:10:51,781
 And so these ASs can control lots of different networks, they don't have to control just

241
00:10:51,781 --> 00:10:52,781
 one, right?

242
00:10:52,781 --> 00:10:57,261
 Like Cloudflare A has like multiple ASs because they're really big, and so does Google and

243
00:10:57,261 --> 00:11:02,581
 so does Amazon, but they can also own networks that aren't contiguous, right?

244
00:11:02,581 --> 00:11:09,581
 They can say I want 12.2/16 and all the stuff below it, but I also want 100.11, whatever.

245
00:11:09,581 --> 00:11:13,261
 And so if you just looked at the IP addresses, you wouldn't know that they're the same entity,

246
00:11:13,261 --> 00:11:14,261
 right?

247
00:11:14,261 --> 00:11:16,341
 So you can imagine where like in Core this might be a problem, right?

248
00:11:16,341 --> 00:11:17,341
 Because if you're...

249
00:11:17,341 --> 00:11:23,141
 A sophisticated civil attacker might split it up across there, the things they control.

250
00:11:23,141 --> 00:11:25,941
 Yeah, or the AS is the attacker themselves, right?

251
00:11:25,941 --> 00:11:31,781
 Or the government knocks on the door of the ASs and says, let's hijack the traffic, right?

252
00:11:31,781 --> 00:11:38,481
 So yeah, so this observation that A, you know, multiple, like ASs could have multiple subnetworks,

253
00:11:38,481 --> 00:11:40,141
 and also the fact that they were trading them more often.

254
00:11:40,141 --> 00:11:44,701
 Like if you looked at just how different ASs trade like...

255
00:11:44,701 --> 00:11:46,221
 IP address ranges and zones.

256
00:11:46,221 --> 00:11:50,301
 Yeah, they break them up, they trade them, and it's happening more frequently than it

257
00:11:50,301 --> 00:11:51,301
 used to.

258
00:11:51,301 --> 00:11:52,301
 Interesting, okay.

259
00:11:52,301 --> 00:11:55,341
 So we're having trouble keeping a static map of the internet valid, you know, like 20 years

260
00:11:55,341 --> 00:11:58,321
 ago or even like, you know, I don't know, 30 years ago for sure was not, it was less

261
00:11:58,321 --> 00:11:59,321
 of a problem.

262
00:11:59,321 --> 00:12:00,321
 But today...

263
00:12:00,321 --> 00:12:03,841
 You could map it out once and you would be fine for a long while, and now it's way more

264
00:12:03,841 --> 00:12:06,941
 dynamic process mapping it out.

265
00:12:06,941 --> 00:12:07,941
 Right.

266
00:12:07,941 --> 00:12:08,941
 Yeah.

267
00:12:08,941 --> 00:12:14,421
 So, you know, the idea with ASMap was, okay, well, what we really want to do is map who

268
00:12:14,421 --> 00:12:15,421
 owns these networks.

269
00:12:15,421 --> 00:12:20,621
 It has the control from an entity standpoint over individual peers.

270
00:12:20,621 --> 00:12:21,621
 And so that's...

271
00:12:21,621 --> 00:12:27,301
 So what we're trying to do, well, what ASMap is doing is using, is building maps between

272
00:12:27,301 --> 00:12:28,301
 the IP...

273
00:12:28,301 --> 00:12:30,821
 Instead of using IP networks, we use the autonomous system.

274
00:12:30,821 --> 00:12:35,941
 So we create maps of who owns what IP networks, and then we bucket the peers in your peers

275
00:12:35,941 --> 00:12:38,461
 by AS and not by IP address, right?

276
00:12:38,461 --> 00:12:45,781
 So we change the whole, like, peering code itself to no longer be... no longer respect,

277
00:12:45,781 --> 00:12:49,701
 like, well, I mean, it does a bit of both, but it can also do by AS if it has an ASMap

278
00:12:49,701 --> 00:12:50,701
 to use.

279
00:12:50,701 --> 00:12:51,701
 Okay.

280
00:12:51,701 --> 00:12:52,701
 And that's considered superior, right?

281
00:12:52,701 --> 00:12:53,701
 Yeah.

282
00:12:53,701 --> 00:12:54,701
 Yeah.

283
00:12:54,701 --> 00:12:55,701
 So, yeah.

284
00:12:55,701 --> 00:12:56,701
 Yeah, yeah, yeah.

285
00:12:56,701 --> 00:12:59,381
 So basically, the goal of the ASMap is to be an extra security for your peering system.

286
00:12:59,381 --> 00:13:00,381
 Yeah.

287
00:13:00,381 --> 00:13:05,421
 So you wanna... like, now you... with what you just described, you basically build up

288
00:13:05,421 --> 00:13:09,621
 the connections, and I want one peer from this zone and one peer from this zone and

289
00:13:09,621 --> 00:13:18,301
 one peer from this zone and so on, and you can define the... like you said, you can connect...

290
00:13:18,301 --> 00:13:22,061
 you can initiate a connection based on the AS as the main criteria.

291
00:13:22,061 --> 00:13:27,861
 Well, it's more like when you hear about a peer and, you know, you get its IP address,

292
00:13:27,861 --> 00:13:31,221
 or you're... in this case, it's AS or in this case, both.

293
00:13:31,221 --> 00:13:37,701
 It says, "Oh, you know what, like, I would get this AS123 that is Hetzner or whatever."

294
00:13:37,701 --> 00:13:38,701
 And you know what?

295
00:13:38,701 --> 00:13:39,701
 That bucket's full.

296
00:13:39,701 --> 00:13:40,701
 Yeah.

297
00:13:40,701 --> 00:13:41,701
 If it was your IP address, I might catch you.

298
00:13:41,701 --> 00:13:42,701
 Okay.

299
00:13:42,701 --> 00:13:43,701
 So you don't establish the connection in the first place.

300
00:13:43,701 --> 00:13:44,701
 Right.

301
00:13:44,701 --> 00:13:45,701
 Well, so you might.

302
00:13:45,701 --> 00:13:46,701
 Yeah.

303
00:13:46,701 --> 00:13:48,501
 Or, you know, you'll keep filling up the bucket until it's full, and so you'll say, "Okay,

304
00:13:48,501 --> 00:13:50,581
 you go in this AS 'cause you're Hetzner clearly."

305
00:13:50,581 --> 00:13:51,581
 Yeah.

306
00:13:51,581 --> 00:13:52,581
 Or something else, right?

307
00:13:52,581 --> 00:13:53,581
 Yeah.

308
00:13:53,581 --> 00:13:57,341
 And so the idea is that it's much harder to do that, like to control entire autonomous

309
00:13:57,341 --> 00:13:59,501
 systems than it is to control an individual IP network.

310
00:13:59,501 --> 00:14:00,501
 Yeah.

311
00:14:00,501 --> 00:14:04,941
 And it's elevating the cost of attack of a peer... the cost of a peering attack by itself.

312
00:14:04,941 --> 00:14:05,941
 Yeah.

313
00:14:05,941 --> 00:14:06,941
 Right.

314
00:14:06,941 --> 00:14:07,941
 Yeah.

315
00:14:07,941 --> 00:14:08,941
 So that's the idea, right?

316
00:14:08,941 --> 00:14:09,941
 Then the...

317
00:14:09,941 --> 00:14:12,141
 So I mean, this has existed, you know, all this backend code and core has existed for

318
00:14:12,141 --> 00:14:13,141
 a long time.

319
00:14:13,141 --> 00:14:16,981
 It's been... you know, 20 was in, I think, 2020, maybe earlier.

320
00:14:16,981 --> 00:14:19,301
 But the whole problem with this is how do you actually build these maps?

321
00:14:19,301 --> 00:14:22,461
 Like, how do you build a map of the internet?

322
00:14:22,461 --> 00:14:26,541
 And this is a problem that kind of dragged along for a while 'cause it's quite difficult

323
00:14:26,541 --> 00:14:27,541
 to do.

324
00:14:27,541 --> 00:14:30,381
 And the answer is, well, it's not really possible to, like, do it correctly or, like, fully,

325
00:14:30,381 --> 00:14:31,381
 right?

326
00:14:31,381 --> 00:14:35,301
 'Cause there's no real address book for the internet that doesn't exist, you know?

327
00:14:35,301 --> 00:14:39,581
 So Fabian went on a bunch of big research bent to kind of, like, figure out how to do

328
00:14:39,581 --> 00:14:40,581
 this optimally.

329
00:14:40,581 --> 00:14:42,861
 Like, we went to network research conferences and everybody was like, "No, don't forget

330
00:14:42,861 --> 00:14:43,861
 about it.

331
00:14:43,861 --> 00:14:45,981
 You can't do it."

332
00:14:45,981 --> 00:14:51,981
 But you know, so, you know, in the spirit of no solutions, only trade-offs, we came up

333
00:14:51,981 --> 00:14:52,981
 with...

334
00:14:52,981 --> 00:14:55,461
 So he started working on this project called Cartograph, right?

335
00:14:55,461 --> 00:14:59,861
 Cartograph with a K is is a Python program that fetches

336
00:15:00,010 --> 00:15:22,010
 We've got a lot of information about who owns what AS, and try to merge it as optimally as possible, like giving the trade-offs of what we think is valid, and I can go more deeply into how we assemble these maps if that's interesting, but basically we try to build these maps up as sanely as possible, and as of the last Core release, we're now embedding an ASMap into Core directly.

337
00:15:22,010 --> 00:15:32,010
 It's not on by default, but if you add ASMap on, or I can't remember where the setting is, then we'll use the default ASMap that we embed with the releases.

338
00:15:32,010 --> 00:15:43,010
 Why was the immediate reaction from the network, guys? It's impossible, basically, and don't do it. What's the main issue that you run into? What is it that makes it impossible?

339
00:15:43,010 --> 00:16:01,010
 There's no authenticated way to prove that you own an IP network. The way that the internet is a network of networks, and the way that we find IP addresses on the internet is that there's all these routers that point you in the right direction.

340
00:16:01,010 --> 00:16:21,010
 These are called BGP routers, Border Gateway Protocol. BGP sucks a lot, and there are "attacks", but it's really just how BGP works is that you can just announce whatever you want, and routers will just replicate it unless they have extra checks behind to try to ensure or have someone to call and say, "Is this really your network?"

341
00:16:21,010 --> 00:16:37,010
 There have been plenty of documented attacks. Yeah, I remember. We had some big internet outages a while ago related to BGP 15 years ago. I still remember that I did all my networking classes at the university and so on.

342
00:16:37,010 --> 00:16:58,010
 Long ago, my BGP knowledge is very bad, but that's what I still remember. It's one of the hilliest hills, basically. If you want to bring stuff down and just cause a bunch of ruckus, BGP is the way to go.

343
00:16:58,010 --> 00:17:10,010
 Yeah, and luckily, obviously, this is a known problem, and a lot of people, all the operators of the internet are well aware of the problem set. There's been a lot of thought about how to do it better, so it's hard to pull off.

344
00:17:10,010 --> 00:17:21,010
 These attacks are visible immediately, I guess, is one way. The other part is that people have tried to come up with solutions to this, so there's stuff like BGPsec, which I don't really know much about it, but it's part of making BGP better.

345
00:17:21,010 --> 00:17:31,010
 The thing we use a lot in cartograph is this thing called RPKI. I'm sorry, I'm throwing a ton of backbones, but that is just the way the internet is. It's just alphabet soup.

346
00:17:31,010 --> 00:17:36,010
 I mean, PKI, people should know what PKI stands for. What's the R?

347
00:17:36,010 --> 00:17:55,010
 So, it's resource public key infrastructure. So, yeah, RPKI is basically a way for regional internet registrars like RIPE and ARIN to attest that a certain AS is authorized to advertise a route.

348
00:17:55,010 --> 00:18:09,010
 So, effectively, we have a public key infrastructure around publishing announcements that say, "Okay, I'm AS444 and I'm allowed to advertise these internet routes."

349
00:18:09,010 --> 00:18:22,010
 So, you have a public attestation that is verifiable against the RIR's public key and this thing called a trust anchor, and anybody can just hook up to RPKI servers and check that all this stuff is valid.

350
00:18:22,010 --> 00:18:30,010
 So, this is really good and it's got some limitations and some soft spots, but it's already a huge leap forward in our ability to verify the address.

351
00:18:30,010 --> 00:18:36,010
 At least you have cryptographic assurances. Of course, the PKI infrastructure is still permissioned and those kind of things.

352
00:18:36,010 --> 00:18:47,010
 Yeah, exactly. I mean, so anybody can pull down from it. The RPKI server is already centralized, so there's multiple programs you can use to go verify RPKI information.

353
00:18:47,010 --> 00:18:57,010
 So, the problem is that RPKI is kind of under what's being rolled out and so it covers about 55% of the internet today, which is good.

354
00:18:57,010 --> 00:19:05,010
 Surprisingly good. Usually, if you roll something new out, like IPv6 had this issue for the longest time and still has it, right?

355
00:19:05,010 --> 00:19:09,010
 It's been around forever, but the internet doesn't like to change.

356
00:19:09,010 --> 00:19:17,010
 Indeed, like herding goats. But yes, that's kind of our basis for Cartograph.

357
00:19:17,010 --> 00:19:25,010
 It's like what we use to be the kind of bedrock of AS map, of how we map an AS to an autonomous system.

358
00:19:25,010 --> 00:19:35,010
 We go through all the RPKI info, we validate it, and then we take those routes and those ASs and that's the basis for our map.

359
00:19:35,010 --> 00:19:41,010
 Yeah, I see. Tell me more about kind of what I briefly mentioned before.

360
00:19:41,010 --> 00:19:49,010
 There's all these GitHub issues where various people run certain things and basically everyone needs to arrive at the same result.

361
00:19:49,010 --> 00:19:51,010
 What's that about? Explain this to people.

362
00:19:51,010 --> 00:20:07,010
 Right, so every month or so we do something called a collaborative run, which is where we all decide...

363
00:20:07,010 --> 00:20:07,010
 It's on a certain UNIX timestamp, a certain second, that we're going to launch this cartograph program at.

364
00:20:07,010 --> 00:20:12,461
 And the goal is to see if we can arrive at the exact same ASMap on the other side, right?

365
00:20:12,461 --> 00:20:18,261
 So for our PKI, it's a decentralized set of servers, we're connecting to a bunch of servers around the world,

366
00:20:18,261 --> 00:20:22,301
 and we're fetching all this our PKI data, we're validating it, we're building it into a map,

367
00:20:22,301 --> 00:20:28,181
 and we're also, since it's only 55% of the internet, we're also pulling other information from other public databases

368
00:20:28,181 --> 00:20:31,061
 that are not cryptographically verified, right?

369
00:20:31,061 --> 00:20:33,821
 And we want to to agreement what we don't have.

370
00:20:33,821 --> 00:20:42,581
 And so doing all those things, we want to see if it's possible to get to a final map that is correct.

371
00:20:42,581 --> 00:20:47,221
 And that works. I mean, usually we're, you know, there are certain things like,

372
00:20:47,221 --> 00:20:51,821
 oh, like a route changed right the second we did this, so some of you got in, some of you didn't.

373
00:20:51,821 --> 00:20:54,781
 Or like straight up, like we failed to connect some servers or whatever.

374
00:20:54,781 --> 00:20:58,621
 Yeah. But we're able to see like how different the maps are.

375
00:20:58,621 --> 00:21:05,141
 And most of the time, you know, we gauge like, okay, 50% of the run, if we have more than six people, five or six people running,

376
00:21:05,141 --> 00:21:12,221
 and we get to 50% of people that have the same map. Yeah, you know, and we look at maps that are really divergent to see like what the problem is.

377
00:21:12,221 --> 00:21:19,501
 So yeah, I looked at the output, I was surprised that from the issues I looked at, most people got the same output.

378
00:21:19,501 --> 00:21:23,301
 I was like, oh, wow, that's amazing. It actually works. You know, it really depends.

379
00:21:23,301 --> 00:21:31,381
 Some months don't go so well. And some months are really easy. And, you know, almost everyone gets it. And it's cool.

380
00:21:31,381 --> 00:21:36,341
 We've we built some better tooling these days to kind of try to see the differences between maps, and they're usually pretty minor.

381
00:21:36,341 --> 00:21:45,461
 So I mean, like the idea of doing this is, is a to give assurance to ourselves that like cartograph is a reliable way of building these maps.

382
00:21:45,461 --> 00:21:49,141
 Yeah, which we're pretty confident with now. And, you know, we can inspect them or whatever.

383
00:21:49,141 --> 00:22:00,301
 And the other is to ensure that, you know, especially when we embed it into core, that we are that that the the map we give core was not just like yellowed on someone's computer.

384
00:22:00,301 --> 00:22:05,861
 Yeah, there's a multiple people have done this. We've looked at the map, like we got the exact same hash outputs.

385
00:22:05,861 --> 00:22:12,981
 This wasn't like invented by me. And there is a, you know, like cryptographically verifiable paper trail, so to speak, like how this map came about, right?

386
00:22:12,981 --> 00:22:28,861
 Yeah, yeah, yeah. And, you know, I've thought about, you know, how can we not purify this or like, you know, what if we did, you know, broader publication of hashes and everybody like hit their hash, but only reveal it together at the same time to make sure that, you know, there's also fun stuff you can do.

387
00:22:28,861 --> 00:22:36,581
 But for us, yeah, GitHub is the paper trail. And that's yeah, yeah, sure. I mean, you know, one step at a time.

388
00:22:36,581 --> 00:22:45,701
 Yeah, so so yeah, I mean, you know, anybody can join like a hop in or one of our collaborative runs, you really just have to like run, like set the program to run at the time.

389
00:22:45,701 --> 00:22:50,341
 And it's one flag, you know, yeah, just a thing. And you get an AS map at the end. And that's it.

390
00:22:50,341 --> 00:22:55,061
 Awesome. Yeah, no, I'll definitely put stuff in the show notes so people can join and participate.

391
00:22:55,061 --> 00:23:02,221
 I think it's a fun little thing to do. Very, very low barrier of entry, very low barrier barrier. Run one command, paste the output.

392
00:23:02,221 --> 00:23:19,101
 Yeah, yeah. And you know, it's like the we had this debate about AS mapping core a lot, because, you know, I was kind of of the of the bent of like, well, you know, the features there, like if people won't just go build their own AS maps when it's this easy, I don't know if they're ever going to like, we're gonna have to embed it into core for them to use it.

393
00:23:19,101 --> 00:23:27,381
 Yeah, but I really wish people would just like, whenever you spin up a new version of core, you just, you know, snap a new AS map that's fresh off the presses for yourself.

394
00:23:27,381 --> 00:23:43,901
 And just use it. Like, you do it, you don't have to wait for core for us or for anybody to build one for you. Yeah, yeah, yeah, I see. Yeah, yeah. But it's it's kind of an esoteric topic. So so very few even very few like hardcore Bitcoin users would will care about it in some sense.

395
00:23:43,901 --> 00:23:57,341
 Yeah, that's fair. Yeah, it's a it's a it's definitely I mean, that's a it's a niche subject. Yeah, for sure. For sure. So how come you already spoke about you thought about, you know, notifying some of those things?

396
00:23:57,341 --> 00:24:11,981
 How did you bump into nuster? And what has you kind of excited about it? And and also important to me personally, how did you bump into sovereign engineering? Well, how come how come you discovered the whole thing and decided to sign up and you're here now?

397
00:24:11,981 --> 00:24:19,901
 Yeah, yeah, yeah. Always blows me away that people actually come, you know, like, there's nothing in it for you. Just hanging out in Madeira for a couple of weeks, basically.

398
00:24:19,901 --> 00:24:37,181
 And I don't know. I think there's a lot in it for you. Yeah, no one come guys. This is already plenty fun. Don't join sovereign engineering. No, so nuster. I mean, I was, I think the first person I heard about it from was probably will casserine.

399
00:24:37,181 --> 00:24:46,861
 When I was following him on on Twitter, and he's always hacking together fun stuff, I was looking at I think it was like, Ellen's socket work or whatever. Yeah. And just poking around his little prototypes, which are always super fun.

400
00:24:46,861 --> 00:24:59,901
 And, and, and he learned a lot from just listening to what will cause casserine outputs. And he started poking at nuster. And I was like, Okay, like, what's this nuster thing? And I read, yeah, if you had jobs, like one page on it. And it was just like a good example of like something that was

401
00:25:00,010 --> 00:25:03,310
 I thought so clearly, so well-exposed, and I was like, "Ah, thank you."

402
00:25:03,310 --> 00:25:08,690
 And it seemed to me that it was at the right time where everyone was trying to find the

403
00:25:08,690 --> 00:25:13,690
 next social media because Twitter was kind of a mess at the time.

404
00:25:13,690 --> 00:25:18,890
 And it just came with like, oh, someone finally saw clearly the right mix of clients and servers

405
00:25:18,890 --> 00:25:19,890
 and...

406
00:25:19,890 --> 00:25:21,450
 Yeah, the right trade-offs, kind of, right?

407
00:25:21,450 --> 00:25:22,490
 The right trade-offs, right?

408
00:25:22,490 --> 00:25:23,930
 Right primitives, right trade-offs.

409
00:25:23,930 --> 00:25:28,790
 It's funny, when you deal with technical solutions like this, you can almost feel when the thing

410
00:25:28,790 --> 00:25:29,790
 is right.

411
00:25:29,790 --> 00:25:33,690
 You get it and you're like, "Ah, immediately, I know this is exactly it."

412
00:25:33,690 --> 00:25:37,090
 So I just watched it, I was like, "Where's this thing going?"

413
00:25:37,090 --> 00:25:41,010
 I'll make a little profile and I'll start listening and see what happens on Nostr.

414
00:25:41,010 --> 00:25:42,450
 You know what a Bitcoiner is going to come up with.

415
00:25:42,450 --> 00:25:48,210
 Maybe it goes nowhere and then in half a month or half a year, people fizzle out or whatever.

416
00:25:48,210 --> 00:25:49,530
 And the exact opposite happens.

417
00:25:49,530 --> 00:25:51,530
 It's dying right now.

418
00:25:51,530 --> 00:25:52,530
 Haven't you heard?

419
00:25:52,530 --> 00:25:53,530
 It's dying right now.

420
00:25:53,530 --> 00:25:54,530
 Nostr is dead.

421
00:25:54,530 --> 00:26:00,250
 If the Nostr hive mind is to believe it's dying right now.

422
00:26:00,250 --> 00:26:01,250
 It's the same with Bitcoin.

423
00:26:01,250 --> 00:26:02,250
 It's dying all the time.

424
00:26:02,250 --> 00:26:04,250
 We're off on our little island in the middle of the Atlantic.

425
00:26:04,250 --> 00:26:05,250
 They just can't find us.

426
00:26:05,250 --> 00:26:06,250
 They don't know we're here.

427
00:26:06,250 --> 00:26:07,250
 Yeah, exactly.

428
00:26:07,250 --> 00:26:08,250
 No, yeah.

429
00:26:08,250 --> 00:26:16,410
 And honestly, over the years, just watching the experimentation and the breakneck pace

430
00:26:16,410 --> 00:26:20,650
 of experimentation that happened on Nostr, right?

431
00:26:20,650 --> 00:26:24,730
 And I was just like, "Yeah, this is how I like to social."

432
00:26:24,730 --> 00:26:25,730
 You know?

433
00:26:25,730 --> 00:26:26,730
 I don't want to be back on Twitter.

434
00:26:26,730 --> 00:26:27,730
 I've been off most social media.

435
00:26:27,730 --> 00:26:32,250
 I think there was only one I had and I was like, "Yeah, this is where me and my friends

436
00:26:32,250 --> 00:26:33,250
 can hang out.

437
00:26:33,250 --> 00:26:38,090
 If it's only just us, this is perfectly fine for me."

438
00:26:38,090 --> 00:26:40,290
 And it has not disappointed.

439
00:26:40,290 --> 00:26:45,370
 So for me, it's like the exact right primitive we can build on and people have done so in

440
00:26:45,370 --> 00:26:51,570
 ways that still blow my mind every day and I think the future is very bright.

441
00:26:51,570 --> 00:26:57,330
 What's your daily driver right now or what's your Nostr usage look like?

442
00:26:57,330 --> 00:27:00,970
 You know, I really wish it was better, but my daily driver is just amethyst.

443
00:27:00,970 --> 00:27:01,970
 All right.

444
00:27:01,970 --> 00:27:02,970
 I'm insane.

445
00:27:02,970 --> 00:27:03,970
 Guilty as charged.

446
00:27:03,970 --> 00:27:04,970
 Yeah.

447
00:27:04,970 --> 00:27:05,970
 Yeah.

448
00:27:05,970 --> 00:27:06,970
 I just like...

449
00:27:06,970 --> 00:27:07,970
 Kitchen sink client for the win.

450
00:27:07,970 --> 00:27:08,970
 Yeah, exactly right.

451
00:27:08,970 --> 00:27:12,810
 Yeah, my phone almost takes off and leaves for the next room every time I open amethyst.

452
00:27:12,810 --> 00:27:13,810
 Yeah, no.

453
00:27:13,810 --> 00:27:14,810
 I mean, it's great.

454
00:27:14,810 --> 00:27:16,650
 I mean, it's a great mobile app, et cetera.

455
00:27:16,650 --> 00:27:18,370
 I wish there were better like desktop clients.

456
00:27:18,370 --> 00:27:20,210
 I never found one I really loved.

457
00:27:20,210 --> 00:27:21,210
 Yeah.

458
00:27:21,210 --> 00:27:26,690
 I don't really want to log into the web for Nostr, like I'd really like to just keep it

459
00:27:26,690 --> 00:27:27,690
 on my own.

460
00:27:27,690 --> 00:27:31,930
 I use Nostr, which is actually an iOS client on my desktop since I'm on Mac as you can

461
00:27:31,930 --> 00:27:33,370
 use the iOS apps as well.

462
00:27:33,370 --> 00:27:34,370
 Okay.

463
00:27:34,370 --> 00:27:36,410
 So that kind of works for me, but I agree.

464
00:27:36,410 --> 00:27:42,570
 The desktop clients are subpar, like I'm very close to building my own.

465
00:27:42,570 --> 00:27:43,570
 Very, very close.

466
00:27:43,570 --> 00:27:45,810
 Great, you know, I might do this in summer engineering.

467
00:27:45,810 --> 00:27:46,810
 We'll see.

468
00:27:46,810 --> 00:27:47,810
 All right.

469
00:27:47,810 --> 00:27:48,810
 All right.

470
00:27:48,810 --> 00:27:49,810
 Yeah.

471
00:27:49,810 --> 00:27:50,810
 Please nerd snipe me on it.

472
00:27:50,810 --> 00:27:51,810
 Like I really don't want to.

473
00:27:51,810 --> 00:27:55,050
 And yeah, the other desktop client I use is co-op, but that's just for Nip17, just talking

474
00:27:55,050 --> 00:27:56,450
 to my agents warm basically.

475
00:27:56,450 --> 00:27:57,450
 Yeah, yeah.

476
00:27:57,450 --> 00:28:03,570
 And those days are numbered probably as well because I'll probably follow Pablo's path

477
00:28:03,570 --> 00:28:06,250
 and switch to Nip29 group chats and so on.

478
00:28:06,250 --> 00:28:07,250
 But we'll see.

479
00:28:07,250 --> 00:28:08,250
 Do you use...

480
00:28:08,250 --> 00:28:09,410
 Are you iOS or Android?

481
00:28:09,410 --> 00:28:10,410
 Android.

482
00:28:10,410 --> 00:28:11,410
 Do you use Zapstore a lot?

483
00:28:11,410 --> 00:28:12,410
 Oh yeah.

484
00:28:12,410 --> 00:28:13,410
 Yeah?

485
00:28:13,410 --> 00:28:14,410
 It's a friends app.

486
00:28:14,410 --> 00:28:15,410
 Shout out.

487
00:28:15,410 --> 00:28:16,410
 We never met, but dude, I love you.

488
00:28:16,410 --> 00:28:17,410
 Yeah, I know.

489
00:28:17,410 --> 00:28:18,410
 He's fantastic.

490
00:28:18,410 --> 00:28:19,410
 He's fantastic.

491
00:28:19,410 --> 00:28:25,210
 And yeah, friends, like Zapstore, because it's one of my main nostril use cases.

492
00:28:25,210 --> 00:28:28,850
 It's like I got a new phone, I install Zapstore, and I'm off to the races, and I love it so

493
00:28:28,850 --> 00:28:29,850
 much.

494
00:28:29,850 --> 00:28:30,850
 Yeah.

495
00:28:30,850 --> 00:28:33,450
 And for me, it's kind of like I'm at a point now, if it isn't on Zapstore, it doesn't exist

496
00:28:33,450 --> 00:28:34,450
 for me.

497
00:28:34,450 --> 00:28:35,450
 Yeah.

498
00:28:35,450 --> 00:28:36,450
 Like I probably don't need the app.

499
00:28:36,450 --> 00:28:37,450
 Yeah.

500
00:28:37,450 --> 00:28:38,450
 Exactly.

501
00:28:38,450 --> 00:28:40,970
 And so it's awesome in that way.

502
00:28:40,970 --> 00:28:41,970
 Yeah.

503
00:28:41,970 --> 00:28:44,090
 And it's got a bright future ahead too.

504
00:28:44,090 --> 00:28:47,970
 There's plenty of things left to build on the whole distribution of apps and whatever.

505
00:28:47,970 --> 00:28:48,970
 Yeah.

506
00:28:48,970 --> 00:28:52,530
 So, I mean, you know, it was funny that nostril is like...

507
00:28:52,530 --> 00:28:54,570
 I think I've ever been more bullish on nostril than right now.

508
00:28:54,570 --> 00:28:55,570
 I know I say this.

509
00:28:55,570 --> 00:28:56,570
 I probably say this a lot.

510
00:28:56,570 --> 00:28:58,690
 But now in the nostril bear market, how dare you?

511
00:28:58,690 --> 00:28:59,690
 Oh, no.

512
00:28:59,690 --> 00:29:00,690
 I am so bullish, dude.

513
00:29:00,690 --> 00:29:05,930
 I mean, the Git thing alone, like Jack is so right to push on this that we should be

514
00:29:05,930 --> 00:29:09,810
 rebuilding GitHub on nostril in a way his new buzz thing that he just released is kind

515
00:29:09,810 --> 00:29:10,810
 of in that direction.

516
00:29:10,810 --> 00:29:11,810
 Oh, it's happening.

517
00:29:11,810 --> 00:29:12,810
 And...

518
00:29:12,810 --> 00:29:15,090
 Did you look into and get and get workshop?

519
00:29:15,090 --> 00:29:16,090
 Yeah.

520
00:29:16,090 --> 00:29:17,090
 Deeply.

521
00:29:17,090 --> 00:29:18,090
 Yeah.

522
00:29:18,090 --> 00:29:19,090
 And I'm going to probably build some stuff on it.

523
00:29:19,090 --> 00:29:20,090
 This cohort.

524
00:29:20,090 --> 00:29:21,090
 Awesome.

525
00:29:21,090 --> 00:29:22,090
 Yeah.

526
00:29:22,090 --> 00:29:23,090
 Like we need to get off...

527
00:29:23,090 --> 00:29:24,090
 Like we need to get off Twitter at some point.

528
00:29:24,090 --> 00:29:25,090
 Yeah.

529
00:29:25,090 --> 00:29:26,090
 No, it's coming.

530
00:29:26,090 --> 00:29:27,090
 It's coming.

531
00:29:27,090 --> 00:29:34,530
 I, myself, I'm very close to this transition because I've been using GitHub forever.

532
00:29:34,530 --> 00:29:35,530
 Just like with...

533
00:29:35,530 --> 00:29:39,970
 Like I deleted my Twitter profile when I was like 15 years old or so and this is similar

534
00:29:39,970 --> 00:29:40,970
 for my GitHub account.

535
00:29:40,970 --> 00:29:46,650
 Like it's ancient and the reason why I deleted my Twitter was nostril was there and it was

536
00:29:46,650 --> 00:29:48,090
 good enough and I just...

537
00:29:48,090 --> 00:29:52,970
 I hated Twitter more and more and the same thing is happening right now to me and GitHub,

538
00:29:52,970 --> 00:29:54,570
 my relationship with GitHub.

539
00:29:54,570 --> 00:29:59,130
 It's like a toxic, abusive relationship and I feel like GitHub is abusing me all day every

540
00:29:59,130 --> 00:30:00,130
 day.

541
00:30:00,130 --> 00:30:10,021
 The stuff that always used to work is breaking and it's time I jump ship and go Nostr only for my git stuff as well.

542
00:30:10,021 --> 00:30:21,021
 I'm not 100% ready yet to pull the trigger because I rely on it so much and every time I tried to go purely Nostr-based,

543
00:30:21,021 --> 00:30:24,021
 I always ran into issues and always something was missing that I needed.

544
00:30:24,021 --> 00:30:28,021
 But it is going to get there. I can feel it in my plums.

545
00:30:28,021 --> 00:30:34,021
 The thing that's interesting with GitHub is that you have to unbundle a bunch of things.

546
00:30:34,021 --> 00:30:39,021
 This is just where my git repo lives and people can push and pull to it.

547
00:30:39,021 --> 00:30:43,021
 It's where I read files and interact with code and then explore it.

548
00:30:43,021 --> 00:30:47,021
 There's CI, which is a huge reason why people get locked into GitHub.

549
00:30:47,021 --> 00:30:51,021
 There's the social part, which to me is also super important.

550
00:30:51,021 --> 00:30:55,021
 I actually really love on GitHub being able to just follow people and see what they push or what they're working on.

551
00:30:55,021 --> 00:31:00,021
 I follow so many people and the star thing, like Justin Moon was talking about this as well.

552
00:31:00,021 --> 00:31:05,021
 It's such a big signal. All those little things are so important.

553
00:31:05,021 --> 00:31:08,021
 We have to either unbundle and re-bundle.

554
00:31:08,021 --> 00:31:11,021
 Issue tracking pull requests, of course. All that kind of thing.

555
00:31:11,021 --> 00:31:14,021
 Oh, that stuff. The social thing will be really interesting too.

556
00:31:14,021 --> 00:31:19,021
 What are the parts of GitHub social that we can reinvent on Nostr primitives that work?

557
00:31:19,021 --> 00:31:25,021
 What works with GitHub that we should keep? There's git dev workshop, which is kind of like,

558
00:31:25,021 --> 00:31:31,021
 let's get exactly the GitHub UI, try to get to parity, like we did with the early Nostr clients.

559
00:31:31,021 --> 00:31:36,021
 Let's try to get to parity with what exists and then build from there, which is awesome.

560
00:31:36,021 --> 00:31:40,021
 Shout out Dan Conway. Works awesome.

561
00:31:40,021 --> 00:31:42,021
 But I would like to see more, more different UIs.

562
00:31:42,021 --> 00:31:45,021
 Yeah, Marty Malmy is building his own thing as well.

563
00:31:45,021 --> 00:31:49,021
 Yeah, he has a git interface. I think even Justin built his own.

564
00:31:49,021 --> 00:31:51,021
 Cool.

565
00:31:51,021 --> 00:31:54,021
 Git interface basically just with the features he needs.

566
00:31:54,021 --> 00:31:55,021
 Why not?

567
00:31:55,021 --> 00:31:58,021
 No, it's definitely happening.

568
00:31:58,021 --> 00:32:01,021
 Cool to see also that you looked into it.

569
00:32:01,021 --> 00:32:05,021
 I think about it for Core a lot. It's come up over the years here and there.

570
00:32:05,021 --> 00:32:11,021
 Core has good resilience plans. If GitHub shut us out tomorrow, what would Core do?

571
00:32:11,021 --> 00:32:16,021
 And there's mirrors. People are backed up, all the issues, the pull requests, the history.

572
00:32:16,021 --> 00:32:20,021
 CI, they've been to movement, different directions.

573
00:32:20,021 --> 00:32:26,021
 But if CI went out tomorrow, we could rebuild it fairly quickly, I think, at least for Linux.

574
00:32:26,021 --> 00:32:31,021
 But yeah, Bitcoin is one of the most important software projects in history

575
00:32:31,021 --> 00:32:34,021
 and it should probably not be dependent on Microsoft's company.

576
00:32:34,021 --> 00:32:38,021
 And I agree the infra work around Core is humongous.

577
00:32:38,021 --> 00:32:43,021
 But I think they're going to have to figure out a solution to move towards it.

578
00:32:43,021 --> 00:32:48,021
 Yeah, step by step. I mean, same for there were discussions around the mailing list as well.

579
00:32:48,021 --> 00:32:58,021
 Using the mailing list as the communication hub and getting booted off of certain providers and so on.

580
00:32:58,021 --> 00:33:02,021
 I would like to see Core push more towards...

581
00:33:02,021 --> 00:33:08,021
 be further on the edge of, yeah, we're going to try to do it right and be as independent as possible

582
00:33:08,021 --> 00:33:12,021
 instead of GitHub and Google doing our crutches until they kick us out.

583
00:33:12,021 --> 00:33:18,021
 Yeah, yeah. But I think it's bound to happen. It's just kind of a matter of time

584
00:33:18,021 --> 00:33:22,021
 and also you have to be a little bit pragmatic about the choices as well.

585
00:33:22,021 --> 00:33:29,021
 It has to work. If it makes everyone only half as productive, then it's a really bad move.

586
00:33:29,021 --> 00:33:37,021
 Yeah, yeah, yeah. So I mean, another thing like on that train of thought about GitHub over Noster is like...

587
00:33:37,021 --> 00:33:44,021
 I don't know about you, but for me, LLMs on having the co-pilot in GitHub has been really nice.

588
00:33:44,021 --> 00:33:48,021
 Like it has changed the way I interact with repositories distinctly, right?

589
00:33:48,021 --> 00:33:51,021
 So like when I have any question about a repo or like not even like where's this file,

590
00:33:51,021 --> 00:33:56,021
 but like where's the part of this code base where this logic happens or like does this do what I think it does or etc.

591
00:33:56,021 --> 00:34:01,021
 I don't bother. And you use co-pilot directly on the GitHub interface or...

592
00:34:01,021 --> 00:34:03,021
 Yeah, yeah. And they give it to you for free, right?

593
00:34:03,021 --> 00:34:07,021
 They had never limited me about some questions on there. It's not a fairly advanced model.

594
00:34:07,021 --> 00:34:12,021
 I've never used it. I've never used it. I do the same thing, but I always use my own stuff.

595
00:34:12,021 --> 00:34:15,021
 I mean, I'm almost never in the GitHub interface.

596
00:34:15,021 --> 00:34:17,021
 So you clone locally and then ask questions about it?

597
00:34:17,021 --> 00:34:22,021
 I have a swarm of agents and I just paste the link to repo and tell me about it.

598
00:34:22,021 --> 00:34:25,021
 Okay, you're further ahead than I am. You have to show me your ways.

599
00:34:25,021 --> 00:34:29,021
 Yeah.

600
00:34:29,021 --> 00:34:32,021
 But I think what I was going to get at was like, you know, there's an opportunity for whoever rebuilds GitHub.

601
00:34:32,021 --> 00:34:39,021
 I mean, this is basically what Jack did with Buzz, right? Is like get agents plus get in one interface and see what happens.

602
00:34:39,021 --> 00:34:40,021
 Yeah.

603
00:34:40,021 --> 00:34:44,021
 And I think it's a really good move, really good move.

604
00:34:44,021 --> 00:34:48,021
 Yeah, it's the right move. I had a long conversation today with Pablo about all this.

605
00:34:48,021 --> 00:34:55,021
 And basically the the you just mentioned, you know, like show me your ways. My ways are very simple.

606
00:34:55,021 --> 00:35:00,021
 I'm a very simple man. I want to be in front of the computer for as little time as possible.

607
00:35:00,021 --> 00:35:10,010
 So all my stuff has to be mobile first and basically what I landed on is like I use NIP17 to DM my agents.

608
00:35:10,010 --> 00:35:15,010
 The agents run in various harnesses that are various systems underneath basically.

609
00:35:15,010 --> 00:35:22,010
 Mostly open-claw still just because like I set it up when it kind of came out and I still like it and it does everything I need to do.

610
00:35:22,010 --> 00:35:35,010
 And so for the thing you described if I want to like literally just I open any Nostr client doesn't matter, Amethyst doesn't, Wisp is more reliable on NIP17-DM so I use Wisp mostly for that.

611
00:35:35,010 --> 00:35:46,010
 And on the desktop I use Co-op and so on and I just you know I pick my, I have one it's called Sovereign Sven and he knows all about the sovereign tech I'm interested in.

612
00:35:46,010 --> 00:35:55,010
 And I will paste in the GitHub link or repo link whatever it doesn't matter you know it can be GitLab, Gitia, whatever, plain also hosted on Nostr repo.

613
00:35:55,010 --> 00:36:04,010
 It's like tell me more about this what's the most interesting, what's the most interesting feature of this thing and so on and then I just go back and forth and I'm on my phone or whatever.

614
00:36:04,010 --> 00:36:08,010
 Also text to speech and voice and so on like this stuff works.

615
00:36:08,010 --> 00:36:18,010
 So these are my ways because then I can do it here you know like I walk up and down or I'm at the beach and I'll just I'll do this stuff and not stare into a screen all the time.

616
00:36:18,010 --> 00:36:20,010
 So it's almost like GitHub just died on its own.

617
00:36:20,010 --> 00:36:26,010
 Like everyone just like walks out of the interface and like if you have agents and if you have Git servers.

618
00:36:26,010 --> 00:36:30,010
 That's part of the agentic world, the death of UI is part of the agentic world.

619
00:36:30,010 --> 00:36:36,010
 I think it's one of the one of the very logical consequences.

620
00:36:36,010 --> 00:36:39,010
 You think voice will become the primary interface?

621
00:36:39,010 --> 00:36:44,010
 I think the differentiation between text and speech will continue to blur.

622
00:36:44,010 --> 00:36:50,010
 It definitely has for me, but I also realized that I'm an insane person and you know like.

623
00:36:50,010 --> 00:36:51,010
 You're just ahead of the curve.

624
00:36:56,010 --> 00:37:05,010
 Dropped off of the curve a long time ago, but to me it kind of doesn't matter for some things voice is way better like if you if you have.

625
00:37:07,010 --> 00:37:11,010
 The good thing about writing is it really helps you think it helps you to distill your thoughts.

626
00:37:11,010 --> 00:37:26,010
 The good thing about a long rambling session is like if you don't even know what the problem is you just just just talking into a microphone for like 15 minutes while you're walking is like fantastic and other lamps are insanely good at distilling the essence of it and so on.

627
00:37:26,010 --> 00:37:33,010
 And so I might one of my workflows just to stick with the example you posted like new repo I want to figure out I don't know how Bitchat works or whatever.

628
00:37:33,010 --> 00:37:46,010
 OK, there's this new thing and I paste it in like look at this repo and then in the same same session like same chat basically same agent everything so have a good overview and then I'll ask a follow up question along ramble voice note.

629
00:37:46,010 --> 00:38:03,010
 You know and I just go like ah that sounds interesting tell me more about you know like the way it utilizes Bluetooth or the way it utilizes and pops like what is an identity in Bitchat and so on and what else so I might come up with and I just that's it.

630
00:38:03,010 --> 00:38:09,010
 I hit send and then five minutes later I get a text message back I might use you know like the phone can also read it to me and so on.

631
00:38:09,010 --> 00:38:24,010
 And so that's I think the line between text to speech will continue to blur and I know it does also for Pablo for example who is even more insane than me and he will have all his agents work and stuff as autonomously as possible.

632
00:38:24,010 --> 00:38:33,010
 And then one of the agents will go over all the work and produce a podcast produce a podcast episode and his podcast player will just pop up.

633
00:38:33,010 --> 00:38:46,010
 Oh you know like daily update of what happened so probably living his own life anymore. No no he made himself a cog in the machine and what the machine is doing no one knows what's the goal of the machine no one knows.

634
00:38:46,010 --> 00:39:01,010
 Did you did you see what you push last night. No. No it's really funny so he's he's very close to the peak of a psychosis I think he's not at the peak yet so yeah we'll see.

635
00:39:01,010 --> 00:39:10,010
 You know what I was thinking about with like speech and text interfaces was you know one interface that has not changed with computers in like 50 fucking years.

636
00:39:10,010 --> 00:39:16,010
 It is a keyboard and mouse. Yeah we've been we've been stuck with like graphical terminals.

637
00:39:16,010 --> 00:39:37,010
 And yeah typey typey on the keyboard and like a little mouse around your little guy. Yeah as the default mode of how most of humanity deals with computing. Yeah right. And I think the fact that we have LMS and we can do like voice speech text is going to radically change the nature of like interfaces with computers for society as a whole right.

638
00:39:37,010 --> 00:39:52,010
 Yeah I think it's a pretty massive change. Yeah no that's definitely happening. And also what what fear Jeff for example doesn't understand is like 98 percent of the human population that interface with compute interfaces on the mobile phone.

639
00:39:52,010 --> 00:39:59,010
 And so so so that most people don't have computers anymore like everyone just has a phone and the phones are still it's still.

640
00:40:00,021 --> 00:40:10,061
 It's still point-and-click basically, but the interface has evolved to all the buttons are bigger, the interfaces are smaller, mobile-friendly, and so on.

641
00:40:10,061 --> 00:40:20,661
 And I think the next step of this evolution will definitely be kind of, people just will have, it's already happening with AirPods and so on, tiny earplugs.

642
00:40:20,661 --> 00:40:33,501
 Or also with, like, everything dematerializes further. You have your watch, you know, your smartwatch, it can display some things, you have something in your ear that can whisper to you, and I think that's just where it's heading, it's very natural.

643
00:40:33,501 --> 00:40:46,781
 Also, some of these rings, you know the Pebble smartwatch? The guy who built the Pebble smartwatch, he's building a ring now, and the ring has one single button and all he does is record a voice note.

644
00:40:46,781 --> 00:40:57,101
 So basically, that will be one of the input interfaces, where you just go like, you know, 20-second voice note, and it will go to your phone and do things and so on.

645
00:40:57,101 --> 00:41:07,741
 I think, you know, some of this, it's not quite like, the thing that's nice about computers and less so on mobile is that they are not on us, or around us, right?

646
00:41:07,741 --> 00:41:11,781
 You have to go to a desk and like interact with the machine. Yes, it's frustrating.

647
00:41:11,781 --> 00:41:22,061
 No, no, I guess, what do you mean? But then like, yeah, and so like the first instinct for people with AI is like, okay, can I like wear it? Can I have it, you know, in my ear? Can I have it in my glasses? Can I have it in my ring?

648
00:41:22,061 --> 00:41:33,781
 Yeah. And I don't know if that's gonna stick, right? Like, I don't think people like that. I mean, some people are like, really into it, and I've seen a lot more wearable AI stuff in the last six months than I have in a long time forever.

649
00:41:33,781 --> 00:41:47,061
 But I wonder if people will not always want computing to be outside of themselves. Like, if we had drones that could be solar powered and just follow us around all day, I think you would much prefer that with their AI than they would to have something on them at all times.

650
00:41:47,061 --> 00:42:08,141
 I think I know what you mean. Like, for me as well, especially if I start a new project, I need to go into the workshop and hammer on the thing intensely for like a couple hours, and do really like rapid iterations and be really close to the matter and literally be in the workshop. There is another way. I can't do it on the go. I can't do it from my phone and so it doesn't work.

651
00:42:08,141 --> 00:42:22,741
 However, once I'm in maintenance mode, for example, and someone reports a bug, and it's, you know, one of my agents fixed it, and I just have to look over it and press merge, I can do that on my phone, or get a briefing like a daily briefing or something, some summary of stuff that happened.

652
00:42:22,741 --> 00:42:29,741
 I'm fine with, you know, getting that on mobile or even in voice message format at the end of the day or something like that.

653
00:42:29,741 --> 00:42:32,941
 So kind of, you know, right, right tool for the job kind of thing.

654
00:42:32,941 --> 00:42:40,461
 Yeah, you could imagine like your phone, because I mean, they're not even phones anymore, right? No, no, no, no, no one's using their phone for phone purposes.

655
00:42:40,461 --> 00:42:55,741
 I mean, we still have, you know, you know, however many gigs of memory and phone plan and data, whatever, it's a full blown supercomputer. Yeah, exactly. So, like, if everything is remote, and it's just like your agent texting you what's up all the time, or telling you, you're giving you little yes or no prompts or whatever, how small can the phone be?

656
00:42:55,741 --> 00:43:11,341
 How much can we remove from this idea of a smartphone? Hey, hey, hey, next guest appearance. Yeah, we're recording. I'll see you around. Take care, man. Small island.

657
00:43:11,341 --> 00:43:20,741
 Yeah, yeah, yeah. He's just pretending he's not that popular. No, really small island. We have to chase away his fans. That's crazy.

658
00:43:20,741 --> 00:43:39,141
 Bitcoin engineer, obviously. Yeah, yeah, yeah. Really good guy. Yeah, he did lots of lighting infrastructure stuff. Yeah, yeah, yeah. Like really deep into the weeds. Yeah, you wouldn't guess it if you see him, right? He looks more like a CrossFit type of person or something.

659
00:43:39,141 --> 00:43:55,741
 Big corners, man, we're all shapes and sizes, and we're everywhere. It's crazy. Yeah, really cool. Yeah, let's see how it evolves on the computing side. I mean, it's definitely exciting times. Everyone has some level of AI psychosis. Stuff is moving really fast.

660
00:43:55,741 --> 00:44:10,741
 Do you have any, like, sci fi references that you think about when you think of AI, like things you'd like to see? Things that have been like, ah, you know, man, I can't wait till we get to that point where like, AI is like that, or like, sounds are like that, or whatever.

661
00:44:10,741 --> 00:44:39,741
 Good question. Really good question. Whoo, I hope not. Lots of the sci fi I really like is like this topic, you know? I was kind of mining for positive examples. Yeah, I don't know. Like, like, the first thing that popped into my head now when you asked a question was snow crash. And basically like something, something that is so sophisticated that it takes over your mind.

662
00:44:39,741 --> 00:45:00,741
 And you see that happening kind of like Bitcoin is a little bit like that, you know, you get infected by the Bitcoin virus. Yeah, yeah, yeah. And, and, and so on. But no, but you mentioned the autonomous solar powered drones, I would love to see that I saw some prototypes years ago, I used to do lots of big scare quotes, like extreme sports.

663
00:45:00,741 --> 00:45:02,610
 Snowboarding, skateboarding, BMX riding, those kinds of things.

664
00:45:02,610 --> 00:45:05,010
 And it would be awesome to have just a little drone

665
00:45:05,010 --> 00:45:08,010
 that you can just throw over your shoulder and it will unfold

666
00:45:08,010 --> 00:45:11,410
 and just follow you automatically and kind of, you know,

667
00:45:11,410 --> 00:45:13,610
 like shoot some cool things or whatever.

668
00:45:13,610 --> 00:45:14,910
 Yeah, yeah, yeah.

669
00:45:14,910 --> 00:45:18,010
 So, yeah, I don't know. On the AI side, I mean...

670
00:45:19,610 --> 00:45:21,810
 I don't know. It's a really great question.

671
00:45:21,810 --> 00:45:24,410
 I think, I mean, I talk about this a lot.

672
00:45:24,410 --> 00:45:28,610
 I think I mentioned it almost every single recorded dialogue we do

673
00:45:28,610 --> 00:45:31,610
 is like, LLMs have their limits.

674
00:45:31,610 --> 00:45:34,110
 And one of the limits is they are not embodied.

675
00:45:34,110 --> 00:45:34,610
 Yeah.

676
00:45:34,610 --> 00:45:37,710
 And because we humans, we're just going to do it.

677
00:45:37,710 --> 00:45:39,510
 We are going to embody LLMs.

678
00:45:39,510 --> 00:45:44,310
 And I talked with Kelly, for example, about different approach

679
00:45:44,310 --> 00:45:48,110
 to building artificial intelligence systems is building robots

680
00:45:48,110 --> 00:45:51,610
 and let them build up world models themselves.

681
00:45:51,610 --> 00:45:54,610
 And I'm kind of...

682
00:45:54,610 --> 00:45:56,610
 I'm looking forward to seeing that evolve.

683
00:45:56,610 --> 00:45:59,510
 And I wouldn't mind having a, I don't know,

684
00:45:59,510 --> 00:46:03,710
 like an autonomous spider robot in my house,

685
00:46:03,710 --> 00:46:06,010
 just like a watchdog for like...

686
00:46:06,010 --> 00:46:08,910
 Instead of having a security camera, have a thing like that.

687
00:46:08,910 --> 00:46:09,410
 Yeah.

688
00:46:09,410 --> 00:46:09,910
 And...

689
00:46:09,910 --> 00:46:12,210
 Was it Snowcrash where there were like robotic dogs?

690
00:46:12,210 --> 00:46:12,910
 Pretty sure.

691
00:46:12,910 --> 00:46:14,210
 Was it another...

692
00:46:15,410 --> 00:46:16,510
 Snowcrash had many things.

693
00:46:16,510 --> 00:46:18,810
 Yeah. They all burned around there.

694
00:46:18,810 --> 00:46:20,610
 But yeah, so, I mean...

695
00:46:20,610 --> 00:46:22,110
 So, I mean, one thing that's interesting with embodiment

696
00:46:22,110 --> 00:46:23,910
 is that they're no longer LLMs, right?

697
00:46:23,910 --> 00:46:24,710
 When we talk about world models,

698
00:46:24,710 --> 00:46:26,310
 we're not talking about large language models anymore.

699
00:46:26,310 --> 00:46:27,510
 Exactly. Yeah, yeah.

700
00:46:27,510 --> 00:46:30,210
 So, to go over the limit,

701
00:46:30,210 --> 00:46:34,010
 to break through that limit that LLMs will always have,

702
00:46:34,010 --> 00:46:36,710
 one of the requirements is embodiment.

703
00:46:36,710 --> 00:46:39,410
 Well, so I'm not sure if they're not just differences in kind, right?

704
00:46:39,410 --> 00:46:42,610
 Because like, the reason why LLMs are so successful at software engineering

705
00:46:42,610 --> 00:46:45,310
 is because it's ultimately written text and code, right?

706
00:46:45,310 --> 00:46:47,710
 That they can manipulate and they're so good at this, right?

707
00:46:47,710 --> 00:46:51,010
 But a world model is no longer language, right?

708
00:46:51,010 --> 00:46:52,710
 And so like, there's no, like...

709
00:46:52,710 --> 00:46:54,510
 But you can also plop an LLM on top

710
00:46:54,510 --> 00:46:57,610
 in the sense of like, our prefrontal cortex is like an LLM,

711
00:46:57,610 --> 00:46:59,610
 but the rest of our brain is not.

712
00:46:59,610 --> 00:47:05,510
 You know, the closer you go to the brain stem,

713
00:47:05,510 --> 00:47:06,910
 the more different it is.

714
00:47:06,910 --> 00:47:07,610
 Yeah, yeah, yeah.

715
00:47:07,610 --> 00:47:10,510
 I'm obviously simplifying by a lot, but you get what I mean.

716
00:47:10,510 --> 00:47:14,510
 Yeah, and I wonder how, what that jump looks like, right?

717
00:47:14,510 --> 00:47:16,110
 Are we going to find out that, oh, shit, actually,

718
00:47:16,110 --> 00:47:19,810
 like the jump from LLM to whatever happens in the brain

719
00:47:19,810 --> 00:47:21,210
 is like actually humongous,

720
00:47:21,210 --> 00:47:23,310
 and we need a totally different science to do it

721
00:47:23,310 --> 00:47:25,910
 than the large language models we've learned so far.

722
00:47:25,910 --> 00:47:30,010
 I mean, you could argue like vision is a good counter example, right?

723
00:47:30,010 --> 00:47:31,610
 Like vision actually works great, right?

724
00:47:31,610 --> 00:47:33,410
 We have self-driving cars, etc.

725
00:47:35,410 --> 00:47:37,210
 But yeah, so I mean, this is Jan LeCun.

726
00:47:37,210 --> 00:47:41,410
 But it's still, you say vision works great.

727
00:47:41,410 --> 00:47:43,810
 Yes, but general purpose vision is really hard

728
00:47:43,810 --> 00:47:47,210
 because a self-driving car cares about a very specific subset

729
00:47:47,210 --> 00:47:48,210
 of the things you can see.

730
00:47:48,210 --> 00:47:48,910
 Yeah, yeah, yeah, sure.

731
00:47:48,910 --> 00:47:51,010
 And so your vision, and like, again,

732
00:47:51,010 --> 00:47:52,910
 quoting my favorite guy, John Ravecki,

733
00:47:52,910 --> 00:47:59,210
 like your whole perceptive apparatus depends on your value structure,

734
00:47:59,210 --> 00:48:00,110
 on your belief system.

735
00:48:00,110 --> 00:48:01,510
 It's like, what is salient to you?

736
00:48:01,510 --> 00:48:03,210
 What do you care about, and so on?

737
00:48:03,210 --> 00:48:04,610
 It's like, you know, if you're thirsty,

738
00:48:04,610 --> 00:48:06,310
 everything becomes a water source.

739
00:48:06,310 --> 00:48:11,210
 The example I was, used is, again, like I used to be,

740
00:48:11,210 --> 00:48:12,510
 like skateboarding was my life.

741
00:48:12,510 --> 00:48:13,410
 I was a skateboarder.

742
00:48:13,410 --> 00:48:15,010
 I started when I was three years old, you know,

743
00:48:15,010 --> 00:48:16,910
 and I did nothing but skateboarding.

744
00:48:16,910 --> 00:48:18,910
 And so everything to me was like a handrail

745
00:48:18,910 --> 00:48:21,310
 and something to grind down to and like a latch

746
00:48:21,310 --> 00:48:23,010
 or a staircase was not a staircase.

747
00:48:23,010 --> 00:48:25,610
 It was like, can I jump it or can I not jump it?

748
00:48:25,610 --> 00:48:27,710
 Even like, this still annoys me.

749
00:48:27,710 --> 00:48:30,610
 Like, you cannot skate on this asphalt, you know,

750
00:48:30,610 --> 00:48:32,110
 like it's absolutely terrible.

751
00:48:32,110 --> 00:48:34,910
 We're looking at really jarred out asphalt

752
00:48:34,910 --> 00:48:35,610
 that was terrible for us.

753
00:48:35,610 --> 00:48:37,510
 It's a gravel road, basically, you know,

754
00:48:37,510 --> 00:48:41,310
 and so that's still, this perceptive structure is still inside me

755
00:48:41,310 --> 00:48:43,910
 and I see, I notice all these kind of things

756
00:48:43,910 --> 00:48:45,310
 because I'm still a skateboarder.

757
00:48:45,310 --> 00:48:45,810
 Yeah, yeah.

758
00:48:45,810 --> 00:48:47,510
 My soul is still skateboarding.

759
00:48:47,510 --> 00:48:47,910
 Sure, sure.

760
00:48:47,910 --> 00:48:48,610
 I'm not anymore.

761
00:48:48,610 --> 00:48:52,610
 I'm too old, but but yeah, perception is super tricky.

762
00:48:52,610 --> 00:48:57,110
 So this will be very interesting to see and like from a cognitive science

763
00:48:57,110 --> 00:49:00,210
 perspective, we know that it will be very different than the world

764
00:49:00,210 --> 00:49:06,210
 of LLMs we have now, but it will also be necessary because the one thing

765
00:49:06,210 --> 00:49:10,110
 that LLMs can't do at all is care about the information they are processing.

766
00:49:10,110 --> 00:49:13,210
 We care about the information we're processing because we don't want to die

767
00:49:13,210 --> 00:49:14,310
 and we care about ourselves.

768
00:49:14,310 --> 00:49:15,410
 We care about our kids.

769
00:49:15,410 --> 00:49:16,610
 We care about society.

770
00:49:16,610 --> 00:49:19,210
 We care about the true, the good and the beautiful and so on.

771
00:49:19,210 --> 00:49:23,810
 And so one challenge will be how do you make any artificial intelligence

772
00:49:23,810 --> 00:49:27,210
 system care about itself and the world and so on.

773
00:49:27,210 --> 00:49:28,710
 So it's going to be a long journey.

774
00:49:28,710 --> 00:49:32,510
 We're not at the end of the, you know, singularity might be near,

775
00:49:32,510 --> 00:49:36,410
 but it's not as near as Silicon Valley Bros would like you to believe.

776
00:49:36,410 --> 00:49:41,210
 Yeah, I mean, the interesting part is that we'll get to bootstrap LLMs

777
00:49:41,210 --> 00:49:44,010
 to write the code to build the world models and embody that.

778
00:49:44,010 --> 00:49:53,710
 Yeah, we're definitely on an exponential and we'll see where it goes.

779
00:49:53,710 --> 00:49:58,110
 But yeah, I'd be interested in kind of.

780
00:50:00,021 --> 00:50:07,461
 Your thoughts on Nostr and some of the things that we touched on in the first week of the

781
00:50:07,461 --> 00:50:08,461
 court?

782
00:50:08,461 --> 00:50:12,061
 Like, you're a new participant, it's been your first week.

783
00:50:12,061 --> 00:50:16,461
 There was a lot of, like, we went very broad, kind of, you know, like, lots of different

784
00:50:16,461 --> 00:50:18,621
 Nostr primitives.

785
00:50:18,621 --> 00:50:23,821
 Naplets were a big part of the first week's sandwich at the Naplet Workshop on Wednesday.

786
00:50:23,821 --> 00:50:24,821
 Fantastic.

787
00:50:24,821 --> 00:50:26,541
 I really like it.

788
00:50:26,541 --> 00:50:30,141
 This workshop I've ever come to, in the sense of, like, everybody got it all running, and

789
00:50:30,141 --> 00:50:32,621
 we all shipped it with it, like, half an hour was crazy.

790
00:50:32,621 --> 00:50:37,421
 80% success rate of the people that were there, it was really good.

791
00:50:37,421 --> 00:50:42,661
 Obviously, lots of FIPS talk, and that's what I kind of want to pick your brain about.

792
00:50:42,661 --> 00:50:49,221
 Since you come from the network world, we went really deep into ASMap, and, you know,

793
00:50:49,221 --> 00:50:54,701
 what is the issue with IP addresses and clustering and all those kind of things?

794
00:50:54,701 --> 00:51:00,101
 And what are your thoughts on FIPS, and what do you know about it up to now, and are you

795
00:51:00,101 --> 00:51:01,101
 excited about it?

796
00:51:01,101 --> 00:51:02,101
 Do you think it's stupid?

797
00:51:02,101 --> 00:51:03,901
 Like, talk to me about it.

798
00:51:03,901 --> 00:51:05,341
 So I love the idea.

799
00:51:05,341 --> 00:51:12,021
 I haven't evaluated, like, the technical internals, not just, like, only, like, service level,

800
00:51:12,021 --> 00:51:16,501
 and I've messed with it slightly, but it's interesting, because you could basically do

801
00:51:16,501 --> 00:51:19,541
 an introduction to FIPS by describing ASMap, right?

802
00:51:19,541 --> 00:51:20,541
 Yeah.

803
00:51:20,541 --> 00:51:23,701
 And, you know, if you have FIPS, well, none of the ASMap problem even exists, right?

804
00:51:23,701 --> 00:51:24,701
 Yeah.

805
00:51:24,701 --> 00:51:27,061
 And so you don't need ASMap as a solution, which is really cool.

806
00:51:27,061 --> 00:51:28,061
 Yeah.

807
00:51:28,061 --> 00:51:37,341
 So, yeah, like, FIPS is this idea that you can create an overlay and skip DNS, and point,

808
00:51:37,341 --> 00:51:40,301
 and create an intranet between cryptographically- Skip IP, see?

809
00:51:40,301 --> 00:51:44,541
 Well, that's what I don't really get, like, I mean, yes, you're routing towards, you're

810
00:51:44,541 --> 00:51:47,221
 obviously routing towards an IPv6 address, or is that just locally?

811
00:51:47,221 --> 00:51:48,221
 No, you don't have to.

812
00:51:48,221 --> 00:51:50,421
 You can also route to an input.

813
00:51:50,421 --> 00:51:52,861
 But why do you need to route to an input?

814
00:51:52,861 --> 00:51:56,581
 The IPv6 just comes in to make it compatible with the existing internet.

815
00:51:56,581 --> 00:51:57,581
 Okay.

816
00:51:57,581 --> 00:51:58,581
 Yeah.

817
00:51:58,581 --> 00:51:59,581
 Okay.

818
00:51:59,581 --> 00:52:00,581
 So I don't quite follow how that works.

819
00:52:00,581 --> 00:52:05,541
 I need to sit down with Arjen, and- Let's say you only have LoRa devices, so whatever,

820
00:52:05,541 --> 00:52:06,981
 and like, you pick your thing.

821
00:52:06,981 --> 00:52:07,981
 Pick your radio.

822
00:52:07,981 --> 00:52:08,981
 Yes, you can pick radio.

823
00:52:08,981 --> 00:52:09,981
 Yeah.

824
00:52:09,981 --> 00:52:12,501
 But if you're using the internet, what are you using?

825
00:52:12,501 --> 00:52:14,181
 But then you just go to radio to radio.

826
00:52:14,181 --> 00:52:17,181
 IP addresses don't even factor.

827
00:52:17,181 --> 00:52:27,221
 You can just, you know, you run into a guy who wants to sell you muscles, that's one

828
00:52:27,221 --> 00:52:29,301
 of the benefits of being on an island.

829
00:52:29,301 --> 00:52:30,301
 It's true.

830
00:52:30,301 --> 00:52:31,301
 Not bad.

831
00:52:31,301 --> 00:52:32,301
 Very fresh.

832
00:52:32,301 --> 00:52:33,301
 He just got it.

833
00:52:33,301 --> 00:52:34,301
 Infinite cetaceans.

834
00:52:34,301 --> 00:52:38,221
 So, I mean, yeah, no, I get like, you know, you can have a transport over a radio or over

835
00:52:38,221 --> 00:52:46,821
 another medium, but if the medium is the internet, how do you route traffic if it's not

836
00:52:46,821 --> 00:52:49,541
 like, what do you mean, route to an NPUB?

837
00:52:49,541 --> 00:52:53,441
 Like an NPUB is in an IP place.

838
00:52:53,441 --> 00:52:54,901
 It's hosted somewhere.

839
00:52:54,901 --> 00:52:55,901
 No.

840
00:52:55,901 --> 00:53:00,841
 So the issue that all mesh networks face, all of those new things is like, you would

841
00:53:00,841 --> 00:53:06,581
 have to rebuild everything on top of the mesh system to make anything talk, any node talk

842
00:53:06,581 --> 00:53:08,441
 to any other node, right?

843
00:53:08,441 --> 00:53:15,261
 To make fibs actually work in practice is like, if we have the internet already, we have compute

844
00:53:15,261 --> 00:53:17,481
 and networking infrastructure already.

845
00:53:17,481 --> 00:53:24,381
 So fibs right now, if you go the IPv6 route, acts kind of as an overlay network on the

846
00:53:24,381 --> 00:53:33,941
 existing stuff and the exit point kind of is an IPv6 IP address.

847
00:53:33,941 --> 00:53:40,461
 And the NPUB maps deterministically to IPv6, but that means then you have to go to, you

848
00:53:40,461 --> 00:53:41,661
 go through like...

849
00:53:41,661 --> 00:53:42,661
 Yeah, okay.

850
00:53:42,661 --> 00:53:43,661
 No, okay.

851
00:53:43,661 --> 00:53:44,661
 Yeah.

852
00:53:44,661 --> 00:53:47,861
 The TCP/IP is then the channel, but you can use...

853
00:53:47,861 --> 00:53:49,141
 It's transport agnostic.

854
00:53:49,141 --> 00:53:55,541
 So you can use a different transport layer and like what sandwich try to show with just

855
00:53:55,541 --> 00:53:57,381
 using sound, using sound waves.

856
00:53:57,381 --> 00:53:59,141
 Yeah, yeah, okay.

857
00:53:59,141 --> 00:54:00,141
 Yeah.

858
00:54:00,141 --> 00:54:01,141
 Sure.

859
00:54:01,141 --> 00:54:02,141
 And then IP addresses don't even come into the picture.

860
00:54:02,141 --> 00:54:08,141
 So it's the layer under IP layer, whatever layer that is.

861
00:54:08,141 --> 00:54:11,221
 You know, you choose me as a network guy as hilarious because I really have very little

862
00:54:11,221 --> 00:54:12,221
 network knowledge.

863
00:54:12,221 --> 00:54:14,421
 I'm not sure if it's up or down.

864
00:54:14,421 --> 00:54:20,461
 But it's basically the beautiful thing that makes it backward compatible is that you can

865
00:54:20,461 --> 00:54:26,661
 deterministically derive an IPv6 address from an NPUB and vice versa.

866
00:54:26,661 --> 00:54:32,101
 But inside of FIPS, all the routing and everything only goes to an NPUB.

867
00:54:32,101 --> 00:54:33,101
 Yeah.

868
00:54:33,101 --> 00:54:37,901
 And there's a whole magic around like doing an actual mesh network via NPUBs and relays

869
00:54:37,901 --> 00:54:38,901
 and whatever.

870
00:54:38,901 --> 00:54:39,901
 Yeah, yeah, yeah.

871
00:54:39,901 --> 00:54:40,901
 Totally.

872
00:54:40,901 --> 00:54:47,541
 And if this is how, as advertised, like this is like we're building a new transport for

873
00:54:47,541 --> 00:54:51,941
 the internet that is going to be far more independent than what we had before.

874
00:54:51,941 --> 00:54:54,701
 And that is so rock and roll.

875
00:54:54,701 --> 00:54:55,701
 Super based.

876
00:54:55,701 --> 00:54:56,701
 Yeah.

877
00:54:56,701 --> 00:54:57,701
 That I'm just about right.

878
00:54:57,701 --> 00:54:59,781
 It's a new addressing mechanism.

879
00:54:59,781 --> 00:55:00,021
 That's what's happening.

880
00:55:00,021 --> 00:55:05,270
 It moves it away from IP addresses, and it doesn't require DNS in the traditional sense.

881
00:55:05,270 --> 00:55:07,910
 That's what makes it so powerful, because DNS is one of the archaeologist's heels of

882
00:55:07,910 --> 00:55:13,250
 the internet, and IP addresses, for all the things we talked about, especially IP before,

883
00:55:13,250 --> 00:55:14,250
 are terrible.

884
00:55:14,250 --> 00:55:19,590
 It's permissioned by default, and so FIPS solves both of these problems, and the rest

885
00:55:19,590 --> 00:55:30,830
 of the, like, again, so that's kind of simple, and the hard part is then, it depends on the

886
00:55:30,830 --> 00:55:31,830
 transport layer.

887
00:55:31,830 --> 00:55:39,070
 Like, making FIPS work over Bluetooth is harder than making FIPS work over, you know, or making

888
00:55:39,070 --> 00:55:42,670
 FIPS work over sound waves is harder than making FIPS work over PCB.

889
00:55:42,670 --> 00:55:43,670
 Sure, sure, sure.

890
00:55:43,670 --> 00:55:44,670
 But then it can be arbitrary, right?

891
00:55:44,670 --> 00:55:45,670
 You can pick whatever you want.

892
00:55:45,670 --> 00:55:46,670
 You can pick whatever you want.

893
00:55:46,670 --> 00:55:52,270
 Like, if you go to the FIPS website, FIPS.network, you'll see a FIPS package, you know, going

894
00:55:52,270 --> 00:55:57,430
 over the wire, then going over Tor, then going over Bluetooth, and you know, on Tor it slows

895
00:55:57,430 --> 00:56:00,210
 down massively, and Bluetooth is slow as well, and so on.

896
00:56:00,210 --> 00:56:04,350
 But it's all FIPS, and it's just the underlying transport layer is what changes.

897
00:56:04,350 --> 00:56:07,670
 It's like, one idea I had that I was talking about right about last night was like, wow,

898
00:56:07,670 --> 00:56:11,410
 it would be cool, you know, like, what if we, you know, add another transport layer to

899
00:56:11,410 --> 00:56:16,950
 a Bitcoin Core, so we have Tor, IQP, and then FIPS, and then I was like, oh, well, wait,

900
00:56:16,950 --> 00:56:20,590
 if it just does that BV6, it already just works out of the box.

901
00:56:20,590 --> 00:56:25,390
 You can have a FIPS peer be a Bitcoin Core node, and there was no problem, right, it

902
00:56:25,390 --> 00:56:26,390
 just works.

903
00:56:26,390 --> 00:56:29,030
 So I want to test this out and make sure I'm right about that.

904
00:56:29,030 --> 00:56:33,270
 That'd be, you can do all sorts of fun things here, right?

905
00:56:33,270 --> 00:56:37,470
 Have one core relay for the island, and then it just FIPS over radio to your entire island,

906
00:56:37,470 --> 00:56:38,470
 or whatever, right?

907
00:56:38,470 --> 00:56:43,750
 Like, I love the versatility, the fact that it's amorphous, right?

908
00:56:43,750 --> 00:56:46,630
 It can change shapes and go in every direction it needs to go.

909
00:56:46,630 --> 00:56:48,390
 It's kind of a bridge building technology.

910
00:56:48,390 --> 00:56:50,590
 It's like, you can bridge anything with anything.

911
00:56:50,590 --> 00:56:54,990
 The archipelago guys, what they showed off, making the archipelago nodes that they're

912
00:56:54,990 --> 00:57:01,070
 building FIPS compatible, that was so amazing to see that they managed to pull it off.

913
00:57:01,070 --> 00:57:05,150
 And also, like, one of the side effects is like, you don't have to care about, you know,

914
00:57:05,150 --> 00:57:09,550
 like hole punching, or any net stuff, or any router configuration to make it connect to

915
00:57:09,550 --> 00:57:11,030
 this IPs and so on.

916
00:57:11,030 --> 00:57:12,590
 It just works.

917
00:57:12,590 --> 00:57:13,590
 It just works.

918
00:57:13,590 --> 00:57:15,990
 You just, you punch in the end pump, that's it.

919
00:57:15,990 --> 00:57:19,230
 You even have a discovery layer, you know, like, you can discover the end pump and so

920
00:57:19,230 --> 00:57:20,230
 on.

921
00:57:20,230 --> 00:57:21,230
 And then it just works.

922
00:57:21,230 --> 00:57:22,230
 Yeah.

923
00:57:22,230 --> 00:57:23,230
 Yeah, it's quite amazing.

924
00:57:23,230 --> 00:57:26,190
 No, so, I mean, you know, you asked me, like, what my overall, like, first week impression

925
00:57:26,190 --> 00:57:27,190
 was.

926
00:57:27,190 --> 00:57:31,390
 I think, like, I am beyond, like, I could not be more bullish about the next couple of

927
00:57:31,390 --> 00:57:33,950
 weeks than just, like, the things we can do on nonster.

928
00:57:33,950 --> 00:57:37,150
 Like one thing that was amazing was, like, stuff like the Archipelago guys was like,

929
00:57:37,150 --> 00:57:40,190
 yeah, it was just, like, easy to integrate, and it made everything better, and it was

930
00:57:40,190 --> 00:57:41,190
 all easier.

931
00:57:41,190 --> 00:57:44,470
 And you see this path, and I've seen this in a couple projects now where you just do

932
00:57:44,470 --> 00:57:46,150
 the, Buzz would be a good example, right?

933
00:57:46,150 --> 00:57:49,990
 Where you take the nonster ideas, and it turns out that they're very simple to integrate

934
00:57:49,990 --> 00:57:55,070
 and give you so much for free that all you have to do is step away from your old way

935
00:57:55,070 --> 00:58:00,230
 of thinking about application building, about the web, about centralized services, SaaS,

936
00:58:00,230 --> 00:58:01,230
 etc.

937
00:58:01,230 --> 00:58:02,230
 Yeah.

938
00:58:02,230 --> 00:58:05,550
 And then you start bubbling up that are so, like, within reach, and make your product

939
00:58:05,550 --> 00:58:06,710
 so much better, right?

940
00:58:06,710 --> 00:58:13,070
 And that's been, that I didn't think about before, but yeah, very bullish on that front,

941
00:58:13,070 --> 00:58:14,070
 right?

942
00:58:14,070 --> 00:58:15,070
 Yeah.

943
00:58:15,070 --> 00:58:16,070
 Did you play around with Buzz already?

944
00:58:16,070 --> 00:58:19,630
 Did you give it a go?

945
00:58:19,630 --> 00:58:25,030
 So I've been, I put NixOS on my laptop, and so I suffer with any package building, and

946
00:58:25,030 --> 00:58:29,110
 I've been trying to build Buzz for a better part of the past week, unsuccessfully.

947
00:58:29,110 --> 00:58:34,910
 But I did learn a lot about half-image packaging.

948
00:58:34,910 --> 00:58:36,990
 So, no, I haven't gotten past the interface.

949
00:58:36,990 --> 00:58:41,510
 The thing that was weird with Buzz was, like, it forces you to get either codecs or, yeah,

950
00:58:41,510 --> 00:58:45,350
 or a clod, and yeah, I didn't really love that.

951
00:58:45,350 --> 00:58:46,350
 Yeah.

952
00:58:46,350 --> 00:58:48,150
 So I don't know why I could just use Goos or something.

953
00:58:48,150 --> 00:58:49,150
 Yeah.

954
00:58:49,150 --> 00:58:50,150
 Yeah.

955
00:58:50,150 --> 00:58:53,150
 You're a kind of maple-only kind of guy, what I heard, right?

956
00:58:53,150 --> 00:58:54,150
 Yeah.

957
00:58:54,150 --> 00:58:55,150
 Is that correct?

958
00:58:55,150 --> 00:58:57,750
 I try to use maple, AI, shout out to Tony.

959
00:58:57,750 --> 00:58:58,750
 Yeah.

960
00:58:58,750 --> 00:58:59,750
 Yeah.

961
00:58:59,750 --> 00:59:03,550
 Because I like it, and it's a great service, and that's how I want to use LLMs, is privately.

962
00:59:03,550 --> 00:59:04,550
 Yeah.

963
00:59:04,550 --> 00:59:05,550
 Yeah.

964
00:59:05,550 --> 00:59:06,550
 Yeah.

965
00:59:06,550 --> 00:59:08,870
 You run local models as well, or you have any type of hardware to speak of?

966
00:59:08,870 --> 00:59:13,030
 No, I don't have very serious hardware, so that's why I'm kind of stuck with the services.

967
00:59:13,030 --> 00:59:14,030
 Yeah.

968
00:59:14,030 --> 00:59:15,230
 That's why maple's the right fit, right?

969
00:59:15,230 --> 00:59:17,030
 You plan on getting one?

970
00:59:17,030 --> 00:59:18,030
 Nah.

971
00:59:18,030 --> 00:59:19,030
 Post-hype Bitcoinization?

972
00:59:19,030 --> 00:59:21,030
 So, I mean, yeah, right.

973
00:59:21,030 --> 00:59:23,030
 Like, right.

974
00:59:23,030 --> 00:59:27,510
 I kind of like, I mean, the thing that's interesting with this whole AI thing is, like, are we going

975
00:59:27,510 --> 00:59:30,950
 to wake up a year from now and all of a sudden there's a model you can run that's very small

976
00:59:30,950 --> 00:59:33,070
 and focused exactly for the task you need?

977
00:59:33,070 --> 00:59:35,310
 You know, like, are we going to wake up and all of a sudden the GPUs didn't actually matter

978
00:59:35,310 --> 00:59:36,310
 that much?

979
00:59:36,310 --> 00:59:37,310
 Mm-hmm.

980
00:59:37,310 --> 00:59:41,830
 Or are we going to have to keep gobbling up hardware until 2030 to run the next and best

981
00:59:41,830 --> 00:59:42,830
 thing?

982
00:59:42,830 --> 00:59:44,710
 Or is it always going to be just in the cloud because no one's going to have a server rack

983
00:59:44,710 --> 00:59:45,710
 in their closet?

984
00:59:45,710 --> 00:59:49,710
 Like, I don't really know where it's going, so I'm not going to invest in, like, hardware,

985
00:59:49,710 --> 00:59:50,710
 you know?

986
00:59:50,710 --> 00:59:51,710
 Yeah.

987
00:59:51,710 --> 00:59:52,710
 I don't know yet.

988
00:59:52,710 --> 00:59:53,710
 Yeah.

989
00:59:53,710 --> 00:59:58,430
 Yeah, the future depends on open source local models.

990
00:59:58,430 --> 01:00:00,030
 I think we talked about this somewhere.

991
01:00:00,030 --> 01:00:27,021
 I think that's where things are heading. I know very little about this stuff, but I obviously talk to people that know a lot about this stuff.

992
01:00:27,021 --> 01:00:38,021
 And the kind of research consensus seems to be models get smarter and smaller and catch up faster to frontier than most people would like to think.

993
01:00:38,021 --> 01:00:43,021
 And I think also the specialization aspect that you mentioned is going to be big going forward.

994
01:00:43,021 --> 01:00:54,021
 Like quoting John Verweke and everything we know about cognitive science is like you can't be a specialist and a generalist at the same time. So we will need to have specialized models.

995
01:00:54,021 --> 01:01:05,021
 And the stuff you talked about obviously looks like highly specialized, like an elixir specialist, you know, like a rust specialist or like even more stupid than that, like, I don't know.

996
01:01:05,021 --> 01:01:16,021
 Only HTML. Yeah, diff merging, code conflict resolution specialist and those kind of things. Yeah. And people will definitely build this and also like distill bigger models into these specialists and so on.

997
01:01:16,021 --> 01:01:23,021
 So that's definitely happening. I mean, is it just that I don't know the like cutting edge LLM training subculture of the Internet and they're all doing this?

998
01:01:23,021 --> 01:01:28,021
 Like, I feel like someone must have done similar things, right? Yeah, I know too little about it myself.

999
01:01:28,021 --> 01:01:32,021
 I'm sure there's people out there that did the first experiments along these lines.

1000
01:01:32,021 --> 01:01:38,021
 You know, there was an interesting moment. Jesus, my sense of time has gone out the window with this fucking LLM stuff.

1001
01:01:38,021 --> 01:01:48,021
 But I want to say a year ago where like I think it was AMP, AMP code. They're like they're kind of a harness for the UI for coding with agents.

1002
01:01:48,021 --> 01:01:59,021
 And they were like, you know what, we're removing all tool calling because soon we see a future where the models will be so good that we won't need to do any like actual tool calling engineering. Interesting.

1003
01:01:59,021 --> 01:02:04,021
 And then like a month later, the exact opposite thing was the status quo in the world.

1004
01:02:04,021 --> 01:02:09,021
 Where everyone was like, you know what, actually the dumbest fucking models, as long as they're good at tool calling, will outperform anything.

1005
01:02:09,021 --> 01:02:18,021
 Because what we really want for software engineering isn't like a hyper super, you know, independent model that understands everything about everything.

1006
01:02:18,021 --> 01:02:25,021
 We want someone that is really good at using the tools at its disposal. Picking the right tool for the right job. Yeah.

1007
01:02:25,021 --> 01:02:35,021
 So I actually I think that's the right thesis. And I think that's why we're seeing all this like agentic engineering stuff take off because that's why I think anthropic.

1008
01:02:35,021 --> 01:02:42,021
 I'm super bearish and anthropic for lots of reasons. But like one of them is that they their main thesis is that the best model will win.

1009
01:02:42,021 --> 01:02:49,021
 There will be one model to rule them. Yeah, that's what I disagree with. Obviously, as well. No, no, no, terrible strategy. Right. Yeah.

1010
01:02:49,021 --> 01:03:00,021
 So so I don't know, I would like to think of the future is like models that are very good at tool calling and managing their environments and can be like their actual size is really, really small.

1011
01:03:00,021 --> 01:03:08,021
 And you need very little memory. Yeah. Yeah. But again, man, I don't know. Yeah, no, we'll see. We'll see. I mean, I think stuff.

1012
01:03:08,021 --> 01:03:18,021
 Well, I mean, Apple is doing that as well. They are ahead of the curve in some sense that they have all the hardware, you know, and like the chips in the iPhones are insanely capable.

1013
01:03:18,021 --> 01:03:27,021
 And so we will have locally I that's where it's trending. Basically, we'll have lots of stuff run locally on on on the supercomputers we have in our pockets and so on.

1014
01:03:27,021 --> 01:03:37,021
 And I don't rule it out that open source models will get good enough and small enough that you can just run it on relatively simple consumer consumer hardware.

1015
01:03:37,021 --> 01:03:55,021
 We are at a strange time right now where just hardware prices are all over the place. And just, you know, if you want to buy something and run your own inference like strong models, you'll have to buy the equivalent of like a medium sized car.

1016
01:03:55,021 --> 01:04:06,021
 I wouldn't like I wouldn't underestimate the hackers, right? They come up with ways to do things with with models that are never like never bet against human ingenuity. Like that's that's yeah.

1017
01:04:06,021 --> 01:04:13,021
 And isn't I know it's kind of like, you know, we have lots of who we're gonna know where this is going to go. But isn't it so fun? Like, I feel like in the next couple of years, it's a lot of fun.

1018
01:04:13,021 --> 01:04:24,021
 Yeah, no, it's a lot of fun. What a time. Yeah, yeah. I have one more one more thing to pick your brain about. You mentioned it already briefly, but it relates to what you demoed this Friday.

1019
01:04:24,021 --> 01:04:38,021
 So you packaged a bunch of nasty stuff for next. So tell me all about this. And what's your what's your plan for that? And and and yeah, how do we get Justin Moon aware of it as fast as possible?

1020
01:04:38,021 --> 01:04:57,021
 So, yeah, I started packaging nosterer software with Nix, right? And so I presented basically, like, and I packaged a bunch of stuff, and I threw it onto a Nix binary cache. So if you use Nix as a package manager, you can just use this binary cache to give you binaries, right?

1021
01:04:57,021 --> 01:04:59,961
 So, you say, yeah, I want to go, you know, I want to build.

1022
01:05:00,010 --> 01:05:06,010
 The developer environment that uses NAC or Engit or FIPS, you know, or all these things,

1023
01:05:06,010 --> 01:05:10,010
 it'll just check the binary cache first and say if it's there, I'll just pull it down, right?

1024
01:05:10,010 --> 01:05:13,010
 So, I mean, we're all used to binary caches as software engineers.

1025
01:05:13,010 --> 01:05:23,010
 So the reason to do this, like, my motivation for doing it was that, like, software packages are the bedrock

1026
01:05:23,010 --> 01:05:26,010
 of how we build all the things we build, right?

1027
01:05:26,010 --> 01:05:33,010
 Bitcoin uses other people's software, we ship software to others, and so we need a transparent tool

1028
01:05:33,010 --> 01:05:37,010
 to be able to manage both, like, publishing packages and downloading packages from the web

1029
01:05:37,010 --> 01:05:41,010
 and make it as easy as possible to use tools, software tools.

1030
01:05:41,010 --> 01:05:44,010
 And to me, the answer to that is NiX, right?

1031
01:05:44,010 --> 01:05:49,010
 You know, I've seen the good and the bad side of NiX on by no means like a NiX zealot,

1032
01:05:49,010 --> 01:05:53,010
 but I think there's only one tool that's a true universal package manager

1033
01:05:53,010 --> 01:06:01,010
 that can handle any language, any output build system, or output target, I mean,

1034
01:06:01,010 --> 01:06:04,010
 and have, like, one language and system to handle all that.

1035
01:06:04,010 --> 01:06:07,010
 And I think that's, like, that's a necessary...

1036
01:06:07,010 --> 01:06:09,010
 I see it as, like, sovereign infrastructure, right?

1037
01:06:09,010 --> 01:06:13,010
 How come you got into NiX in the first place?

1038
01:06:13,010 --> 01:06:21,010
 But it feels like it's your first love affair, something like that.

1039
01:06:21,010 --> 01:06:26,010
 So, I was... I mean, I've been coding in Elixir for a bunch of years,

1040
01:06:26,010 --> 01:06:30,010
 and so it looks just a functional language, and I've been looking at, like...

1041
01:06:30,010 --> 01:06:33,010
 I honestly kind of stumbled upon it, and I had a couple of...

1042
01:06:33,010 --> 01:06:39,010
 I guess I just, like, went reading about it, and I was like, this looks like a total mess,

1043
01:06:39,010 --> 01:06:41,010
 and I definitely don't want to get nerd sniped by this,

1044
01:06:41,010 --> 01:06:45,010
 because this feels like this is going to suck in years of my life, and no thank you.

1045
01:06:45,010 --> 01:06:46,010
 So you were right about that, probably.

1046
01:06:46,010 --> 01:06:51,010
 So I was right, and, yeah, it did take a while to just be with NiX,

1047
01:06:51,010 --> 01:06:55,010
 and I wish I had waited for LLMs to just teach me, but whatever.

1048
01:06:55,010 --> 01:07:00,010
 And, yeah, I mean, there was just something that the creator of NiX saw

1049
01:07:00,010 --> 01:07:05,010
 about how to build software, like, a question about, really,

1050
01:07:05,010 --> 01:07:09,010
 what does it mean to build software, and what is the optimal way to do it,

1051
01:07:09,010 --> 01:07:11,010
 or what is at least a sound way to do it,

1052
01:07:11,010 --> 01:07:13,010
 better than all the other examples we have today,

1053
01:07:13,010 --> 01:07:15,010
 which are all kind of shitty for different ways,

1054
01:07:15,010 --> 01:07:17,010
 for different reasons.

1055
01:07:17,010 --> 01:07:21,010
 And so I read his... He had a PhD thesis, which isn't, like,

1056
01:07:21,010 --> 01:07:24,010
 it really isn't that long, or whatever, and he just exposes...

1057
01:07:24,010 --> 01:07:26,010
 - So readable PhD thesis? - Yeah, totally readable.

1058
01:07:26,010 --> 01:07:29,010
 - Those are few and far between. - There are few.

1059
01:07:29,010 --> 01:07:34,010
 And it just laid it out, and it was all, like, based on cryptographic identifiers,

1060
01:07:34,010 --> 01:07:38,010
 and, you know, having a graph and map of different...

1061
01:07:38,010 --> 01:07:42,010
 - Dependencies? - Yeah, that created a graph of dependencies.

1062
01:07:42,010 --> 01:07:45,010
 And so it was, you know, they used Merkle trees,

1063
01:07:45,010 --> 01:07:47,010
 they used all sorts of things that are really familiar to us.

1064
01:07:47,010 --> 01:07:49,010
 - All this stuff you already loved. - Exactly.

1065
01:07:49,010 --> 01:07:51,010
 And I was like, "Oh, well, like, I understand this pattern,

1066
01:07:51,010 --> 01:07:55,010
 and, like, I see how it fits, and, like, does it really work for building packages?"

1067
01:07:55,010 --> 01:07:57,010
 And it turns out it does.

1068
01:07:57,010 --> 01:07:59,010
 And you get some really nice things out of it.

1069
01:07:59,010 --> 01:08:03,010
 One of them is reproducibility, so you're able to, like...

1070
01:08:03,010 --> 01:08:06,010
 If I write it on my machine, and we have the same machine type,

1071
01:08:06,010 --> 01:08:10,010
 and you write it on your machine, if you did it correctly,

1072
01:08:10,010 --> 01:08:13,010
 then we'll get the exact same binary bits a bit, right?

1073
01:08:13,010 --> 01:08:18,010
 And that's really hard to do on most other systems or other build systems.

1074
01:08:18,010 --> 01:08:20,010
 And Nix has, like, a way to do that.

1075
01:08:20,010 --> 01:08:22,010
 And that's important for certain trust assumptions,

1076
01:08:22,010 --> 01:08:25,010
 so you want to know that the stuff you're executing,

1077
01:08:25,010 --> 01:08:27,010
 there is no funny business going on,

1078
01:08:27,010 --> 01:08:30,010
 there is no man-in-the-middle injecting stuff into your software, and so on.

1079
01:08:30,010 --> 01:08:34,010
 - Yeah, exactly. - And that's one way to make sure that...

1080
01:08:34,010 --> 01:08:36,010
 And, you know, it should be how all software works.

1081
01:08:36,010 --> 01:08:41,010
 I know that the fact that we live without this is kind of like the Wild West...

1082
01:08:41,010 --> 01:08:43,010
 The world has been running on the YOLO mode for a long time.

1083
01:08:43,010 --> 01:08:45,010
 Exactly.

1084
01:08:45,010 --> 01:08:49,010
 And so, you know, with Nix, I found a little, like, island of sanity

1085
01:08:49,010 --> 01:08:53,010
 in terms of how to distribute and think about build systems and building packages.

1086
01:08:53,010 --> 01:08:56,010
 And, you know, the Nix community took this idea and went very far with it, right?

1087
01:08:56,010 --> 01:08:59,010
 It went from a package manager to, actually, you know what?

1088
01:08:59,010 --> 01:09:01,010
 We can use the same theory to build operating systems,

1089
01:09:01,010 --> 01:09:04,010
 so there's, like, a Linux distribution called Nix OS.

1090
01:09:04,010 --> 01:09:07,010
 We can actually use this to be able to deploy management systems

1091
01:09:07,010 --> 01:09:11,010
 that are incredibly powerful and useful because they don't have to recompute everything every time

1092
01:09:11,010 --> 01:09:14,010
 and they can see transparently what's happening on both machines, blah, blah, blah.

1093
01:09:14,010 --> 01:09:16,010
 All sorts of stuff.

1094
01:09:16,010 --> 01:09:19,010
 And so, yeah, I just became comfortable with it.

1095
01:09:19,010 --> 01:09:21,010
 I mean, I worked for Nix Company at some point.

1096
01:09:21,010 --> 01:09:23,010
 I worked for the Nix OS Foundation briefly.

1097
01:09:23,010 --> 01:09:24,010
 Oh, wow.

1098
01:09:24,010 --> 01:09:27,010
 Yeah, I just... You know, and then when I came into Bitcoin...

1099
01:09:27,010 --> 01:09:31,010
 I mean, I'd heard about this project, but Jonas Nick and Eric Arvstadt built Nix Bitcoin,

1100
01:09:31,010 --> 01:09:38,010
 which was like a whole Bitcoin deploy thing with lots of Bitcoin stuff packaged.

1101
01:09:38,010 --> 01:09:40,010
 So that was kind of my inspiration for Nostra package.

1102
01:09:40,010 --> 01:09:43,010
 It was like, "Well, it'd be cool if we had a similar thing of, like..."

1103
01:09:43,010 --> 01:09:44,010
 Nix Nostra.

1104
01:09:44,010 --> 01:09:46,010
 Yeah, Nix Nostra. Exactly.

1105
01:09:46,010 --> 01:09:47,010
 Where, like, stuff is packaged.

1106
01:09:47,010 --> 01:09:48,010
 Let's go.

1107
01:09:48,010 --> 01:09:49,010
 There's recipes to deploy it.

1108
01:09:49,010 --> 01:09:50,010
 Awesome.

1109
01:09:50,010 --> 01:09:51,010
 You can deploy your own nodes.

1110
01:09:51,010 --> 01:09:55,010
 So, like, what I'm going to work on next is, like, that deploy story stuff.

1111
01:09:55,010 --> 01:09:59,010
 Like, can we make a really easy, reproducible way of shipping a FIPS node

1112
01:09:59,010 --> 01:10:01,590
 with a

1113
01:10:01,590 --> 01:10:07,161
 And get GRASP server with, you know, a blossom drive with cashew mints.

1114
01:10:07,161 --> 01:10:12,681
 You know, can you compose these services and ship them composably mix and match etc.

1115
01:10:12,681 --> 01:10:13,841
 Yeah, awesome.

1116
01:10:13,841 --> 01:10:19,581
 And so like the whole goal with that is to make it easy and safe to deploy your own machines.

1117
01:10:19,581 --> 01:10:21,261
 It's also very agent friendly, right?

1118
01:10:21,261 --> 01:10:24,861
 Like agents are very good at deploying nix stuff.

1119
01:10:24,861 --> 01:10:30,861
 So if you have an infrastructure agent and, you know, you just like you just said, you

1120
01:10:30,861 --> 01:10:35,081
 know, like, okay, like, I want from this, I want the blossom server and I want the cashew

1121
01:10:35,081 --> 01:10:39,181
 mint and maybe I want the relay, just deploy it and it will just go ahead and do everything.

1122
01:10:39,181 --> 01:10:40,181
 Definitely.

1123
01:10:40,181 --> 01:10:44,541
 Like the fact that nix is like a written configuration means the clankers are just really good at

1124
01:10:44,541 --> 01:10:45,541
 it.

1125
01:10:45,541 --> 01:10:46,541
 It's all text and they love text.

1126
01:10:46,541 --> 01:10:47,541
 Yeah.

1127
01:10:47,541 --> 01:10:48,541
 Fantastic.

1128
01:10:48,541 --> 01:10:49,541
 Yeah.

1129
01:10:49,541 --> 01:10:54,661
 So yeah, it's been, it's been also another happenstance of LLMs that they are, they are

1130
01:10:54,661 --> 01:10:56,541
 quite in love with nix.

1131
01:10:56,541 --> 01:10:57,541
 Yeah.

1132
01:10:57,541 --> 01:10:59,661
 Match made in heaven.

1133
01:10:59,661 --> 01:11:02,821
 You might know a little bit about geeks as well.

1134
01:11:02,821 --> 01:11:09,621
 Since we spoke about building Bitcoin and I always thought that was interesting because

1135
01:11:09,621 --> 01:11:12,061
 that's a very hard problem to solve as well.

1136
01:11:12,061 --> 01:11:18,501
 How can you start with as little dependencies as possible and, and, and, you know, like build,

1137
01:11:18,501 --> 01:11:25,221
 yeah, bootstrap the software into existence, like with every step, making sure that there's

1138
01:11:25,221 --> 01:11:26,821
 no funny business going on.

1139
01:11:26,821 --> 01:11:28,661
 Like can you talk to me about that a little bit?

1140
01:11:28,661 --> 01:11:30,581
 How much do you know about geeks?

1141
01:11:30,581 --> 01:11:31,581
 Uh, g-u-i-x.

1142
01:11:31,581 --> 01:11:32,581
 Yeah.

1143
01:11:32,581 --> 01:11:33,581
 G-u-x.

1144
01:11:33,581 --> 01:11:36,701
 I mean, conceptually, I don't know much about the internals apart from like they forked off

1145
01:11:36,701 --> 01:11:40,501
 of nix super early and did their own kind of, they took that same idea I was talking about

1146
01:11:40,501 --> 01:11:43,941
 at nix and just implemented it very differently.

1147
01:11:43,941 --> 01:11:44,941
 And it's what we use in core.

1148
01:11:44,941 --> 01:11:45,941
 Right.

1149
01:11:45,941 --> 01:11:46,941
 Yeah.

1150
01:11:46,941 --> 01:11:51,781
 And people will like rebuild core locally on their machine with geeks and they publish

1151
01:11:51,781 --> 01:11:54,261
 the hash to make sure they have the same one.

1152
01:11:54,261 --> 01:11:55,261
 Right.

1153
01:11:55,261 --> 01:11:56,261
 Yeah.

1154
01:11:56,261 --> 01:11:58,741
 And so, you know, what's really interesting is, uh, so that works great.

1155
01:11:58,741 --> 01:12:02,421
 And like, you know, people use geeks, you could go verify the attestations and all that.

1156
01:12:02,421 --> 01:12:07,601
 But recently, uh, oxb10c, uh, went and did the same thing with nix because we always

1157
01:12:07,601 --> 01:12:10,341
 thought that it'd be possible to do this thing with nix.

1158
01:12:10,341 --> 01:12:14,941
 Um, and he was like, can I actually find the geeks hash with nix?

1159
01:12:14,941 --> 01:12:15,941
 Okay.

1160
01:12:15,941 --> 01:12:21,261
 Like, can I do this cross, uh, cross tool, come up with the same output for core and

1161
01:12:21,261 --> 01:12:22,261
 the mad lad did it.

1162
01:12:22,261 --> 01:12:23,261
 Uh, so he figured out...

1163
01:12:23,261 --> 01:12:24,261
 Mad lad is right.

1164
01:12:24,261 --> 01:12:25,261
 Yeah, dude.

1165
01:12:25,261 --> 01:12:27,141
 He's such a mad lad.

1166
01:12:27,141 --> 01:12:31,861
 And uh, so I think now we have it for all builds for macOS or nix, where you can actually

1167
01:12:31,861 --> 01:12:37,021
 just, yeah, you get a correct reproducible build, whether you use nix or geeks, nix or

1168
01:12:37,021 --> 01:12:38,021
 geeks.

1169
01:12:38,021 --> 01:12:39,021
 Awesome.

1170
01:12:39,021 --> 01:12:40,021
 Yeah.

1171
01:12:40,021 --> 01:12:41,021
 He's just insane.

1172
01:12:41,021 --> 01:12:46,581
 And stuff he does in the Bitcoin world, it's like, um, yeah.

1173
01:12:46,581 --> 01:12:53,841
 To me, it's always fascinating that those type of people even exist, you know, just

1174
01:12:53,841 --> 01:12:56,901
 doing hardcore stuff and being so consistent for so long.

1175
01:12:56,901 --> 01:12:57,901
 Yeah.

1176
01:12:57,901 --> 01:13:01,741
 And, and he's also, I think he's like picked up the LLMs and he's kind of just having fun.

1177
01:13:01,741 --> 01:13:03,101
 He's like, yeah, man, I just threw cloth on it.

1178
01:13:03,101 --> 01:13:04,501
 It was like, go figure it out.

1179
01:13:04,501 --> 01:13:09,301
 So it's more like about having all these great ideas and just, uh, yeah, yeah, like knowing

1180
01:13:09,301 --> 01:13:11,661
 how the internals work of all these systems.

1181
01:13:11,661 --> 01:13:12,661
 Yeah.

1182
01:13:12,661 --> 01:13:15,221
 Uh, no, God bless them for sure.

1183
01:13:15,221 --> 01:13:16,221
 Yeah.

1184
01:13:16,221 --> 01:13:17,901
 Um, yeah.

1185
01:13:17,901 --> 01:13:23,141
 So when do you think, uh, will, uh, nix noster be a thing?

1186
01:13:23,141 --> 01:13:26,421
 When can we send the link to Justin and he can install all this stuff?

1187
01:13:26,421 --> 01:13:30,501
 Um, well, I think I'm going to go ham on it next week.

1188
01:13:30,501 --> 01:13:31,501
 All right.

1189
01:13:31,501 --> 01:13:33,781
 Um, and I have the plans for it.

1190
01:13:33,781 --> 01:13:39,541
 So there's going to be quite a bit of kind of drudgery work and testing around the config

1191
01:13:39,541 --> 01:13:43,781
 stuff, but conceptually, I don't see any big blockers.

1192
01:13:43,781 --> 01:13:44,781
 All right.

1193
01:13:44,781 --> 01:13:45,781
 Yeah.

1194
01:13:45,781 --> 01:13:46,781
 All right.

1195
01:13:46,781 --> 01:13:47,781
 Yeah.

1196
01:13:47,781 --> 01:13:50,541
 I was looking at the next Bitcoin code base again, because they have all this extra niceties

1197
01:13:50,541 --> 01:13:55,301
 around, uh, secret management and like making sure the configs don't conflict.

1198
01:13:55,301 --> 01:13:59,421
 And like, uh, there, there, there is a lot of, you have to dig into the specifics of

1199
01:13:59,421 --> 01:14:00,421
 each software.

1200
01:14:00,421 --> 01:14:01,421
 Yeah.

1201
01:14:01,421 --> 01:14:05,981
 I think I'm going to mess with the others in, and you have to be careful about stuff.

1202
01:14:05,981 --> 01:14:06,981
 Yeah.

1203
01:14:06,981 --> 01:14:07,981
 I see.

1204
01:14:07,981 --> 01:14:08,981
 I want to say I'm going to go see how far I get next week.

1205
01:14:08,981 --> 01:14:09,981
 Yeah.

1206
01:14:09,981 --> 01:14:10,981
 Um, but I'll definitely have a demo on that front.

1207
01:14:10,981 --> 01:14:11,981
 Cool.

1208
01:14:11,981 --> 01:14:17,381
 Speaking of demos, you saw your, you had your first demo day this Friday.

1209
01:14:17,381 --> 01:14:18,381
 Yeah.

1210
01:14:18,381 --> 01:14:19,381
 Yeah.

1211
01:14:19,381 --> 01:14:20,381
 Yesterday.

1212
01:14:20,381 --> 01:14:21,381
 Wow.

1213
01:14:21,381 --> 01:14:22,381
 It was only yesterday.

1214
01:14:22,381 --> 01:14:23,381
 It wasn't a long day again.

1215
01:14:23,381 --> 01:14:24,381
 Yeah.

1216
01:14:24,381 --> 01:14:25,381
 Yeah.

1217
01:14:25,381 --> 01:14:26,381
 Tell me about it.

1218
01:14:26,381 --> 01:14:28,821
 And, um, yeah, time, time kind of moves differently.

1219
01:14:28,821 --> 01:14:33,001
 Like Bitcoin time is different than, you know, like real world time.

1220
01:14:33,001 --> 01:14:37,601
 And you mentioned, you know, the, your, the whole AI LLM stuff again is like a different

1221
01:14:37,601 --> 01:14:38,981
 sense of time again.

1222
01:14:38,981 --> 01:14:43,621
 And every time I'm, I'm participating in the court, I lose all sense of time entirely.

1223
01:14:43,621 --> 01:14:46,021
 That's just how it goes.

1224
01:14:46,021 --> 01:14:52,581
 But, um, yeah, Friday demo day, what were some things that are still bouncing around

1225
01:14:52,581 --> 01:14:53,581
 in your mind?

1226
01:14:53,581 --> 01:14:59,021
 Some stuff that you thought was, you know, like interesting or, or funny or well done

1227
01:14:59,021 --> 01:15:00,021
 or an interesting idea.

1228
01:15:00,021 --> 01:15:01,870
 What's something that stood out to you?

1229
01:15:01,870 --> 01:15:04,410
 We saw 21 demos, so I have to ask.

1230
01:15:04,410 --> 01:15:06,410
 Yeah, yeah, yeah. I mean, like, the...

1231
01:15:07,610 --> 01:15:11,250
 I mean, there's stuff like FIPS over sound waves. It's like pure hacker, you know,

1232
01:15:11,250 --> 01:15:16,130
 we were at the bar just like masterminding all the possible different ways we could send information, you know,

1233
01:15:16,130 --> 01:15:19,530
 like, and to see fun little hacker projects like that is just...

1234
01:15:19,530 --> 01:15:22,290
 Someone on Austro was like, "Oh, okay, so just like dial-up."

1235
01:15:22,290 --> 01:15:23,010
 Yeah, that's hot.

1236
01:15:23,010 --> 01:15:26,290
 And cheese-arminty was like, "Yes, but slower."

1237
01:15:27,570 --> 01:15:32,570
 But look, you know, it's fun, right? It's like the creativity of hacking stuff together, that's the fun part.

1238
01:15:32,570 --> 01:15:36,850
 The fact that you can actually do it and you can, like, you're...

1239
01:15:36,850 --> 01:15:40,250
 The other participants in the FIPS network wouldn't even know.

1240
01:15:40,250 --> 01:15:40,730
 Exactly.

1241
01:15:40,730 --> 01:15:42,890
 That's amazing, you know? Like, that's amazing.

1242
01:15:42,890 --> 01:15:50,610
 The other... I mean, the one that really stuck with me and that hit me very deep in the soul was the Archipelago demo,

1243
01:15:52,290 --> 01:15:57,090
 which was the UI one where they showed you the thing for the nodes, you know that?

1244
01:15:57,090 --> 01:15:58,010
 Yeah, yeah.

1245
01:15:58,010 --> 01:16:09,970
 And so this isn't released yet, but the start-up screen of this, like, home node system from Archipelago

1246
01:16:09,970 --> 01:16:17,930
 is the first time I've seen aesthetic and design so badass and so beautiful

1247
01:16:17,930 --> 01:16:23,970
 that it really lives up to, like, the mission that it's trying to embody.

1248
01:16:23,970 --> 01:16:29,170
 Yeah. Like, I think we've failed so much at aesthetics in Bitcoin and design

1249
01:16:29,170 --> 01:16:35,010
 and really getting into people's, you know, grabbing them from the first time they look at the screen or something, you know?

1250
01:16:35,010 --> 01:16:37,850
 All our software is fine. I mean, it works. It's what you're used to.

1251
01:16:37,850 --> 01:16:39,010
 Yeah, it's not exciting.

1252
01:16:39,010 --> 01:16:39,650
 It's not exciting.

1253
01:16:39,650 --> 01:16:40,490
 It doesn't pull you in.

1254
01:16:40,490 --> 01:16:45,610
 This is the first time, and I won't spoil it because people are going to have the experience of loading it themselves.

1255
01:16:45,610 --> 01:16:46,930
 But, man, I can't...

1256
01:16:46,930 --> 01:16:52,610
 It's an experience. It's like you're sitting in the cinema at, like, the THX sound place or something. You're ready.

1257
01:16:52,610 --> 01:16:58,050
 And it was also something different, something I had never seen before, right?

1258
01:16:58,050 --> 01:17:01,690
 And that's cool. That's very cool.

1259
01:17:01,690 --> 01:17:05,250
 I told them six months ago already, "Please release it. It's so polished already.

1260
01:17:05,250 --> 01:17:10,010
 Like, please make it available to the public." And they're like, "I don't know. There's still some rough edges."

1261
01:17:10,010 --> 01:17:13,210
 So there are some perfectionists at work here.

1262
01:17:13,210 --> 01:17:14,970
 Well, does this mean, like, it's not, you know...

1263
01:17:14,970 --> 01:17:20,170
 They promised me now July 31st, right? I think is what they said.

1264
01:17:20,170 --> 01:17:21,130
 I thought you said October 31st.

1265
01:17:21,130 --> 01:17:23,730
 October 31st. God damn it. Rock pulled by a month again.

1266
01:17:23,730 --> 01:17:25,930
 But, OK, I'll take it. October 31st.

1267
01:17:25,930 --> 01:17:29,770
 Yeah. But, I mean, you know, I don't see this perfectionism, right?

1268
01:17:29,770 --> 01:17:34,850
 I mean, it's like it's obviously like really amazing top shelf work by far.

1269
01:17:34,850 --> 01:17:42,690
 But it's I think as engineers, we often think about the systems themselves and like the the techniques behind it and how it works.

1270
01:17:42,690 --> 01:17:45,370
 And, oh, look at it. Isn't it amazing that it works?

1271
01:17:45,370 --> 01:17:52,170
 But the rest of the world doesn't belong on to things or doesn't like appropriate things for themselves because of the rational merits of it.

1272
01:17:52,170 --> 01:17:53,370
 This is aesthetics.

1273
01:17:53,370 --> 01:18:00,290
 It's aesthetics. Yeah. And like, you know, we fail to communicate like the badassery of cyberpunk aesthetics.

1274
01:18:00,290 --> 01:18:05,530
 Yeah, like it's it's it's cyberpunk, self-sovereign aesthetics. It's with what Archipelago?

1275
01:18:05,530 --> 01:18:09,370
 Yeah, 100 percent. And it's one of the first times I've really seen it.

1276
01:18:09,370 --> 01:18:14,970
 Yeah. You know, I could honestly I don't know many good examples of cyberpunk aesthetics that were properly

1277
01:18:14,970 --> 01:18:18,890
 designed and communicated that well. Yeah. Yeah. Shout out to Zaza.

1278
01:18:18,890 --> 01:18:27,370
 He's the main designer behind this, like he makes all this work and pretty and like he's a he comes from the AI world.

1279
01:18:27,370 --> 01:18:43,970
 And he I think he mentioned he started with flash animations, you know, like those this era of the Internet where we saw more of these kind of flashy intro heavy websites and so on.

1280
01:18:43,970 --> 01:18:51,970
 So so that definitely comes through. Yeah, no, I'm I'm very excited to see it progress like I've known about the project for quite a while now.

1281
01:18:51,970 --> 01:18:55,970
 And it's amazing to me how well it works.

1282
01:18:55,970 --> 01:19:01,170
 And it's also like you said, you know, the design and the aesthetics of it is top notch.

1283
01:19:01,170 --> 01:19:06,370
 And I'm very much looking forward to it being out in the wild and other people like getting up and running with it.

1284
01:19:06,370 --> 01:19:12,370
 And we see some Archipelago notes pop up. And right now it's internal testing.

1285
01:19:12,370 --> 01:19:20,370
 I think they mentioned the cohort will get, you know, like exclusive access in terms of testing and just running it for ourselves.

1286
01:19:20,370 --> 01:19:25,370
 And yeah, a little bit over a month from now, it will be hopefully out in the world. Nice. Really cool.

1287
01:19:25,370 --> 01:19:30,770
 One of my favorite node projects. I don't even know about it until yesterday. Yeah, it's really cool.

1288
01:19:30,770 --> 01:19:40,570
 What else stood out to you? Let's see. What else? What else was what else happened yesterday?

1289
01:19:40,570 --> 01:19:45,770
 I can't keep up with Pablo, man. He demoed three projects in four minutes. It was crazy.

1290
01:19:45,770 --> 01:19:54,770
 Oh, well, this was pretty cool, actually, was so Pablo demoed this thing that was basically how to compose different filters together.

1291
01:19:54,770 --> 01:19:57,370
 Yeah. And so the end of the back end can compose them intelligently.

1292
01:19:57,370 --> 01:19:59,970
 So you basically it was a UI where you use you can.

1293
01:20:00,021 --> 01:20:08,021
 I want long-form nodes from this person, but also single nodes from this other person,

1294
01:20:08,021 --> 01:20:14,021
 and just throw them into one bucket, and then the event starts streaming to you in a given way.

1295
01:20:14,021 --> 01:20:18,021
 Long-form nodes from someone I follow, but it's blocked by VHF.

1296
01:20:18,021 --> 01:20:22,021
 Yeah, so really arbitrary composition of filters.

1297
01:20:22,021 --> 01:20:24,021
 And so it was drag and drop, right?

1298
01:20:24,021 --> 01:20:28,021
 I don't think it's the right UI for it, because I don't think anybody wants to sit there,

1299
01:20:28,021 --> 01:20:33,021
 but I can see it as a speech thing, right?

1300
01:20:33,021 --> 01:20:38,021
 And say, yeah, just give me, all right, hey, good morning, Nostr.

1301
01:20:38,021 --> 01:20:46,021
 Get me the last memes by Corn DeLorean, or whatever, but nothing, you know,

1302
01:20:46,021 --> 01:20:48,021
 only the things that got zapped, or whatever.

1303
01:20:48,021 --> 01:20:52,021
 And also give me anything about the World Cup, and anything happening on Madeira, boom.

1304
01:20:52,021 --> 01:20:58,021
 I think that'll be something that a speech UI like that, but you've got to design the backend

1305
01:20:58,021 --> 01:21:00,021
 to be intelligent enough to compose them and stream them back.

1306
01:21:00,021 --> 01:21:05,021
 Yeah, that's what he's building out, in the sense that it's infinitely composable.

1307
01:21:05,021 --> 01:21:12,021
 And depending on, I mean, I think you like it because you come from the functional programming side, right?

1308
01:21:12,021 --> 01:21:13,021
 Like a very similar idea.

1309
01:21:13,021 --> 01:21:18,021
 And then the output of the whole function is of a certain kind, let's say like a list of profiles,

1310
01:21:18,021 --> 01:21:20,021
 or a list of kind one nodes, or a list of whatever.

1311
01:21:20,021 --> 01:21:26,021
 And depending on the kind, the way you display it, it's going to be different.

1312
01:21:26,021 --> 01:21:29,021
 Like a list of images is just going to be a gallery, those kind of things.

1313
01:21:29,021 --> 01:21:30,021
 Cool.

1314
01:21:30,021 --> 01:21:36,021
 And the part that's tricky is you can also nest kinds, which means you have to nest representations,

1315
01:21:36,021 --> 01:21:40,021
 because you can have a kind one node that mentions an article, for example,

1316
01:21:40,021 --> 01:21:42,021
 or embeds an article, you know, and all those kind of things.

1317
01:21:42,021 --> 01:21:46,021
 And then it gets tricky, so that's what he's currently banging his head against,

1318
01:21:46,021 --> 01:21:50,021
 because then you need, dependent on the representation context,

1319
01:21:50,021 --> 01:21:53,021
 you need a different kind of representation for the thing.

1320
01:21:53,021 --> 01:21:56,021
 Like if I embed an article, you don't want to see the whole article,

1321
01:21:56,021 --> 01:21:58,021
 you might want to have the article name, you know?

1322
01:21:58,021 --> 01:21:59,021
 Yeah, yeah, yeah.

1323
01:21:59,021 --> 01:22:02,021
 Or how do you display a profile when it's just mentioned in a kind one,

1324
01:22:02,021 --> 01:22:07,021
 it's different to how you show a profile if it's part of a block list, for example, or whatever.

1325
01:22:07,021 --> 01:22:09,021
 Yeah, yeah, yeah.

1326
01:22:09,021 --> 01:22:10,021
 No, that's interesting for sure.

1327
01:22:10,021 --> 01:22:11,021
 Yeah, it's an interesting problem.

1328
01:22:11,021 --> 01:22:21,021
 But the whole system already works, like the composability, and the kind of like how basically,

1329
01:22:21,021 --> 01:22:27,021
 it's basically relay filters, but composed, and queries composed infinitely deep.

1330
01:22:27,021 --> 01:22:31,021
 And the cool thing you could do on top of that is like then start having arbitrary queries, right?

1331
01:22:31,021 --> 01:22:32,021
 Yeah.

1332
01:22:32,021 --> 01:22:36,021
 Say, okay, give me anything that mentions x, but nothing that mentions y.

1333
01:22:36,021 --> 01:22:37,021
 Yeah.

1334
01:22:37,021 --> 01:22:43,021
 And yeah, then you're running, not SQL, but you're running some kind of query on top of a composable query of the actual.

1335
01:22:43,021 --> 01:22:45,021
 Yeah, and it's extremely performant as well.

1336
01:22:45,021 --> 01:22:47,021
 Like it was quite impressive.

1337
01:22:47,021 --> 01:22:49,021
 Yeah, he's always cooking up something crazy.

1338
01:22:49,021 --> 01:22:51,021
 Let's see where that goes.

1339
01:22:51,021 --> 01:22:54,021
 And you know, it's a whole class of things I've always thought that's really well suited to Noster,

1340
01:22:54,021 --> 01:22:56,021
 because you can't query Twitter.

1341
01:22:56,021 --> 01:22:57,021
 Yeah, of course.

1342
01:22:57,021 --> 01:23:00,021
 Unless you pay for the API, like, you know, everybody's experience with Twitter would be so much better

1343
01:23:00,021 --> 01:23:03,021
 if you could just like compose your own UIs over it or whatever.

1344
01:23:03,021 --> 01:23:09,021
 And I think once there's a little bit, once people get that composability in the tools you can build on top of a Noster,

1345
01:23:09,021 --> 01:23:10,021
 I think will be a really big one.

1346
01:23:10,021 --> 01:23:11,021
 Yeah.

1347
01:23:11,021 --> 01:23:12,021
 Yeah.

1348
01:23:12,021 --> 01:23:13,021
 Yeah, I mean, it's already happening.

1349
01:23:13,021 --> 01:23:19,021
 Like the Noster, we have, you know, more Noster clients than users, basically.

1350
01:23:19,021 --> 01:23:21,021
 And some of them are really different.

1351
01:23:21,021 --> 01:23:23,021
 I think they're built a little too hard.

1352
01:23:23,021 --> 01:23:29,021
 And on the Twitter side, lots of people use Notec, you know, and lots of people use alternative Twitter clients.

1353
01:23:29,021 --> 01:23:33,021
 But unfortunately, of course, they shut all that down, basically, with restricting the API access.

1354
01:23:33,021 --> 01:23:35,021
 But that's all just an incentives game.

1355
01:23:35,021 --> 01:23:39,021
 You know, they need to squeeze the users, keep them in, make money somehow, and so on.

1356
01:23:39,021 --> 01:23:40,021
 Yeah, yeah.

1357
01:23:40,021 --> 01:23:41,021
 That's just how it goes.

1358
01:23:41,021 --> 01:23:45,021
 No, I think when finally people wander into the Noster tents, they'll be kind of like, you guys built what?

1359
01:23:45,021 --> 01:23:47,021
 Like, holy shit.

1360
01:23:47,021 --> 01:23:50,021
 Yeah, that'd be fun.

1361
01:23:50,021 --> 01:23:52,021
 Yeah, lots of things going on.

1362
01:23:52,021 --> 01:23:55,021
 What else happened yesterday that was cool?

1363
01:23:55,021 --> 01:23:56,021
 I feel bad now.

1364
01:23:56,021 --> 01:24:00,021
 I'm like blanking on all the awesome projects, but I mean, one that was cool was the data one.

1365
01:24:00,021 --> 01:24:07,021
 So why don't we have data sets as like blossom bits, and you can just query for data sets over Noster?

1366
01:24:07,021 --> 01:24:08,021
 Yeah.

1367
01:24:08,021 --> 01:24:09,021
 Which is a super obvious one.

1368
01:24:09,021 --> 01:24:13,021
 But I mean, there's some technical things you got to figure out with like, he used Parquet as a data format.

1369
01:24:13,021 --> 01:24:14,021
 Exactly.

1370
01:24:14,021 --> 01:24:15,021
 And how you compose chunks and all that stuff.

1371
01:24:15,021 --> 01:24:18,021
 But ultimately, like, I would love to have a public data repo.

1372
01:24:18,021 --> 01:24:24,021
 And it's just like stuff I can compose and query and cross-reference over Noster, like, it makes a whole sense.

1373
01:24:24,021 --> 01:24:26,021
 Yeah, that was Wispahash, that was his demo.

1374
01:24:26,021 --> 01:24:28,021
 And I spoke to him today about it, actually.

1375
01:24:28,021 --> 01:24:30,021
 And he'll get it done for sure.

1376
01:24:30,021 --> 01:24:33,021
 Like, he's highly motivated.

1377
01:24:33,021 --> 01:24:43,021
 And his main concern is he still has too many blind spots in terms of Noster technicalities, like how to structure the events right and so on.

1378
01:24:43,021 --> 01:24:46,021
 But I told him, like, you know, all the right people are here.

1379
01:24:46,021 --> 01:24:47,021
 Yeah, yeah.

1380
01:24:47,021 --> 01:24:48,021
 Talk to Hazard.

1381
01:24:48,021 --> 01:24:49,021
 Talk to Pablo.

1382
01:24:49,021 --> 01:24:50,021
 It's like you're Gucci.

1383
01:24:50,021 --> 01:24:51,021
 I locked you guys in a room.

1384
01:24:51,021 --> 01:24:52,021
 Yeah, you're fine. You're fine.

1385
01:24:52,021 --> 01:24:53,021
 It's not an issue.

1386
01:24:53,021 --> 01:24:54,021
 Talk to your expert.

1387
01:24:54,021 --> 01:24:59,021
 Yeah, and that's a good example of someone who's, like, a data person and, like, a specialist.

1388
01:25:00,010 --> 01:25:14,730
 What surprised me was kind of the FIPS world map that was really really surprising to me.

1389
01:25:14,730 --> 01:25:23,130
 How could I forget? Dude this was mind-blowing. So this was basically like, I thought FIPS was

1390
01:25:23,130 --> 01:25:28,970
 like 12 nodes and there was what 2500? Yeah something like that way more way more than

1391
01:25:28,970 --> 01:25:32,570
 you would think. And so it was a live it was a live globe of like all the active FIPS connections.

1392
01:25:34,650 --> 01:25:38,250
 I think Arian must have just given a talk or something and like everyone has gone off and

1393
01:25:38,250 --> 01:25:46,410
 tried it because this thing just blew up like it's crazy. Crazy yeah it's I still I still don't know

1394
01:25:46,410 --> 01:25:52,170
 what happened or why there are so many nodes someone mentioned it's probably Marty Malmy's

1395
01:25:52,170 --> 01:25:59,210
 NostrVPN which basically bundles a FIPS node. No way he's got like even a thousand nodes. I don't

1396
01:25:59,210 --> 01:26:05,210
 know like I don't have a working theory for why there are so many there are way more than I would

1397
01:26:05,210 --> 01:26:10,330
 have expected. I think Jonathan might be unhappy about this because the last time I spoke to him

1398
01:26:10,330 --> 01:26:14,250
 he mentioned like there's way too many people using this thing it's not ready everything will break

1399
01:26:15,930 --> 01:26:20,650
 and it feels like it just gets worse by the minute in terms of like people are actually adopting it.

1400
01:26:20,650 --> 01:26:24,810
 I mean this is this will only spur him to death faster like I see no downsides here. Yeah

1401
01:26:27,050 --> 01:26:32,250
 and what else do we have we had some fun ones like the NostrTamagotchi thing that can potentially

1402
01:26:32,250 --> 01:26:39,530
 kill itself. Yeah I was like that's fine man but you know like I you know I mean we live in

1403
01:26:39,530 --> 01:26:46,650
 Europe and like I'm very you know like the France is like forcing social media for kids

1404
01:26:46,650 --> 01:26:53,450
 or doing identity verification and I'm like I want to build the like infra stuff that matters

1405
01:26:53,450 --> 01:26:59,450
 so that we escape we have all the tools escape the gulags right so you know all you know respect to

1406
01:27:00,330 --> 01:27:06,890
 whoever built that I yeah I won't be building Tamagotchis you know I think there's lots of

1407
01:27:06,890 --> 01:27:13,210
 lots of it would be a nice tagline like anti-gulag technology yeah oh yeah for sure well I'm so I'm

1408
01:27:13,210 --> 01:27:17,770
 so like I'm not disappointed I'm just like I can't believe it's a real AGI anti-gulag infrastructure

1409
01:27:17,770 --> 01:27:24,570
 yeah but like I want young hackers to come out and be like how do I escape this shit

1410
01:27:24,570 --> 01:27:28,250
 of the this country making all these stupid decisions about my use of the computer and

1411
01:27:28,250 --> 01:27:33,930
 social media yeah and I want them to like wander in and have like a vibrant tool set ideas and

1412
01:27:33,930 --> 01:27:39,850
 people building in it yeah and they see that it's not just like you know four hackers somewhere

1413
01:27:39,850 --> 01:27:45,690
 often wherever yeah that kind of got a side project that sort of works yeah so I'm kind

1414
01:27:45,690 --> 01:27:52,730
 of feeling the pressure to like run fast and and build build this stuff we need to build yeah yeah

1415
01:27:52,730 --> 01:27:59,050
 build those cape hatches yeah yeah what's your what's your take on the future in general like

1416
01:27:59,050 --> 01:28:07,130
 the next I don't know five years ten years speaking of European Union decisions and

1417
01:28:07,770 --> 01:28:13,530
 the digital gulags we find ourselves in um well it's never been a better time to be a cypherpone

1418
01:28:15,290 --> 01:28:19,210
 but I think I'll just quote Hunter S. Thompson when he said uh when the going gets weird the

1419
01:28:19,210 --> 01:28:25,610
 weird turn pro yeah yeah um I don't know I mean it's such a weird time to be alive and so many

1420
01:28:25,610 --> 01:28:30,570
 things are changing so quickly yeah that's kind of all I got yeah I'm along for the ride yeah

1421
01:28:30,570 --> 01:28:36,970
 you have kids yet no not yet no kids one and a half kids or no uh not sure yet we'll see

1422
01:28:38,410 --> 01:28:45,370
 no comment no comment because that's kind of it for me for me obviously

1423
01:28:46,090 --> 01:28:53,050
 supercharge is my kind of enthusiasm and motivation to actually get this right and

1424
01:28:54,250 --> 01:28:58,330
 build something that my kids can use that no one else can you know take away from them or

1425
01:28:58,330 --> 01:29:06,090
 shut down or whatever yeah and I grew up in a free and open internet mostly and I watched

1426
01:29:07,450 --> 01:29:15,050
 you know with dismay kind of things get really weird and the doors closing in and yeah just

1427
01:29:15,050 --> 01:29:22,330
 most things going to shit and I mean obviously everyone knows and everyone realizes

1428
01:29:23,530 --> 01:29:28,890
 kind of the conundrum we're in Cory Doctor for example wrote the book and shitification you know

1429
01:29:28,890 --> 01:29:34,170
 I think it's a nice term yeah like how every single online thing kind of slowly but surely

1430
01:29:34,170 --> 01:29:42,650
 turn to shit and yeah I don't want to see my kids grow up in the in the gulags digital or otherwise

1431
01:29:42,650 --> 01:29:49,370
 so yeah I'm highly motivated to keep working on this stuff and you know hopefully help it along

1432
01:29:49,370 --> 01:29:54,010
 a little bit and push things in the right direction a little bit so yeah I mean for

1433
01:29:54,010 --> 01:29:59,850
 me it's clear my kids won't have a bank account for example they're three and six now yeah so

1434
01:30:00,021 --> 01:30:06,021
 They're not going to need a bank account, especially my family, like the last couple of years.

1435
01:30:06,021 --> 01:30:09,021
 We moved to Madeira because there's a Bitcoin circular economy here, basically,

1436
01:30:09,021 --> 01:30:11,021
 and you can live on a Bitcoin standard.

1437
01:30:11,021 --> 01:30:17,021
 So, you know, like if push comes to shove, I still have a bank account, but I hardly use it.

1438
01:30:17,021 --> 01:30:22,021
 And I want to maneuver myself into a position where I can just always fully live off Bitcoin,

1439
01:30:22,021 --> 01:30:25,021
 and my kids will be by default in the same position, basically.

1440
01:30:25,021 --> 01:30:31,021
 And the same for, you know, whatever, communication, social media, and all those kind of things.

1441
01:30:31,021 --> 01:30:36,021
 Like the chat control that's coming, it just won't face me.

1442
01:30:36,021 --> 01:30:39,021
 And I had, you know, the talk with the wife already and so on.

1443
01:30:39,021 --> 01:30:44,021
 Her life is already complicated because, you know, my life is peculiar.

1444
01:30:44,021 --> 01:30:51,021
 But I just told her, just so you know, you know, like the EU just passed this thing and so on.

1445
01:30:51,021 --> 01:30:56,021
 So going forward very soon, you know, like the mother-in-law and father-in-law

1446
01:30:56,021 --> 01:31:00,021
 and everyone who still wants to get baby pictures and so on, and pictures of the kids,

1447
01:31:00,021 --> 01:31:04,021
 they'll have to install special software, just like prepare them for it, you know.

1448
01:31:04,021 --> 01:31:07,021
 It's going to be some nip.

1449
01:31:07,021 --> 01:31:12,021
 And that's just, you know, that's just what you have to do.

1450
01:31:12,021 --> 01:31:16,021
 Yeah, and you know, like, yeah, like Bitcoin's only one part of the puzzle, you know.

1451
01:31:16,021 --> 01:31:17,021
 Oh, absolutely.

1452
01:31:17,021 --> 01:31:21,021
 I feel like I'm very worried about the fact, I mean, not really, right?

1453
01:31:21,021 --> 01:31:24,021
 It's just that the rest of society truly is kind of oblivious to it.

1454
01:31:24,021 --> 01:31:26,021
 Like they're going to let chat control just happen, right?

1455
01:31:26,021 --> 01:31:29,021
 They're just going to be like, yeah, well, you know, shucks, what are we going to do?

1456
01:31:29,021 --> 01:31:33,021
 You know, I'm like, okay, well, if that's the level we're dealing with,

1457
01:31:33,021 --> 01:31:35,021
 it's going to be hard to get you to fight against anything else, you know.

1458
01:31:35,021 --> 01:31:37,021
 Yeah, yeah, yeah.

1459
01:31:37,021 --> 01:31:40,021
 So, yeah, who knows?

1460
01:31:40,021 --> 01:31:44,021
 Yeah, the general population doesn't have any type of revolution in them.

1461
01:31:44,021 --> 01:31:51,021
 I think the testosterone levels are too low for any kind of violent overthrow of the European government, for example.

1462
01:31:51,021 --> 01:31:54,021
 I'm just asking them to, you know, follow their own laws as they're stated,

1463
01:31:54,021 --> 01:31:57,021
 and they can't even do that, you know.

1464
01:31:57,021 --> 01:32:00,021
 No, it's a sad state of affairs.

1465
01:32:00,021 --> 01:32:03,021
 But I want, you know, it's like, that's why I want us to build this stuff

1466
01:32:03,021 --> 01:32:10,021
 and have it ready for the young people that come in trying to figure out how to navigate this world

1467
01:32:10,021 --> 01:32:13,021
 and find out that we have been thinking about this for a long time,

1468
01:32:13,021 --> 01:32:15,021
 and we've been through a lot of thinking and built a lot of things

1469
01:32:15,021 --> 01:32:21,021
 and a lot of tools that are available to you to navigate in a way that isn't insane.

1470
01:32:21,021 --> 01:32:27,021
 What do you think would be the best way to get the young people in and interested in this stuff?

1471
01:32:27,021 --> 01:32:29,021
 Build tools they love, you know.

1472
01:32:29,021 --> 01:32:32,021
 Like, that's why the archipelago thing, like I said, it hit me really hard

1473
01:32:32,021 --> 01:32:37,021
 because I know that if you were 16 years old and you hit, like, show me the badass messaging app

1474
01:32:37,021 --> 01:32:41,021
 or show me, you know, the tool that doesn't track me,

1475
01:32:41,021 --> 01:32:46,021
 if it's just another bland, like, Google material-themed, you know, Android application,

1476
01:32:46,021 --> 01:32:49,021
 they're going to just be like, whatever, you know.

1477
01:32:49,021 --> 01:32:54,021
 But if it's something that really grabs them in the first 0.3 seconds that they open the application

1478
01:32:54,021 --> 01:32:57,021
 and speaks to them directly, then you've hooked them, right?

1479
01:32:57,021 --> 01:33:00,021
 And then you can teach them about anything, right?

1480
01:33:00,021 --> 01:33:01,021
 Yeah.

1481
01:33:01,021 --> 01:33:02,021
 So I think it's...

1482
01:33:02,021 --> 01:33:03,021
 Yeah, that's a good point.

1483
01:33:03,021 --> 01:33:04,021
 That's a good point.

1484
01:33:04,021 --> 01:33:07,021
 For me, definitely, there was gaming and, like, certain types of games and so on.

1485
01:33:07,021 --> 01:33:11,021
 That's how I learned about networking and, like, you know, plugging computers together

1486
01:33:11,021 --> 01:33:13,021
 and all these kind of things.

1487
01:33:13,021 --> 01:33:17,021
 And, yeah, no, definitely getting the pull right.

1488
01:33:17,021 --> 01:33:19,021
 Yeah, that's a good point.

1489
01:33:19,021 --> 01:33:21,021
 I think a lot of them will also learn by pain.

1490
01:33:21,021 --> 01:33:26,021
 Like, again, just, you know, not being able to connect to their existing online communities

1491
01:33:26,021 --> 01:33:29,021
 because of, like, age verification checks and those kind of things.

1492
01:33:29,021 --> 01:33:32,021
 And then we'll kind of see what happens.

1493
01:33:32,021 --> 01:33:33,021
 But, yeah, I'm not too worried.

1494
01:33:33,021 --> 01:33:37,021
 Would you say you're optimistic or, like, where do you fall on the line

1495
01:33:37,021 --> 01:33:39,021
 between optimist, pessimist, realist?

1496
01:33:42,021 --> 01:33:48,021
 Yeah, I wouldn't say I'm optimistic, but I also don't think it really matters

1497
01:33:48,021 --> 01:33:50,021
 whether you're optimistic or pessimistic about it.

1498
01:33:50,021 --> 01:33:52,021
 Are you optimistic for yourself at least?

1499
01:33:52,021 --> 01:33:54,021
 For your future trajectory?

1500
01:33:54,021 --> 01:33:55,021
 I am.

1501
01:33:55,021 --> 01:33:57,021
 Not landing in the gulags?

1502
01:33:57,021 --> 01:34:00,021
 I mean, you know, that's a future I'm willing to accept.

1503
01:34:00,021 --> 01:34:03,021
 You know, I've chosen what I'm going to work on, and, you know,

1504
01:34:03,021 --> 01:34:05,021
 one day you might be branded a pariah for doing so.

1505
01:34:05,021 --> 01:34:09,021
 And I think anyone that works in Bitcoin, to some extent, has that talk with themselves.

1506
01:34:09,021 --> 01:34:11,021
 Yeah, sure.

1507
01:34:11,021 --> 01:34:16,021
 And you just got to live your life the way that you can feel good about.

1508
01:34:16,021 --> 01:34:17,021
 Yeah.

1509
01:34:17,021 --> 01:34:21,021
 And whether that means you get thrown in jail for using social media

1510
01:34:21,021 --> 01:34:26,021
 or using non-approved social media platforms, well, fine, whatever, you know.

1511
01:34:26,021 --> 01:34:28,021
 Well, sign notes from jail.

1512
01:34:28,021 --> 01:34:30,021
 Yeah, sign notes from jail, right?

1513
01:34:30,021 --> 01:34:32,021
 Whatever.

1514
01:34:32,021 --> 01:34:36,021
 But I hope that, you know, it's our job to kind of start building those things

1515
01:34:36,021 --> 01:34:38,021
 that when people come to need them, that they're there for them.

1516
01:34:38,021 --> 01:34:40,021
 Yeah, absolutely.

1517
01:34:40,021 --> 01:34:44,021
 And so I feel very lucky to be able to work on that.

1518
01:34:44,021 --> 01:34:46,021
 Awesome. Likewise.

1519
01:34:46,021 --> 01:34:48,021
 Likewise. That's a fantastic closing word.

1520
01:34:48,021 --> 01:34:50,021
 So let's keep it at that.

1521
01:34:50,021 --> 01:34:51,021
 GG. Thank you.

1522
01:34:51,021 --> 01:34:52,021
 Thanks so much.

1523
01:34:52,021 --> 01:34:54,021
 Thanks for making the time and indulging me.

1524
01:34:54,021 --> 01:34:56,021
 That was fantastic.

1525
01:34:56,021 --> 01:34:59,861
 i'm i'm i'm very tired now lots of luck