1
00:00:00,000 --> 00:00:12,720
Trust is a function. It is more than just a feeling. In a world of infinite noise,

2
00:00:12,720 --> 00:00:20,240
the only signal that matters is the one you verify. We are here to map the graph,

3
00:00:20,240 --> 00:00:25,920
order the protocol, and build the immune system for the decentralized web.

4
00:00:25,920 --> 00:00:29,760
Welcome to the signal processing layer.

5
00:00:30,860 --> 00:00:33,720
This is Say What?

6
00:00:41,780 --> 00:00:45,380
Welcome to the very first episode of Say What?

7
00:00:46,380 --> 00:00:53,020
For the last decade, we've been trapped in black boxes fed by algorithms optimized for addiction.

8
00:00:53,020 --> 00:01:01,200
We know the current model is broken, but the exit won't be to a new app or a new platform.

9
00:01:01,780 --> 00:01:08,220
That will be recreating the same problems of gatekeepers and high priests in a different venue.

10
00:01:08,780 --> 00:01:14,280
The exit will be to a new primitive on an open protocol.

11
00:01:14,640 --> 00:01:17,480
It is the web of trust on NOSTA.

12
00:01:18,480 --> 00:01:24,280
This is the signal processing layer of the pioneers and settlers who make up this network.

13
00:01:24,820 --> 00:01:34,560
We're here to dissect the graph, debate the NIPs, and engineer the systems that replace centralized control with cryptographic proof.

14
00:01:34,560 --> 00:01:42,400
For this inaugural deep dive, I wanted to talk to someone deep in the engine room of this new reality.

15
00:01:42,400 --> 00:01:52,560
Joining me in the graph today are three people literally writing the code, or in some cases, vibing it, for our future.

16
00:01:53,280 --> 00:02:04,800
Vitor Pamplona from Amethyst, a newly fiat job liberated Derek Ross from SoBox, and David Strahan from Nosfabrica.

17
00:02:05,120 --> 00:02:07,780
Welcome, gentlemen, to the first episode of Say What.

18
00:02:09,780 --> 00:02:11,020
Thanks for having me.

19
00:02:11,020 --> 00:02:12,660
It's great to be here.

20
00:02:41,020 --> 00:02:48,140
scores. So maybe why don't we take a step back, Vitar, and for folks who are unfamiliar,

21
00:02:49,020 --> 00:02:54,440
explain what a trusted assertion is, because it is part of a NIP that you offered,

22
00:02:54,860 --> 00:03:00,560
and then what integrating that into Amethyst actually means.

23
00:03:01,400 --> 00:03:08,880
Awesome. Yeah, so trusted assertions are a way for users to define who they trust,

24
00:03:08,880 --> 00:03:17,480
to run algorithms on Noster that just have more data than what the phone can download.

25
00:03:18,000 --> 00:03:18,860
So picture this.

26
00:03:18,920 --> 00:03:24,340
If you are processing, if you're trying to look for posts that talk about chess, for instance,

27
00:03:24,980 --> 00:03:30,200
and the phone won't be able to download everything to processes

28
00:03:30,200 --> 00:03:37,020
and then get you to the top 10 or top 20 best posts about chess on Noster.

29
00:03:37,020 --> 00:03:43,340
It's just impossible for the phone or any other Nostra client really to do on that client side.

30
00:03:44,060 --> 00:03:48,060
So for these types of activities that we need to process the entirety of Nostra,

31
00:03:48,180 --> 00:03:54,800
we need to go out there and look at the information that each relay has and download it and process it first

32
00:03:54,800 --> 00:03:59,660
and keep downloading, keep processing forever.

33
00:03:59,660 --> 00:04:14,900
For these types of activities, we are much better off helping users select, pick providers to do those services and then just give us the results back to the clients who clients can use as well.

34
00:04:14,900 --> 00:04:36,280
So in this particular activity is trusted assertions allows the user to pick a provider for scores for a provider that will look at all endpubs out there and will just look at what they post, look at what they talk about, look at who they mute or who they follow.

35
00:04:36,280 --> 00:04:41,760
and essentially converge from your point of view.

36
00:04:42,080 --> 00:04:44,940
So if I am choosing a provider, for instance,

37
00:04:45,580 --> 00:04:47,260
from Vitor's point of view,

38
00:04:47,660 --> 00:04:50,820
who are all the other people that Vitor should trust

39
00:04:50,820 --> 00:04:52,380
and by how much?

40
00:04:52,540 --> 00:04:55,480
So by how much part, it becomes a score.

41
00:04:55,900 --> 00:04:58,020
And you can actually score all the other keys

42
00:04:58,020 --> 00:05:02,040
based on what I like to see and what I don't like to see.

43
00:05:02,200 --> 00:05:04,800
And it's individual to my user.

44
00:05:04,800 --> 00:05:14,700
And if you help users to just pick these providers, then everything else kind of falls into place on Nostar because for clients, it's a lot easier to just download those results.

45
00:05:15,500 --> 00:05:22,660
If they already know that their user trusted this particular provider, you can just download and display it on the screen.

46
00:05:23,160 --> 00:05:31,600
So for Amethyst, that translates into scores being displayed into or in the bottom of each one of the profile pictures.

47
00:05:31,600 --> 00:05:36,920
and essentially says, you know what, for you that is using the app,

48
00:05:37,220 --> 00:05:41,980
that you have picked this other provider, this other provider is running an algorithm for you,

49
00:05:41,980 --> 00:05:48,660
you have chosen to give this particular score to everybody else, right?

50
00:05:48,760 --> 00:05:54,120
So, you know, what, say, Derek, to me, is an 85 score.

51
00:05:54,760 --> 00:05:56,800
David is a 100 score, right?

52
00:05:56,860 --> 00:06:01,580
So it's, to me, but, you know, for other people, this could be inverted.

53
00:06:01,600 --> 00:06:02,440
it could be flipped.

54
00:06:02,900 --> 00:06:06,760
So it's not really a centralized scoring provider per se.

55
00:06:06,760 --> 00:06:09,740
This allows users to pick and choose

56
00:06:09,740 --> 00:06:13,000
which algorithms they want to run those scores

57
00:06:13,000 --> 00:06:15,760
and then show that in their screens.

58
00:06:17,100 --> 00:06:21,500
So, Vitor, just to remove any ambiguity there might be,

59
00:06:22,440 --> 00:06:25,080
Amethyst did not calculate any scores, right?

60
00:06:25,180 --> 00:06:28,280
The trusted assertions, which is part of your update,

61
00:06:28,280 --> 00:06:31,800
was just a mechanism, was a template,

62
00:06:32,280 --> 00:06:36,660
which a user can then choose from a score provider of their choice

63
00:06:36,660 --> 00:06:39,800
to populate that.

64
00:06:39,960 --> 00:06:41,620
And that's what shows up on their profile.

65
00:06:41,840 --> 00:06:43,640
Amethyst is not assigning any scores.

66
00:06:43,860 --> 00:06:44,220
Correct.

67
00:06:44,380 --> 00:06:47,040
Not only assigning, but we are not even picking providers,

68
00:06:47,040 --> 00:06:48,960
not even showing providers,

69
00:06:49,080 --> 00:06:51,500
a list of providers on our own screens, for instance.

70
00:06:52,060 --> 00:06:57,120
We don't interfere in any form of that choice

71
00:06:57,120 --> 00:06:58,560
that the user has to make.

72
00:06:59,580 --> 00:07:04,820
And this is why so many people didn't see anything on our new version.

73
00:07:05,300 --> 00:07:10,920
It's because we're kind of forcing them to learn what Web of Trust is,

74
00:07:11,640 --> 00:07:15,520
figure out what the providers are today, which are very few,

75
00:07:15,940 --> 00:07:20,380
and then go to the provider's own website, make an account.

76
00:07:21,200 --> 00:07:25,020
And when you do so, the provider then creates this little event

77
00:07:25,020 --> 00:07:28,740
that essentially makes you trust

78
00:07:28,740 --> 00:07:30,720
or tells everybody else,

79
00:07:30,940 --> 00:07:32,320
all your clients per se,

80
00:07:32,960 --> 00:07:35,300
that you are trusting this provider

81
00:07:35,300 --> 00:07:37,340
to run your algorithm for you.

82
00:07:37,740 --> 00:07:39,500
So until you go through all those hoops

83
00:07:39,500 --> 00:07:42,060
and you understand which one you're picking

84
00:07:42,060 --> 00:07:44,560
and how far this goes,

85
00:07:45,480 --> 00:07:47,200
you don't even know anything.

86
00:07:48,660 --> 00:07:52,380
Nothing like this happens without user approval.

87
00:07:53,100 --> 00:07:54,360
And Ametist doesn't do anything

88
00:07:54,360 --> 00:07:56,040
before you do all those steps,

89
00:07:56,080 --> 00:07:58,280
including go to somebody else's website

90
00:07:58,280 --> 00:08:00,640
to sign up for their own services.

91
00:08:02,680 --> 00:08:07,900
Derek, you are a skeptic, I think, or maybe not.

92
00:08:08,460 --> 00:08:11,680
But I know you have some tough questions.

93
00:08:12,480 --> 00:08:14,400
So based on what Vitor said,

94
00:08:14,420 --> 00:08:18,000
and then we'll pull David into the conversation, Derek.

95
00:08:18,000 --> 00:08:21,280
And I believe, Derek, you're an Amethyst user as well.

96
00:08:21,900 --> 00:08:24,040
What are your thoughts on this?

97
00:08:24,040 --> 00:08:24,780
What are your questions?

98
00:08:25,860 --> 00:08:31,200
I would say that I'm a medium skeptic.

99
00:08:31,280 --> 00:08:36,180
There are some things that I have questions about.

100
00:08:36,180 --> 00:09:02,240
And I also believe that Web of Trust can solve a lot of the issues that we have on Noster from community support, community tools surrounding content moderation, as well as spam prevention and so forth.

101
00:09:02,240 --> 00:09:17,380
And I think that Noster allows us to have attention over our feeds and have attention over the content that we want to see.

102
00:09:17,760 --> 00:09:23,560
So the more user tools that we have that allow us to achieve this, the better.

103
00:09:23,700 --> 00:09:28,220
And I think Web of Trust falls into that realm of possibilities.

104
00:09:28,220 --> 00:09:31,960
So I say possibilities because, as you said, a little skeptical.

105
00:09:32,240 --> 00:09:37,180
I do have some questions, and I think that a lot of people on Noster have a lot of questions

106
00:09:37,180 --> 00:09:40,320
because this was kind of just dropped on people out of nowhere.

107
00:09:41,080 --> 00:09:47,700
And immediately we started seeing people talking about social credit scores

108
00:09:47,700 --> 00:09:51,520
and how this was a social credit score and how we don't like assigning numbers to people.

109
00:09:52,320 --> 00:09:58,760
And no matter how I was trying to frame this as user configurable and so forth,

110
00:09:58,760 --> 00:10:05,940
you know it just brought up more questions so i'd like to talk you know some about some of these

111
00:10:05,940 --> 00:10:12,180
questions and see if we can kind of figure this out and so i guess first and foremost we'll just

112
00:10:12,180 --> 00:10:17,900
we'll just you know come after the elephant in the room here per se so a lot of people think

113
00:10:17,900 --> 00:10:23,840
that web of trust is essentially a social credit system a social credit score is it

114
00:10:23,840 --> 00:10:26,360
is it not

115
00:10:26,360 --> 00:10:28,840
so why don't we start

116
00:10:28,840 --> 00:10:31,140
this is a good time to bring David on

117
00:10:31,140 --> 00:10:33,360
Vitor had mentioned in his answer

118
00:10:33,360 --> 00:10:35,520
about how Amethyst

119
00:10:35,520 --> 00:10:37,540
with this new update

120
00:10:37,540 --> 00:10:38,880
users can choose

121
00:10:38,880 --> 00:10:40,380
a service provider

122
00:10:40,380 --> 00:10:42,180
which is essentially

123
00:10:42,180 --> 00:10:44,260
a provider that calculates

124
00:10:44,260 --> 00:10:45,700
these scores

125
00:10:45,700 --> 00:10:48,440
and David you've spent several years

126
00:10:48,440 --> 00:10:49,960
thinking about this problem

127
00:10:49,960 --> 00:10:51,320
developing your

128
00:10:51,320 --> 00:10:52,360
brainstorm

129
00:10:52,360 --> 00:10:54,900
brainstorm web of trust engine.

130
00:10:55,640 --> 00:10:58,000
Do you want to take a stab at answering Derek's question

131
00:10:58,000 --> 00:11:00,500
of how these are not social credit scores?

132
00:11:01,200 --> 00:11:05,840
Yeah, so I think it's a fair question to just say,

133
00:11:05,960 --> 00:11:08,440
okay, what's the difference between this and social credit score,

134
00:11:08,600 --> 00:11:11,280
Black Mirror, Dystopian, stuff like that.

135
00:11:12,860 --> 00:11:16,160
Yeah, we've seen that episode, right, from Black Mirror,

136
00:11:16,160 --> 00:11:21,620
the woman, her social credit score keeps getting lower and lower over time

137
00:11:21,620 --> 00:11:24,780
to where she basically can't do anything based on her score.

138
00:11:25,200 --> 00:11:28,000
Yeah, she wants a really cute picture of something or other

139
00:11:28,000 --> 00:11:30,660
so that she can be invited to the right parties.

140
00:11:30,940 --> 00:11:32,380
I don't remember the details of the episode,

141
00:11:32,380 --> 00:11:34,740
but so you can go get a loan at a bank.

142
00:11:35,840 --> 00:11:39,280
So you're having to be an influencer on social media.

143
00:11:39,420 --> 00:11:40,260
Nobody wants that.

144
00:11:40,400 --> 00:11:42,260
That's the fiat system.

145
00:11:42,420 --> 00:11:44,460
That's not the system we're trying to rebuild.

146
00:11:44,940 --> 00:11:48,600
And so it's fair to ask, what's the difference?

147
00:11:48,600 --> 00:11:58,000
And to me, the difference boils down to whether it's centralized or decentralized.

148
00:11:58,780 --> 00:12:06,260
Or another way of putting it is whether the scores are global scores or whether they're personalized.

149
00:12:07,540 --> 00:12:14,600
And so what that means is in a black mirror social credit score dystopian,

150
00:12:14,600 --> 00:12:20,840
You've got, you know, the scores are all calculated by the government or by some big central entity.

151
00:12:20,840 --> 00:12:21,140
Yes.

152
00:12:21,940 --> 00:12:43,220
Or, you know, what the world is used to right now is you go to Amazon and you see average scores or you go to eBay, you go to Google, wherever you go, you have these large institutions that are centralized that calculate what I call global scores.

153
00:12:44,200 --> 00:12:47,880
And in a way, global really just means it's personalized to Google.

154
00:12:48,440 --> 00:12:56,720
You know, you're seeing the world from Google's perspective, from Jeff Bezos' perspective, you know, from Amazon's perspective.

155
00:12:57,180 --> 00:12:58,660
So it's a centralized score.

156
00:12:58,820 --> 00:13:00,060
It's a centralized score.

157
00:13:00,300 --> 00:13:01,280
Yeah, it's global.

158
00:13:01,480 --> 00:13:02,260
It's centralized.

159
00:13:02,620 --> 00:13:05,060
Everybody has the same score no matter who.

160
00:13:05,700 --> 00:13:10,820
Well, I mean, not all the same number, but sure, there's only one interpretation of that score.

161
00:13:11,240 --> 00:13:11,680
Exactly.

162
00:13:11,680 --> 00:13:21,060
If you go to Amazon and I go to Amazon and we look at the rating of a product, I think we're going to see the same score.

163
00:13:22,000 --> 00:13:32,680
And if the tech company decides to show you and me different scores, well, it's still centralized because they're the ones who we don't have any idea how they're doing that.

164
00:13:33,140 --> 00:13:36,040
We don't have any control over how the scores are calculated.

165
00:13:36,680 --> 00:13:38,500
We don't know how the scores are calculated.

166
00:13:38,720 --> 00:13:39,720
That's the dystopia.

167
00:13:39,720 --> 00:14:04,230
Yeah And so the opposite of that is scores that are personalized which means that you and I are going to see different potentially see different scores for the same person the same product the same business And you and I are going to have as much control as we want over how the scores are calculated

168
00:14:04,890 --> 00:14:16,070
So that means we, if we want to dive into the details, look under the hood and figure out how the scores are calculated and then change the details, we'll be able to do that.

169
00:14:16,070 --> 00:14:27,310
Now, if we don't want to do that, we will have the ability to delegate those details to our web of trust.

170
00:14:27,850 --> 00:14:31,630
And so still, I'm at the center of my web of trust.

171
00:14:31,750 --> 00:14:33,250
You're at the center of your web of trust.

172
00:14:33,990 --> 00:14:38,110
And so you and I have ultimate control over all of these details.

173
00:14:39,410 --> 00:14:44,350
That doesn't make it difficult because we can delegate all those hard decisions.

174
00:14:44,350 --> 00:14:55,050
But it's personalized. The scores need to be personalized. That, to me, is what makes it different than Black Mirror social credit score.

175
00:14:55,050 --> 00:15:12,130
So you're saying then every single user is in control of their own system, their own web of trust, metrics, weights, provider, everything.

176
00:15:12,130 --> 00:15:21,290
So, for example, I might see, I'll just say a number, I look at you and for me, I see you maybe at 100.

177
00:15:22,050 --> 00:15:28,190
Vieter looks at you and he sees you at 50, just based on Vieter's own metrics, your relationship and so forth.

178
00:15:28,510 --> 00:15:35,670
So every single person on NOSTER would see people at their own personal score level and it's not some central entity.

179
00:15:35,670 --> 00:15:41,610
And probably nobody's two scores would be the same from one person to the next.

180
00:15:42,130 --> 00:15:44,830
So, yes, that's correct.

181
00:15:45,370 --> 00:15:48,890
You would see one score, Vitor would see another score.

182
00:15:49,050 --> 00:15:51,070
Now, they don't have to be different.

183
00:15:51,310 --> 00:15:55,870
It is very, you know, it's possible that your grapevine and Vitor's grapevine

184
00:15:55,870 --> 00:15:59,510
are going to give me the same score, or maybe almost exactly the same score.

185
00:15:59,870 --> 00:16:02,090
You know, maybe it's off by one digit.

186
00:16:03,130 --> 00:16:06,310
And in fact, that's...

187
00:16:06,310 --> 00:16:11,290
David, you use the word grapevine, and many folks listening might not be familiar with it.

188
00:16:11,290 --> 00:16:13,330
Do you want to expand on that before you continue?

189
00:16:13,850 --> 00:16:16,530
Yeah, I should probably just say web of trust.

190
00:16:17,010 --> 00:16:21,670
When I say Grapevine, I mean that it means a web of trust,

191
00:16:21,670 --> 00:16:27,550
but it's the web of trust as implemented by using the tools

192
00:16:27,550 --> 00:16:29,650
that we are building at Nosfabrica,

193
00:16:29,850 --> 00:16:32,950
so the GrapeRank algorithm, for example.

194
00:16:33,430 --> 00:16:35,790
But I probably should have just used web of trust

195
00:16:35,790 --> 00:16:37,470
as a more generic term.

196
00:16:37,910 --> 00:16:40,750
So your web of trust and Vitor's web of trust

197
00:16:40,750 --> 00:16:45,590
and Navi's web of trust might give me the same reputation score

198
00:16:45,590 --> 00:16:49,030
or maybe off just by a little bit, or it could be completely different.

199
00:16:51,210 --> 00:16:53,430
But yeah, that's what it means to be personalized.

200
00:16:53,990 --> 00:16:56,470
Now, some people are going to hear that and think, well, that's chaos.

201
00:16:57,030 --> 00:16:59,050
Everyone sees a different answer.

202
00:17:01,730 --> 00:17:05,050
Using the tools that we've, the brainstorm prototype,

203
00:17:05,270 --> 00:17:09,530
the prototype tools that we've built at NosFabrica so far,

204
00:17:09,530 --> 00:17:23,170
What we're seeing is that if you and I, you know, if five people use the same algorithm to calculate a trust score, then using our tools, those five people are going to, you know, calculate Alice's trust score.

205
00:17:23,410 --> 00:17:36,450
And they're going to get probably almost exactly the same score if they're using the same algorithms and the same parameters, which is actually very exciting to see that.

206
00:17:36,450 --> 00:17:43,370
It means it's a mechanism for all of us to agree on things, on a question, like, is this person trustable?

207
00:17:44,030 --> 00:17:50,490
Even though there's nothing forcing us to agree on the answer to any given question.

208
00:17:51,830 --> 00:17:53,430
I don't know, does that make sense?

209
00:17:54,030 --> 00:17:55,190
It does, David.

210
00:17:55,290 --> 00:17:58,830
But I think what will be helpful here is we're talking about scores.

211
00:17:59,370 --> 00:18:01,750
And Vitor mentioned service providers.

212
00:18:01,750 --> 00:18:10,650
I think what I believe should be clear to the audience is the scores come from the service providers who are running some algorithm.

213
00:18:11,710 --> 00:18:17,830
Now, I think a very obvious follow-up question for anyone listening would be, well, how are these scores calculated?

214
00:18:18,670 --> 00:18:20,930
Or I also think, who are the providers too?

215
00:18:21,510 --> 00:18:25,230
Like how do you get a provider, find a provider?

216
00:18:25,370 --> 00:18:26,230
Who are the providers?

217
00:18:26,230 --> 00:18:33,430
There's a lot of questions about how that number is displayed.

218
00:18:33,430 --> 00:18:40,750
We see the end result, a number, but there's several missing pieces of the puzzle to finding out how we get to the end result.

219
00:18:42,570 --> 00:18:42,630
Yeah.

220
00:18:44,150 --> 00:18:50,010
One of the things that makes it so challenging to talk about is that we're talking about all these very abstract ideas.

221
00:18:50,010 --> 00:18:58,370
And this is why I think it's so great that Amethyst is actually showing these scores because it gives us something concrete to talk about.

222
00:18:59,530 --> 00:19:07,690
So right now, there are three that I know of, different service providers that calculate personalized.

223
00:19:08,250 --> 00:19:10,450
I call these scores personalized and portable.

224
00:19:10,450 --> 00:19:27,010
So it's your scores from your perspective, but it's portable, meaning calculated by a service provider, and then you can consume them someplace else like Amethyst or any client that decides to use these scores.

225
00:19:27,770 --> 00:19:34,430
Now, also, I should just caveat, these service providers ideally are open source, which means you can be your own service provider.

226
00:19:34,750 --> 00:19:37,290
Just like you can run your own Bitcoin node.

227
00:19:37,410 --> 00:19:38,730
You don't have to, but you can.

228
00:19:38,730 --> 00:19:45,350
You can run your own lightning node, your own stir fry, your own, I'm sorry, your own Nostra relay.

229
00:19:45,790 --> 00:19:47,850
You don't have to, but you always have that power.

230
00:19:48,470 --> 00:19:53,250
So right now there are three service providers at various stages of development.

231
00:19:53,630 --> 00:19:56,910
There's Brainstorm, which is what we're building at Nost Fabrica.

232
00:19:57,370 --> 00:20:02,090
And there's a prototype like pre-alpha version that's available right now.

233
00:20:03,570 --> 00:20:04,870
So there's Brainstorm is one.

234
00:20:04,870 --> 00:20:16,370
There is Vertex, which is a second one, which Papelia launched a year or so ago-ish, something like that, which calculates page rank, personalized and global page rank scores.

235
00:20:16,750 --> 00:20:23,130
And then a third one is Relator, which there's a team of people behind that.

236
00:20:23,290 --> 00:20:33,970
And they, I think just last week, just, is that right, Avi and Vitor published, started to publish scores in a way that they can be consumed by Amethyst?

237
00:20:33,970 --> 00:20:37,490
Yeah, Related just shipped a couple days ago, actually.

238
00:20:38,190 --> 00:20:45,190
The module within Related that computes the trust scores that we use on Amethyst.

239
00:20:47,130 --> 00:20:54,870
Okay, so right now, if you go to Amethyst, you have to have your trusted assertions calculated by a service provider

240
00:20:54,870 --> 00:21:02,510
and then published using the NIP85 trusted assertions NIP, which you, Vitor, authored a little over a year ago.

241
00:21:02,510 --> 00:21:13,270
And right now there's maybe 30 or 40 people who have had those scores calculated and published at the Brainstorm prototype.

242
00:21:14,570 --> 00:21:19,530
Vitor, do you know how many people have published them using Relator?

243
00:21:20,910 --> 00:21:22,130
I have no idea.

244
00:21:22,130 --> 00:21:28,330
I think most people with Relator, since you have to deploy your own server with Relator,

245
00:21:28,790 --> 00:21:36,810
it's more of a tool for developers or some higher tech people that can actually do the work of keeping a server alive.

246
00:21:37,650 --> 00:21:43,670
And other folks who just use Brainstorm or Vertex or other service provider can just, you know,

247
00:21:44,250 --> 00:21:47,490
you click a button, they'll sign up and they calculate it for you.

248
00:21:47,490 --> 00:21:52,410
And you're kind of bound to the services they provide behind this algorithm.

249
00:21:53,470 --> 00:21:54,410
So how do I do that?

250
00:21:54,490 --> 00:22:07,550
I can go to Vertex or Brainstorm and sign in with Nostr or whatever and then request that my web of trust, my trusted assertions be generated and published?

251
00:22:08,550 --> 00:22:15,450
Yeah, so you could go, Derek, you could go right now to my straycat.brainstorm.social,

252
00:22:15,450 --> 00:22:22,130
which is in my Noster profile, and click a button to sign in,

253
00:22:22,530 --> 00:22:26,570
click a button to sign up for my prototype service,

254
00:22:27,030 --> 00:22:29,850
click another button to activate trusted assertions,

255
00:22:30,270 --> 00:22:33,770
and then I emphasize this is a prototype.

256
00:22:33,770 --> 00:22:50,930
It's very slow, but I can go and then I will have it calculate all your grapevine, your web of trust scores, and it'll publish those using the NIP85 spec.

257
00:22:51,610 --> 00:22:57,970
Then you can go to Amethyst and you can see that Alice has a score of 89 and Bob has a score of 92.

258
00:22:57,970 --> 00:23:01,470
And those scores are personalized to you.

259
00:23:01,890 --> 00:23:08,710
And since you're using brainstorm, those scores are calculated using follows and mutes and reports.

260
00:23:09,250 --> 00:23:10,590
That's just the way.

261
00:23:10,770 --> 00:23:14,350
I remember doing that a couple months ago whenever we first started talking about this.

262
00:23:14,450 --> 00:23:15,770
So I didn't remember the process.

263
00:23:15,950 --> 00:23:16,890
Thanks for explaining it.

264
00:23:16,890 --> 00:23:27,270
Um, follow up question to that is, do I need to do something, you know, every so often to

265
00:23:27,270 --> 00:23:29,430
get these types of things refreshed?

266
00:23:29,430 --> 00:23:35,970
Or is that something that you continually crawl my, my web of trust and recalculate

267
00:23:35,970 --> 00:23:40,230
every X amount of, you know, days, hours, months or something like, how does this stay

268
00:23:40,230 --> 00:23:40,790
up to date?

269
00:23:40,790 --> 00:23:43,630
so we're rebuilding it right now

270
00:23:43,630 --> 00:23:44,490
from scratch

271
00:23:44,490 --> 00:23:47,750
and the new version of Brainstorm

272
00:23:47,750 --> 00:23:49,130
is going to

273
00:23:49,130 --> 00:23:51,690
keep track of all the follows and mutants

274
00:23:51,690 --> 00:23:52,750
reports in real time

275
00:23:52,750 --> 00:23:55,130
and for

276
00:23:55,130 --> 00:23:57,730
all of them that we can find in the universe

277
00:23:57,730 --> 00:23:59,730
basically, all follows and mutants reports

278
00:23:59,730 --> 00:24:01,190
that we have access to

279
00:24:01,190 --> 00:24:04,070
and then when you sign up

280
00:24:04,070 --> 00:24:05,750
we'll calculate your scores

281
00:24:05,750 --> 00:24:06,670
when you sign up

282
00:24:06,670 --> 00:24:09,750
and we haven't yet

283
00:24:09,750 --> 00:24:14,050
decided how often those are going to be recalculated. It depends on

284
00:24:14,050 --> 00:24:18,450
how many people sign up and how computationally expensive it turns out to be.

285
00:24:20,550 --> 00:24:22,490
Probably they will

286
00:24:22,490 --> 00:24:25,690
automatically be calculated at some set interval.

287
00:24:26,510 --> 00:24:29,130
I don't know if that'll be every day, every week, whatever, something.

288
00:24:30,050 --> 00:24:34,350
Then maybe at some distant point in the future we could start

289
00:24:34,350 --> 00:24:38,810
having a paid service. It'll calculate, it'll update once an hour, something like that.

290
00:24:38,810 --> 00:24:55,910
Or we might have flags that get triggered where let's say that some highly trusted user suddenly gets reported as an imposter by 10 trusted accounts.

291
00:24:56,310 --> 00:25:06,810
Then maybe we'll have a flag that will recalculate that score for everybody who signs up and publish that quick because that's the kind of thing that would need to be updated quickly.

292
00:25:06,810 --> 00:25:15,370
so how often remains to be seen that's going to be in the way that we do it in the way that vertex

293
00:25:15,370 --> 00:25:20,170
vertex currently does not publish trusted assertions i've been trying to convince

294
00:25:20,170 --> 00:25:28,830
to do that and maybe he might do that i think at some point and the way relator does that is going

295
00:25:28,830 --> 00:25:41,430
And, you know, I think Relator and Vertex, they place a high value on you request the scores and you get them calculated quickly.

296
00:25:41,430 --> 00:25:42,450
So it's up to date.

297
00:25:42,450 --> 00:25:48,390
Like, you know, the scores are going to be done five seconds ago.

298
00:25:50,650 --> 00:25:54,490
And so different providers are going to have different answers to that question,

299
00:25:54,490 --> 00:26:01,150
Which is why I think it's so important that we have an ecosystem of lots of service providers.

300
00:26:02,090 --> 00:26:13,410
And as you, Derek, know, people, listeners may not know, at NOS Fabrica, we have a Web of Trust hackathon, which we are in the beginning of.

301
00:26:13,410 --> 00:26:14,530
It's going through April.

302
00:26:14,930 --> 00:26:23,870
And part of our goal is to develop the ecosystem of service providers within NOSTER.

303
00:26:24,490 --> 00:26:41,830
Noster has matured to the point where it is ready to have Web of Trust scores that are personalized and portable, which means Brainstorm, Vertex, Relator, hopefully other entrants into the field will calculate scores.

304
00:26:42,350 --> 00:26:45,130
And then you'll be able to consume them right now at Amethyst.

305
00:26:45,430 --> 00:26:48,050
And Nostria has started incorporating those.

306
00:26:48,050 --> 00:26:51,630
hopefully we'll see more and more clients incorporating these scores

307
00:26:51,630 --> 00:26:57,250
and doing so in a manner where the client doesn't tell the user,

308
00:26:57,550 --> 00:26:59,870
oh, we only integrate with Vertex.

309
00:27:00,090 --> 00:27:01,690
We only integrate with Brainstorm.

310
00:27:02,330 --> 00:27:06,550
What's great about the way that it's working at Amethyst right now

311
00:27:06,550 --> 00:27:08,430
is what Vitor just talked about.

312
00:27:09,590 --> 00:27:12,250
Amethyst doesn't say we're using Brainstorm scores.

313
00:27:12,250 --> 00:27:17,850
It's the end user who decides which service provider that user wants to use.

314
00:27:18,050 --> 00:27:30,070
David, I think it's worth, and Derek, we'll get into the questions again, but this is a good point just to take stock of some of this because there have been a lot of terms we've used.

315
00:27:31,050 --> 00:27:35,510
So we've talked about the service providers and we talked about trusted assertions.

316
00:27:35,630 --> 00:27:58,320
And I think it worth highlighting in simple terms that a trusted assertion is just a Nostra event It is a delivery mechanism for scores that once generated they sit on relays then it up to a client like Amethyst and potentially other clients to then integrate that and show those scores in some manner right to improve the user experience

317
00:27:58,880 --> 00:28:02,280
The trusted assertions themselves do not calculate the scores.

318
00:28:02,680 --> 00:28:04,220
It's the service providers who do.

319
00:28:04,840 --> 00:28:09,740
David mentioned three, Vertex, Relator, and Brainstorm Under Nos Fabrica.

320
00:28:09,740 --> 00:28:14,880
all of them have different methodologies for calculating their scores.

321
00:28:15,080 --> 00:28:18,840
Vertex today uses PageRank, both personalized and global.

322
00:28:19,160 --> 00:28:21,980
They don't publish trusted assertions as things stand.

323
00:28:22,100 --> 00:28:26,300
So if you want the Vertex scores, you'll actually, I believe it's through API,

324
00:28:26,440 --> 00:28:30,240
but I could correct me if I'm wrong here, but Vertex is via API.

325
00:28:30,580 --> 00:28:36,180
Relator, I'm not sure what their methodology is for calculating Web of Trust scores,

326
00:28:36,180 --> 00:28:43,440
but they've just, as of recently, Jesus, the developer of Relator,

327
00:28:43,800 --> 00:28:47,080
he's publishing these as trusted assertions.

328
00:28:47,220 --> 00:28:53,160
So in theory, an amethyst user who's got their scores calculated at Relator

329
00:28:53,160 --> 00:28:57,720
can then show them, they will be displayed on that user's profile.

330
00:28:58,200 --> 00:29:01,180
And then maybe, David, you could talk a little bit about Brainstorm.

331
00:29:01,180 --> 00:29:08,180
brainstorm you said follow a brainstorm uses follows uh mutes and reports to calculate scores

332
00:29:08,180 --> 00:29:14,400
i think it'll be helpful for people to understand that methodology a little bit how how the weights

333
00:29:14,400 --> 00:29:21,580
get assigned for each event uh or each of those uh whether it's a follow mute or a report and the

334
00:29:21,580 --> 00:29:26,600
fact that it's actually extensible meaning this is just one version of brainstorm a user can run

335
00:29:26,600 --> 00:29:32,500
their own, or a client or anyone, right, can run their custom version of Brainstorm that incorporates

336
00:29:32,500 --> 00:29:39,000
different signals. It can be replies, it can be zaps, it can be repos, or whatever it is. It doesn't

337
00:29:39,000 --> 00:29:46,240
have to be follows, meets, and reports. Yeah, so why don't, so different service providers will use

338
00:29:46,240 --> 00:29:51,460
completely different algorithms to calculate scores. And so, yeah, I want to start out with

339
00:29:51,460 --> 00:29:54,880
the way scores are calculated right now

340
00:29:54,880 --> 00:29:56,900
using Brainstorm.

341
00:29:57,500 --> 00:29:58,640
So follows and mutants reports,

342
00:29:58,720 --> 00:30:00,940
if I'm going to calculate Derek's scores,

343
00:30:01,720 --> 00:30:03,420
then everybody's,

344
00:30:03,560 --> 00:30:06,360
it's going to end up with a score between 0 and 1

345
00:30:06,360 --> 00:30:08,060
or between 0 and 100,

346
00:30:08,240 --> 00:30:11,420
depending on just how we communicate it.

347
00:30:12,320 --> 00:30:14,900
Everybody's score by default is a 0.

348
00:30:15,360 --> 00:30:16,940
So if you go to Nostradot band,

349
00:30:17,040 --> 00:30:19,400
there's I think 45 million pub keys.

350
00:30:19,400 --> 00:30:33,880
So every single one of them has a default score of zero from Derek's perspective, except Derek's own pub key has a score which is pinned at one or 100%, however you communicate it.

351
00:30:34,140 --> 00:30:41,080
Meaning Derek is 100% confident that Derek is not a spammer or not a bad actor.

352
00:30:41,860 --> 00:30:49,620
So now if Derek, everybody who Derek follows, the follow will get those and pubs off of zero.

353
00:30:50,060 --> 00:30:54,280
Not all the way up to 100, but we'll get it off zero.

354
00:30:54,840 --> 00:31:02,160
And then if they follow someone, then those, you know, so now Derek's follows and their follows have a score that's above zero.

355
00:31:02,160 --> 00:31:08,220
And every time somebody trusted follows you, it gets your score higher.

356
00:31:08,220 --> 00:31:12,540
and now the score asymptotically approaches 100%.

357
00:31:12,540 --> 00:31:15,960
So if you have 100 trusted people following you,

358
00:31:16,280 --> 00:31:18,360
your score might be 98.

359
00:31:18,640 --> 00:31:21,220
And then if 10,000 trusted people follow you,

360
00:31:21,320 --> 00:31:22,900
your score gets to be 99.

361
00:31:23,740 --> 00:31:28,520
So it's not the kind of score that shows how popular you are.

362
00:31:29,040 --> 00:31:30,920
The purpose of this score is,

363
00:31:31,760 --> 00:31:36,220
can we say with confidence that you're not a bot or a bad actor?

364
00:31:36,220 --> 00:31:45,200
and so follows boost your score up a mute will push your score down and a report will

365
00:31:45,200 --> 00:31:52,920
also push your score down now a follow people follow pretty indiscriminately um but if if

366
00:31:52,920 --> 00:32:00,100
somebody reports someone they usually really mean it and so our default parameters are that a follow

367
00:32:00,100 --> 00:32:06,960
only has a little bit of a weight, but a report carries a big weight. So if, if a hundred people

368
00:32:06,960 --> 00:32:14,700
follow what turns out to be a bad actor, and then 10 people report that as an impersonator,

369
00:32:15,180 --> 00:32:21,720
then that part, that NPUB is going to have a score of zero because the 10 reports are going to

370
00:32:21,720 --> 00:32:26,200
outweigh the 100 follows. Now I'm assuming these are reports and follows by trusted people.

371
00:32:26,200 --> 00:32:27,340
if you

372
00:32:27,340 --> 00:32:30,120
if you're a bad actor

373
00:32:30,120 --> 00:32:32,600
and you spin up a thousand bots

374
00:32:32,600 --> 00:32:34,440
and they all follow you

375
00:32:34,440 --> 00:32:36,420
well that's not going to get your score

376
00:32:36,420 --> 00:32:38,040
off zero because they're bots

377
00:32:38,040 --> 00:32:39,020
and their score is zero

378
00:32:39,020 --> 00:32:41,320
which means we ignore everything that they say

379
00:32:41,320 --> 00:32:44,160
so follows

380
00:32:44,160 --> 00:32:46,120
yeah so right now

381
00:32:46,120 --> 00:32:48,200
this uses follows mutants and reports

382
00:32:48,200 --> 00:32:50,500
to calculate scores

383
00:32:50,500 --> 00:32:51,740
from you

384
00:32:51,740 --> 00:32:52,900
from Derek's perspective

385
00:32:52,900 --> 00:32:55,600
but like you said Avi

386
00:32:55,600 --> 00:33:14,060
This is using the grape rank algorithm, which is designed to be extensible, which means that when we get ready, once we get to the point where we're ready to do this, we'll be able to incorporate zaps into the score or maybe reactions.

387
00:33:14,060 --> 00:33:24,600
A lot of people object to using follows and they say, I'd rather use how much interaction there is.

388
00:33:24,600 --> 00:33:33,140
If somebody replies to me or if I reply to somebody or if I react, give thumbs up to all their posts, well, that's a better indicator than a follow.

389
00:33:33,900 --> 00:33:42,340
And so GrapeRank is designed that it can take that and it can take data no matter where you want to find it.

390
00:33:42,960 --> 00:33:50,720
And it can turn, it can interpret that, meaning you take the follow or you take the zap or the reaction or whatever.

391
00:33:50,720 --> 00:33:53,340
somebody has to write a script

392
00:33:53,340 --> 00:33:55,260
to translate that into

393
00:33:55,260 --> 00:33:57,040
a format

394
00:33:57,040 --> 00:33:58,960
that is ready to be

395
00:33:58,960 --> 00:34:01,180
incorporated into the basic

396
00:34:01,180 --> 00:34:02,580
grape rank

397
00:34:02,580 --> 00:34:05,200
algorithm. And so now you have a

398
00:34:05,200 --> 00:34:07,400
score that is based off of

399
00:34:07,400 --> 00:34:09,400
some other sources of data

400
00:34:09,400 --> 00:34:11,280
that you want to use

401
00:34:11,280 --> 00:34:12,160
for your scores.

402
00:34:13,720 --> 00:34:15,340
So tell me

403
00:34:15,340 --> 00:34:16,900
do you have questions?

404
00:34:17,080 --> 00:34:17,780
Is that making sense?

405
00:34:17,780 --> 00:34:19,500
Eric?

406
00:34:21,080 --> 00:34:25,600
I think that, yes, that makes sense.

407
00:34:25,780 --> 00:34:28,240
And it brings about two questions.

408
00:34:28,680 --> 00:34:35,420
I think first and foremost that when you're out of this, I'll say beta, alpha, whatever phase,

409
00:34:35,800 --> 00:34:41,420
I think the ability to have user-customized weights is more important.

410
00:34:41,420 --> 00:34:46,800
Because, as you said, maybe people follow quite liberally.

411
00:34:47,780 --> 00:34:51,340
So maybe follows aren't the best metric, at least for me.

412
00:34:51,640 --> 00:34:57,640
Maybe my best metric is zaps and comments and so forth.

413
00:34:57,680 --> 00:35:02,120
Or maybe I want to be my web of trust.

414
00:35:02,280 --> 00:35:05,000
I don't care about kind one messages or something like that.

415
00:35:05,060 --> 00:35:06,400
My comments and stuff like that.

416
00:35:06,480 --> 00:35:10,520
I care about producing content on Divine, perhaps.

417
00:35:10,520 --> 00:35:21,800
Or maybe it's I'm a developer, so I want, you know, vibe coding apps with Shakespeare, you know, publishing apps and repositories to be my signal.

418
00:35:22,100 --> 00:35:30,960
So I think being able to truly configure that for every single person is going to be the game changer there.

419
00:35:30,960 --> 00:35:39,460
So when I can truly build a web of trust for myself based off of the types of content that I value.

420
00:35:39,460 --> 00:35:43,940
I think that's when we see a lot of success with Web of Trust.

421
00:35:44,380 --> 00:35:51,900
Now, one of the reasons I've always been skeptical of Web of Trust is new users.

422
00:35:52,320 --> 00:35:55,520
New users have essentially a zero Web of Trust.

423
00:35:55,860 --> 00:36:01,120
They join the ecosystem, and they've never participated in it before.

424
00:36:01,280 --> 00:36:03,860
So they essentially have a score of zero.

425
00:36:04,260 --> 00:36:07,960
Does that mean that they're locked out of the ecosystem?

426
00:36:07,960 --> 00:36:15,300
And does that mean that they're essentially a pariah until they somehow get seen and interacted with?

427
00:36:15,480 --> 00:36:19,980
How does a brand new user compete in this scoring system?

428
00:36:20,120 --> 00:36:20,860
How do they survive?

429
00:36:21,640 --> 00:36:25,640
So that's one of many, many great, great questions.

430
00:36:26,400 --> 00:36:32,620
And GrapeRank, the basic GrapeRank score, was designed with that in mind.

431
00:36:32,620 --> 00:36:43,100
So if you're a new user, let's say a friend brought you on.

432
00:36:43,600 --> 00:36:48,500
So somebody who is trusted within the Nostra ecosystem brings you on.

433
00:36:48,900 --> 00:36:50,620
You sign up, that person follows you.

434
00:36:50,940 --> 00:36:56,480
You've got one follow, and maybe you have like two or three others because of just people you know.

435
00:36:56,480 --> 00:37:04,080
So your score, those first handful of follows are going to boost your great rank score a lot.

436
00:37:05,180 --> 00:37:15,400
And, you know, the last, you know, once you get to Jack Dorsey or Odell or any of us, you know, a single follow doesn't boost our score much at all.

437
00:37:15,500 --> 00:37:19,820
But a small handful of follows is going to get your score way off of zero.

438
00:37:20,260 --> 00:37:24,060
And one of the parameters that you can change is a parameter called rigor.

439
00:37:24,060 --> 00:37:31,900
and rigor basically means how many follows does somebody have to have for you to feel confident

440
00:37:31,900 --> 00:37:39,080
like this person is worth your attention so a high rigor means i'm not going to pay attention

441
00:37:39,080 --> 00:37:46,500
to anybody who doesn't have a hundred trusted follows because they could be a bot but a uh

442
00:37:46,500 --> 00:37:52,200
the opposite end of the extreme a low rigor means if two trusted people follow you well then

443
00:37:52,200 --> 00:37:58,960
I'm convinced you're not a bot and you're worth just as much of my attention as anybody else.

444
00:37:59,500 --> 00:38:02,680
So that rigor, it's a parameter between zero and one.

445
00:38:02,780 --> 00:38:06,780
You can shift it to be whatever you want it to be.

446
00:38:07,260 --> 00:38:10,340
How much attention do you want to pay to new users?

447
00:38:10,560 --> 00:38:13,480
And Derek, I know you've always been great at welcoming new users.

448
00:38:13,480 --> 00:38:20,940
So you might want to set your score to your rigor parameter to be very low

449
00:38:20,940 --> 00:38:24,820
so that you'll see new users as they come up.

450
00:38:25,560 --> 00:38:25,580
Okay.

451
00:38:26,400 --> 00:38:28,980
Now, if I interact with that user without following them,

452
00:38:29,100 --> 00:38:33,660
like if I comment on their post, does that boost their potential score

453
00:38:33,660 --> 00:38:37,620
because Derek commented on their post?

454
00:38:37,620 --> 00:38:42,780
Well, so right now, we're only using follows and mutants reports.

455
00:38:43,280 --> 00:38:45,480
But we have to take this in baby steps.

456
00:38:45,600 --> 00:38:47,600
We can't build out everything all at once.

457
00:38:47,600 --> 00:38:53,820
And so in the future, yes, that will, that the answer is going to be yes.

458
00:38:53,880 --> 00:38:58,660
If you react, if you interact with someone, if you like their posts or answer to them,

459
00:38:58,660 --> 00:39:02,420
then it will boast that person's score.

460
00:39:02,940 --> 00:39:05,440
And I should say that there's not just one score.

461
00:39:05,520 --> 00:39:06,600
There's going to be multiple scores.

462
00:39:07,260 --> 00:39:13,680
There's, you'll have one grape rank score that basically means, is this person, can

463
00:39:13,680 --> 00:39:16,040
we say with confidence that this is not a bad actor?

464
00:39:16,040 --> 00:39:19,500
And that's the score that you can see at Amethyst right now.

465
00:39:19,860 --> 00:39:23,760
If you go to Amethyst and someone has a score of 95,

466
00:39:24,500 --> 00:39:29,200
then that means this is not a bad actor according to your web of trust.

467
00:39:29,660 --> 00:39:31,840
And we say that with 95% confidence.

468
00:39:32,680 --> 00:39:35,580
But in the future, we're going to have a different score

469
00:39:35,580 --> 00:39:42,700
that might mean, is this person, is this NPUB an expert on,

470
00:39:42,700 --> 00:39:45,240
is this a Nostra developer?

471
00:39:46,040 --> 00:39:47,440
And that's a different score.

472
00:39:48,600 --> 00:39:54,420
You might decide to experiment with different methods of calculating these.

473
00:39:54,420 --> 00:39:57,740
So you could have, let's say, one grape rank score

474
00:39:57,740 --> 00:40:01,760
that's calculated using follows and mutants reports.

475
00:40:02,320 --> 00:40:05,180
Then maybe you want to have another one that's calculated

476
00:40:05,180 --> 00:40:11,980
only using zaps and replies and NIP25, Kine7 reactions.

477
00:40:12,920 --> 00:40:15,800
And maybe you use these two scores for the same purpose,

478
00:40:16,040 --> 00:40:18,720
which is to decide who's worth your attention,

479
00:40:19,020 --> 00:40:20,280
just in a very generic sense.

480
00:40:20,660 --> 00:40:23,560
But you use them both just to figure out which one works better.

481
00:40:24,240 --> 00:40:29,060
So you can have the same question that you answer using different algorithms,

482
00:40:29,060 --> 00:40:32,020
or you might have a completely different question.

483
00:40:32,020 --> 00:40:36,560
Is this person an expert in, you know,

484
00:40:36,620 --> 00:40:41,320
is this an account I want to follow because they generate great content on some topic?

485
00:40:41,420 --> 00:40:45,840
That's a different question than whether is this account a Nostra developer,

486
00:40:46,040 --> 00:40:49,360
Different question than can I say with confidence that this is not a bot.

487
00:40:51,400 --> 00:40:57,260
One of the things that I'm really looking forward to in this new age of Noster,

488
00:40:57,960 --> 00:41:02,380
if we do get this Web of Trust scores out there from multiple providers

489
00:41:02,380 --> 00:41:04,880
with very different algorithms and so on,

490
00:41:05,580 --> 00:41:09,460
each one will have their own individual ways to tweak the results

491
00:41:09,460 --> 00:41:11,740
or to include and remove different things.

492
00:41:11,740 --> 00:41:15,260
They will offer different levers to the users.

493
00:41:16,040 --> 00:41:19,720
there will be, at some point, there will be a client

494
00:41:19,720 --> 00:41:23,840
that simply subscribes the user to all of the providers

495
00:41:23,840 --> 00:41:27,080
and puts all of the scores side by side.

496
00:41:27,080 --> 00:41:29,680
So you can see that in provider A,

497
00:41:29,680 --> 00:41:34,560
your best friend is score higher than on provider B.

498
00:41:34,560 --> 00:41:51,990
And you can go there and maybe tweak some of the levers over there and see if you can match it But the goal here is of course to make you aware of what the different ways of scoring people

499
00:41:51,990 --> 00:41:57,370
and different ways that these providers are offering those tools to you.

500
00:41:57,470 --> 00:41:59,570
Because, of course, they're still trusted.

501
00:41:59,910 --> 00:42:04,470
You as a user, somebody else is running this computation for you.

502
00:42:04,470 --> 00:42:08,930
you cannot download all the data to verify if those scores are correct or not.

503
00:42:09,410 --> 00:42:13,910
So by the simple fact that they are publishing these scores as non-stored events,

504
00:42:14,430 --> 00:42:18,070
any other client can just download those events and say, you know what?

505
00:42:18,750 --> 00:42:22,190
Service provider A and service provider B are very similar.

506
00:42:22,690 --> 00:42:27,650
They do have a few things different and you can emphasize those differences inside that client.

507
00:42:28,170 --> 00:42:33,410
But service provider C is trying to do something else and it may not be what you want.

508
00:42:33,410 --> 00:42:36,670
It may only work for a few different folks.

509
00:42:37,310 --> 00:42:42,150
I was using this example with one of the members in our community that said, you know what?

510
00:42:42,230 --> 00:42:48,690
I would like a provider that is aligned with the ideals of my church.

511
00:42:49,850 --> 00:43:01,590
So the scores reflect not only trustworthy you are in the Nostra ecosystem, but also how religious you are in the followings of that church.

512
00:43:01,590 --> 00:43:06,650
and I was like well so be it right you can go there and you can choose one maybe create a

513
00:43:06,650 --> 00:43:12,130
provider maybe the church can have their own ways of doing this and they offer it as a service to

514
00:43:12,130 --> 00:43:18,070
their own parish and you know they can choose to do so the church cannot force them to use

515
00:43:18,070 --> 00:43:25,010
right and no one else can uh we can now influence that decision but no one else can go there and

516
00:43:25,010 --> 00:43:30,290
decide for them so they will need to go there um hopefully they will have a system that they can see

517
00:43:30,290 --> 00:43:35,290
how the scores are different and how each provider is calculating those scores.

518
00:43:36,230 --> 00:43:41,150
And if we can have a client that compares them in a very interesting screen

519
00:43:41,150 --> 00:43:45,930
that can show the differences to users, that would just be highly beneficial to Noster.

520
00:43:46,150 --> 00:43:51,310
And it helps us to understand ourselves, understand what we want to see in Noster

521
00:43:51,310 --> 00:43:59,230
and how much certain types of users should be scored as opposed to others.

522
00:43:59,230 --> 00:44:06,250
You know, what are your preferences? Do you have topics of interest? Do you have, you know, certain types of discussions that you want to see?

523
00:44:07,370 --> 00:44:11,830
Another one of our users were just talking about, OK, I don't want to I don't want to have conflict.

524
00:44:12,010 --> 00:44:14,970
I'm not here to debate people. I'm here to just have fun.

525
00:44:15,390 --> 00:44:27,490
OK, so I want scores that match the level of fun this provider or the level of fun this key is giving me through memes or through discussions or through, you know, some interesting post.

526
00:44:27,490 --> 00:44:29,750
I don't care about the politics.

527
00:44:29,950 --> 00:44:33,230
I don't care about anything that people are talking about on Nostrad.

528
00:44:33,270 --> 00:44:34,490
I don't care about Nostrad itself.

529
00:44:34,570 --> 00:44:35,530
I don't care about anything else.

530
00:44:35,930 --> 00:44:38,350
I only care about me being happy.

531
00:44:38,690 --> 00:44:45,410
And if you are one of those people, I'm sure in the future you're going to find a provider that can actually compute those things for you,

532
00:44:45,850 --> 00:44:48,430
turn them into, say, numbers on your screens,

533
00:44:48,430 --> 00:44:58,490
and you'll be able to see which people to follow and which people to not follow based on your own way of seeing how NOSTA should work.

534
00:44:59,130 --> 00:45:08,090
I think, Vitor, that maybe a question that a lot of people have then is, you know, the first step is to show numbers, which you're doing.

535
00:45:08,090 --> 00:45:19,330
But what features are you thinking of implementing to utilize those numbers?

536
00:45:20,910 --> 00:45:35,870
Sure, spam prevention might be something that's easy, but I think the tools built after we have the numbers are the things that are going to make this web of trust, trusted assertions, really click for people.

537
00:45:35,870 --> 00:45:39,630
what type of use cases because we say oh it's configurable you can have it mean anything

538
00:45:39,630 --> 00:45:47,110
okay so now i have all this data all these numbers what can i do with it so so the easy way here is

539
00:45:47,110 --> 00:45:52,990
like if you are on your notification screen and you're getting hammered by some type of scammer

540
00:45:52,990 --> 00:45:58,450
out there trying to sell you some crypto coin right and these people are getting very smart

541
00:45:58,450 --> 00:46:04,790
they're creating new accounts they're creating uh randomized posts that go around our filters

542
00:46:04,790 --> 00:46:11,170
on Oster and you would like you don't want to see those things as a user you just make the decision

543
00:46:11,170 --> 00:46:15,510
you have made the decision many many years ago but you know the clients are still catching up

544
00:46:15,510 --> 00:46:20,190
to that need to block certain types of contents given what the user has chosen

545
00:46:20,190 --> 00:46:27,570
by having these scores in the app there could be a possibility for on your notification screen to

546
00:46:27,570 --> 00:46:34,270
simply have a filter to not show anything under say 10 on the score range and that and that may

547
00:46:34,270 --> 00:46:39,670
not be your default. Your default may be fully open. You are always seeing everything. But when,

548
00:46:39,670 --> 00:46:45,030
you know, in those times in NOS that we have seen before, where, you know, for a month or two or so,

549
00:46:45,130 --> 00:46:50,790
there's just this bots that they keep coming. And for those two months, you switch to this new view

550
00:46:50,790 --> 00:46:56,210
of like, oh, I only see, you know, scores that are at least 10. And then all of that disappears

551
00:46:56,210 --> 00:47:01,930
from the client because, you know, they don't have enough scores to show up in your content yet.

552
00:47:01,930 --> 00:47:05,950
Of course, during that time, you may also miss some new users.

553
00:47:06,270 --> 00:47:09,710
So there is a back and forth that you have to be thoughtful about.

554
00:47:10,490 --> 00:47:27,370
But in the long run, we hope to get enough or good enough with that tooling that you can actually pick and choose which items to activate and deactivate if you're being attacked or if you're being mocked by other people or not.

555
00:47:27,370 --> 00:47:31,370
And how to protect yourself against these attacks.

556
00:47:31,930 --> 00:47:35,270
on a decentralized network that no one has control over anything.

557
00:47:35,430 --> 00:47:38,250
So everything has to be done kind of by you on your phone.

558
00:47:38,750 --> 00:47:43,430
And these scores just help the client figure out, you know, what are the possibilities there.

559
00:47:43,950 --> 00:47:48,790
If you've already chosen a provider, those scores are already out there.

560
00:47:49,330 --> 00:47:50,710
Clients just need to use it.

561
00:47:50,770 --> 00:47:56,130
So it's all about downloading them, combining them, and, you know, being on a search

562
00:47:56,130 --> 00:48:00,790
or when you're tagging people to help you choose the right Derek on Noster.

563
00:48:01,930 --> 00:48:05,750
You can easily use this course to filter those things out.

564
00:48:05,930 --> 00:48:09,570
I wouldn't call it filters because filters implies that we are hiding things from users.

565
00:48:10,010 --> 00:48:15,470
We're just sorting them in a way that the first ones are the ones that you should be paying attention to.

566
00:48:15,470 --> 00:48:19,390
You can still see the rest, but they're all likely imposters anyway.

567
00:48:20,050 --> 00:48:22,590
So they're just going to be further behind on the list.

568
00:48:23,270 --> 00:48:26,070
And so those filters will exist.

569
00:48:26,270 --> 00:48:29,490
I don't know if we know all of them yet.

570
00:48:29,490 --> 00:48:31,830
we are still exploring a lot of this

571
00:48:31,830 --> 00:48:33,990
this new user

572
00:48:33,990 --> 00:48:36,130
experiences but you know

573
00:48:36,130 --> 00:48:38,210
since we have things out there since we have

574
00:48:38,210 --> 00:48:40,310
this course everything becomes a lot easier

575
00:48:40,310 --> 00:48:42,270
because now clients don't need to compute this

576
00:48:42,270 --> 00:48:44,250
course and develop the feature at

577
00:48:44,250 --> 00:48:45,790
the same time and it

578
00:48:45,790 --> 00:48:48,310
helps Noster in staying

579
00:48:48,310 --> 00:48:50,270
decentralized because we don't actually

580
00:48:50,270 --> 00:48:52,190
want clients computing

581
00:48:52,190 --> 00:48:54,370
scores or doing

582
00:48:54,370 --> 00:48:55,830
that level of processing

583
00:48:55,830 --> 00:48:58,210
we want to decentralize so that your

584
00:48:58,210 --> 00:49:02,150
your Amethyst and your Primo or your Deimos,

585
00:49:02,670 --> 00:49:04,570
we all use the same scores.

586
00:49:06,150 --> 00:49:07,890
And so we just do the screen.

587
00:49:08,030 --> 00:49:10,370
The UI is just a little different within ourselves.

588
00:49:10,810 --> 00:49:14,770
But everybody else is using the same key, the same score set.

589
00:49:15,370 --> 00:49:20,130
I think this is a critical distinction that he hit upon, Vitor,

590
00:49:20,290 --> 00:49:23,770
which is clients not computing the scores.

591
00:49:24,030 --> 00:49:28,070
And David, you had an article called Separation of Trust and Client.

592
00:49:28,210 --> 00:49:36,770
And, you know, when we think about Nostra, obviously the two first class citizens in Nostra are the client and the relay.

593
00:49:37,550 --> 00:49:44,370
And I wonder if because, you know, and we haven't even gotten to curation and discovery yet.

594
00:49:44,410 --> 00:49:50,810
We're just talking about, you know, healthy experience without spam and scammers and so on.

595
00:49:50,810 --> 00:50:01,170
we don't want the client to get into the business of assigning scores calculating scores because

596
00:50:01,170 --> 00:50:08,110
that just it just a bad set of incentives at that point right and as i said we had the two

597
00:50:08,110 --> 00:50:15,990
first class additions the clients and the relays now with web of trust coming in for cure for i guess

598
00:50:15,990 --> 00:50:19,110
spam removal, better browsing experience, and so on.

599
00:50:19,390 --> 00:50:23,930
Do you think, David, that the web of trust service providers,

600
00:50:24,090 --> 00:50:26,390
I don't know if they've become first-class citizens

601
00:50:26,390 --> 00:50:31,170
at the level of Relay, but maybe a 1.5 or something like that?

602
00:50:32,990 --> 00:50:36,230
Yeah, I think they definitely are going to be a whole new thing,

603
00:50:36,230 --> 00:50:40,850
like a whole new entity that will be very important.

604
00:50:41,170 --> 00:50:45,910
Yeah, maybe equal to the others, maybe, like you said, 1.5.

605
00:50:45,990 --> 00:50:47,730
I don't know, something like that.

606
00:50:47,870 --> 00:50:50,630
But I think a question could be,

607
00:50:50,850 --> 00:50:54,210
well, why shouldn't the clients calculate all these scores?

608
00:50:55,170 --> 00:50:59,430
And to me, the answer is the word portability.

609
00:50:59,430 --> 00:51:04,750
I want my web of trust scores to be portable,

610
00:51:04,990 --> 00:51:09,290
meaning I want to go to Primal and Amethyst and see the same score.

611
00:51:10,430 --> 00:51:15,730
Because an 89 out of 100 is going to mean different things

612
00:51:15,730 --> 00:51:18,910
depending on what algorithm you're using

613
00:51:18,910 --> 00:51:21,130
and what parameters you're using.

614
00:51:21,990 --> 00:51:25,270
So if you shift the rigor from one extreme to the other,

615
00:51:25,490 --> 00:51:28,170
an 89 might mean one thing, might mean something else.

616
00:51:28,390 --> 00:51:33,910
So I just want to have to make those choices,

617
00:51:34,130 --> 00:51:36,590
set the algorithm and the parameters once,

618
00:51:37,450 --> 00:51:44,590
figure out that a score of 50 means not worth my time

619
00:51:44,590 --> 00:51:47,450
or maybe a score of 10 means yes, worth my time.

620
00:51:47,850 --> 00:51:50,990
And then see the same score on all the different clients.

621
00:51:51,810 --> 00:51:56,850
So portability is one reason why service providers, I think,

622
00:51:57,070 --> 00:52:02,630
should be an entirely new class of citizens separate from clients.

623
00:52:03,330 --> 00:52:05,790
And then another reason is that a lot of these scores

624
00:52:05,790 --> 00:52:09,250
are just very computationally expensive.

625
00:52:10,090 --> 00:52:12,990
If you're going to calculate page rank or grape rank

626
00:52:12,990 --> 00:52:17,110
or a lot of these scores, you've got to gather together a large amount of data,

627
00:52:17,510 --> 00:52:18,950
do a lot of number crunching,

628
00:52:19,090 --> 00:52:25,970
and it is just not feasible to do all of that on the fly for every score that exists.

629
00:52:26,250 --> 00:52:31,830
Now, historically, there are clients like Coracle and Wikifredia

630
00:52:31,830 --> 00:52:37,170
and some other ones that were pioneers in calculating personalized trust scores.

631
00:52:37,690 --> 00:52:39,930
And they deserve a lot of credit for that pioneering.

632
00:52:39,930 --> 00:52:45,570
I don't know any place outside of Nostra that calculates personalized scores the way we have done.

633
00:52:46,190 --> 00:52:47,310
But they weren't portable.

634
00:52:47,930 --> 00:52:51,310
I would see a different score in one client than another one.

635
00:52:51,870 --> 00:52:59,130
So yeah, where exactly are service providers going to end up in our minds?

636
00:52:59,250 --> 00:52:59,710
I don't know.

637
00:52:59,910 --> 00:53:01,650
Probably something pretty high.

638
00:53:02,850 --> 00:53:08,530
And to be clear, saying that we don't want clients to compute scores

639
00:53:08,530 --> 00:53:14,150
doesn't mean that we can't have scores computed locally, say, on your phone,

640
00:53:14,370 --> 00:53:21,250
and then send it back to the network as trusted assertions to be seen by Amethyst Primal and so on.

641
00:53:21,290 --> 00:53:23,150
There could be a client that does that locally.

642
00:53:23,910 --> 00:53:27,190
What we're trying to avoid is what David was saying,

643
00:53:27,630 --> 00:53:33,630
that the score itself gets committed to the client and can't leave that space

644
00:53:33,630 --> 00:53:39,250
because it's too attached to features or things that were developed for that client.

645
00:53:40,430 --> 00:53:44,170
So if you can, it's possible, probably likely,

646
00:53:44,670 --> 00:53:48,210
that in the future we will see some type of client that does this calculation locally

647
00:53:48,210 --> 00:53:52,610
and just tries to offer that as a solution that you don't need a server

648
00:53:52,610 --> 00:53:56,750
or you don't need a third party to run this course for you.

649
00:53:56,830 --> 00:54:01,250
Your own phone could compute this course while you're sleeping, for instance,

650
00:54:01,250 --> 00:54:02,710
while it's charging on the wall.

651
00:54:03,630 --> 00:54:07,710
so that it doesn't use your battery, but it runs locally,

652
00:54:07,710 --> 00:54:11,550
so you can fully trust the results of that algorithm

653
00:54:11,550 --> 00:54:13,910
because you are seeing the source code,

654
00:54:14,090 --> 00:54:16,990
you are controlling the execution environment,

655
00:54:17,550 --> 00:54:19,670
you are controlling your phone, you know what it does,

656
00:54:20,130 --> 00:54:22,410
and you can actually trust what's happening with it

657
00:54:22,410 --> 00:54:25,030
because it is running locally.

658
00:54:25,170 --> 00:54:27,030
So that is still possible.

659
00:54:27,810 --> 00:54:30,950
The only thing we are saying is that the outcome of that processing

660
00:54:30,950 --> 00:54:32,350
becomes a not-start event,

661
00:54:32,350 --> 00:54:35,750
that then can be seen and not manipulated,

662
00:54:35,910 --> 00:54:38,810
but seen and used by every other client out there.

663
00:54:39,310 --> 00:54:41,690
Yeah, I should probably rephrase what I said earlier.

664
00:54:41,850 --> 00:54:44,650
So I actually do want clients to have the ability

665
00:54:44,650 --> 00:54:47,210
to calculate personalized trust scores.

666
00:54:47,510 --> 00:54:49,990
I just don't want them to be the only ones who do it.

667
00:54:50,190 --> 00:54:52,190
They shouldn't have to carry the entire burden.

668
00:54:52,190 --> 00:54:58,390
So for example, I can imagine one day going to a music app like Toonster

669
00:54:58,390 --> 00:55:03,730
or let's say maybe an e-commerce site

670
00:55:03,730 --> 00:55:06,330
that has five-star ratings on products

671
00:55:06,330 --> 00:55:12,210
and the client might calculate the average five-star rating

672
00:55:12,210 --> 00:55:14,070
of this particular product.

673
00:55:14,330 --> 00:55:15,990
And that's not computationally expensive

674
00:55:15,990 --> 00:55:18,810
because let's say there's 20 ratings,

675
00:55:19,710 --> 00:55:22,030
then it's not a whole lot of data to gather

676
00:55:22,030 --> 00:55:23,990
and you can process it on the fly.

677
00:55:23,990 --> 00:55:33,210
But that calculation, I want that to filter out the spam and I want to ignore authors that are not.

678
00:55:33,240 --> 00:55:34,760
trusted by my web of trusts.

679
00:55:35,200 --> 00:55:39,320
So that calculation would use scores

680
00:55:39,320 --> 00:55:42,300
that are first calculated by a service provider

681
00:55:42,300 --> 00:55:45,300
and then use those for their own calculations

682
00:55:45,300 --> 00:55:48,780
that are done locally on their client.

683
00:55:48,900 --> 00:55:51,260
And then, like Vitor was saying,

684
00:55:51,620 --> 00:55:54,100
ideally, in some cases,

685
00:55:54,880 --> 00:55:57,380
that client might only want to calculate

686
00:55:57,380 --> 00:55:59,900
and crunch the numbers and only use it locally.

687
00:55:59,900 --> 00:56:05,120
they have the option of also publishing the results of that calculation

688
00:56:05,120 --> 00:56:07,180
so that it could be used elsewhere.

689
00:56:10,380 --> 00:56:14,600
Derek, where do you land on service providers

690
00:56:14,600 --> 00:56:19,600
as a key primitive, I guess, in the protocol?

691
00:56:20,600 --> 00:56:22,380
We have clients and relays,

692
00:56:22,380 --> 00:56:26,800
and now we have things like, I guess, Blossom servers,

693
00:56:26,800 --> 00:56:30,520
which are becoming increasingly important.

694
00:56:31,380 --> 00:56:34,500
Where do you see service providers fall on that list?

695
00:56:35,260 --> 00:56:37,200
That's just another part of the ecosystem, right?

696
00:56:37,500 --> 00:56:41,400
Like we started out by just having clients and relays.

697
00:56:42,160 --> 00:56:43,740
We expanded the protocol.

698
00:56:44,820 --> 00:56:50,940
We expanded the ecosystem and the protocol a little bit with media hosting.

699
00:56:50,940 --> 00:56:58,620
Now part of your decentralized social communication arsenal is your media server.

700
00:57:00,000 --> 00:57:18,260
We see different apps popping up that allow us to do what we're doing exactly right now, live streaming, which uses essentially a centralized media host, whether that is coming from like Zap.Stream or from Primal.

701
00:57:18,260 --> 00:57:21,740
I'm using my own Sovereign Sobo

702
00:57:21,740 --> 00:57:22,440
Okay

703
00:57:22,440 --> 00:57:24,160
There you go, okay

704
00:57:24,160 --> 00:57:25,440
But yes, I get your point

705
00:57:25,440 --> 00:57:28,900
Well, the same point is I can run my own Relay

706
00:57:28,900 --> 00:57:31,460
I can run my own Blossom Media server

707
00:57:31,460 --> 00:57:36,120
I can run my own, you know, Cast Livestream server

708
00:57:36,120 --> 00:57:39,560
Maybe you run your own Trusted Assertions server

709
00:57:39,560 --> 00:57:42,620
Because the only person you want to trust is yourself

710
00:57:42,620 --> 00:57:46,860
Maybe it becomes part of the technology stack

711
00:57:46,860 --> 00:57:53,820
I hate to say it because it's a meme at this point, but we're still early.

712
00:57:54,480 --> 00:58:01,920
And I think that we're kind of figuring all of this out as we go forward.

713
00:58:01,920 --> 00:58:13,680
And maybe in the near future, you will have clients that are full-blown service providers that provide half a dozen different value-added services.

714
00:58:13,680 --> 00:58:17,020
They provide the client, but then they also provide the relay.

715
00:58:17,140 --> 00:58:18,320
They provide the streaming service.

716
00:58:18,460 --> 00:58:22,000
They provide the identity service for Nostra addresses.

717
00:58:22,180 --> 00:58:25,500
They provide maybe a lightning wallet for Zaps and all sorts of things.

718
00:58:25,560 --> 00:58:26,940
These are all pieces of the puzzle.

719
00:58:27,600 --> 00:58:37,640
And if we have to add another piece of the puzzle, I think that that's fine because now we don't have one centralized entity that just does everything.

720
00:58:37,880 --> 00:58:42,980
There's half a dozen pieces of the puzzle, and it's all spread out around the ecosystem.

721
00:58:42,980 --> 00:58:44,580
I think that's fine.

722
00:58:45,000 --> 00:58:53,580
I think that is the premises of Noster, that you can run as little or as much of the ecosystem that you want to run.

723
00:58:53,940 --> 00:59:03,120
And if this becomes something that many people want, we will start seeing various service providers running this.

724
00:59:03,120 --> 00:59:06,420
Maybe they charge for their algorithm.

725
00:59:06,420 --> 00:59:16,700
Maybe their algorithm that they're using is superior for some reason, and they're charging you to have assertions published by them.

726
00:59:17,180 --> 00:59:20,740
That very well could be a business model that we see in the future.

727
00:59:21,120 --> 00:59:25,600
Then maybe you have open source models that anybody and everybody can run.

728
00:59:25,940 --> 00:59:29,620
I think that we're still early applies here.

729
00:59:32,920 --> 00:59:34,800
Yeah, that's absolutely right.

730
00:59:34,800 --> 00:59:39,120
And the monetization that you talk about, I think we're going to see it.

731
00:59:40,860 --> 00:59:45,880
So there could be a, you know, you could have a, for free,

732
00:59:45,980 --> 00:59:48,380
you can get your scores calculated infrequently,

733
00:59:48,680 --> 00:59:53,920
but then if you want them calculated every day or every hour or every minute.

734
00:59:54,080 --> 00:59:54,500
Absolutely.

735
00:59:55,100 --> 00:59:55,800
I like that.

736
00:59:56,380 --> 00:59:57,140
Then you pay more.

737
00:59:57,540 --> 00:59:59,620
And you can do them on your own if you want.

738
00:59:59,620 --> 01:00:00,960
You could do them on your phone.

739
01:00:00,960 --> 01:00:06,420
but your phone's not connected all the time.

740
01:00:06,700 --> 01:00:07,740
So you have options.

741
01:00:08,500 --> 01:00:10,380
You can do them on their phone just periodically

742
01:00:10,380 --> 01:00:12,740
just to check to make sure that your paid service

743
01:00:12,740 --> 01:00:14,900
is not lying to you, for example.

744
01:00:16,240 --> 01:00:19,420
And what you said, I would imagine

745
01:00:19,420 --> 01:00:22,360
any one user is going to use

746
01:00:22,360 --> 01:00:23,840
lots of different service providers.

747
01:00:23,840 --> 01:00:29,120
So I might want Brainstorm to calculate my basic metric

748
01:00:29,120 --> 01:00:31,700
and then maybe a trust score.

749
01:00:31,820 --> 01:00:37,240
Maybe I want Toonster to calculate my curated list

750
01:00:37,240 --> 01:00:42,580
of trusted musicians or music curators, for example.

751
01:00:43,580 --> 01:00:46,840
I can at the same time have lots of different service providers

752
01:00:46,840 --> 01:00:50,880
who are calculating scores using my web of trust metrics

753
01:00:50,880 --> 01:00:54,560
because they can effectively talk to each other.

754
01:00:56,400 --> 01:00:58,380
You mentioned curation,

755
01:00:58,380 --> 01:01:01,300
And I think this is a good time to talk about that.

756
01:01:01,640 --> 01:01:05,280
So when we think about, or at least up until now in this conversation,

757
01:01:05,480 --> 01:01:07,800
and when people think about web of trust, right,

758
01:01:07,840 --> 01:01:12,240
it's about, well, can we make the experience better by filtering out the spam

759
01:01:12,240 --> 01:01:16,820
or by dealing with the spam somehow and all of that.

760
01:01:17,020 --> 01:01:23,540
And that's an important part of having a good user experience for someone on Noster.

761
01:01:23,540 --> 01:01:27,540
But I believe an equally important part is the discovery itself,

762
01:01:28,380 --> 01:01:34,880
Like, for example, in independent music, which is, you know, a Nostra adjacent ecosystem, right?

763
01:01:34,940 --> 01:01:38,100
Value for value, independent music, or even podcasts for that matter.

764
01:01:39,220 --> 01:01:48,620
If I suddenly find myself saying, look, I have 40 hours a week open, I need to find my podcast, I need to find music to listen to.

765
01:01:48,920 --> 01:01:51,060
And I wanted to be in this ecosystem, right?

766
01:01:51,080 --> 01:01:56,960
I don't want to go to Spotify, I don't want to go to Apple Podcasts or Apple Music or whatever it is.

767
01:01:56,960 --> 01:01:59,620
I want to support independent creators.

768
01:01:59,860 --> 01:02:00,140
Great.

769
01:02:00,560 --> 01:02:02,000
You show up on an app like Fountain.

770
01:02:02,160 --> 01:02:06,980
You show up on a lot or, you know, I don't know what the music app is.

771
01:02:07,400 --> 01:02:09,900
Maybe Derek, one of the ones you vibe coded, right?

772
01:02:10,200 --> 01:02:11,520
I think you did a music app.

773
01:02:12,440 --> 01:02:13,860
That's the beauty of the ecosystem.

774
01:02:14,020 --> 01:02:14,760
It's always growing.

775
01:02:15,500 --> 01:02:19,700
So you show up there and you're like, okay, well, there's all this music.

776
01:02:19,800 --> 01:02:21,120
How do I find something?

777
01:02:21,500 --> 01:02:22,620
Or there are all these podcasts.

778
01:02:22,760 --> 01:02:25,780
How do I find something that I would actually like?

779
01:02:25,780 --> 01:02:28,900
and maybe we start with you, David.

780
01:02:29,520 --> 01:02:31,020
How does web of trust,

781
01:02:31,240 --> 01:02:32,540
and I know there's this,

782
01:02:32,680 --> 01:02:34,840
we're going to throw one more curveball here,

783
01:02:35,200 --> 01:02:35,960
a little more complexity.

784
01:02:36,420 --> 01:02:38,700
There's contextual web of trust, right?

785
01:02:39,700 --> 01:02:42,800
So maybe it might be helpful to talk about that

786
01:02:42,800 --> 01:02:45,660
and how that can help with curation and discovery

787
01:02:45,660 --> 01:02:48,160
and make the user experience fun.

788
01:02:50,440 --> 01:02:53,200
So I think just the ability to filter out the bots

789
01:02:53,200 --> 01:02:59,940
is very, very useful to start to solve this problem.

790
01:03:00,440 --> 01:03:02,900
I mean, in the regular, you know, the fiat world,

791
01:03:03,200 --> 01:03:05,500
the internet as we've known it for so many years,

792
01:03:06,180 --> 01:03:09,360
people make apps where you do, you know,

793
01:03:09,420 --> 01:03:11,460
they're designed to curate something or other.

794
01:03:11,520 --> 01:03:13,980
Like you can go online and you can get ratings of doctors.

795
01:03:14,660 --> 01:03:19,260
You can go to Yelp, you can get ratings of different businesses.

796
01:03:19,840 --> 01:03:22,020
And so you can find what you're looking for that way.

797
01:03:22,020 --> 01:03:24,520
The problem is that the bots and the bad actors creep in.

798
01:03:25,300 --> 01:03:28,280
And so just using similar methods,

799
01:03:28,840 --> 01:03:33,200
but with the added ability to say,

800
01:03:33,340 --> 01:03:38,160
okay, I only want my trusted users to be able to contribute to this.

801
01:03:39,440 --> 01:03:41,900
So let's say if your content,

802
01:03:43,720 --> 01:03:50,380
let's say you want to find people who post cat memes or something,

803
01:03:50,380 --> 01:04:05,992
Then one way to do that would be just do a keyword search for all the events that let say that have the word you looking for And then you just ignore anybody who based on some trust score cutoff

804
01:04:06,512 --> 01:04:10,312
And so now you're going to get probably worthwhile content.

805
01:04:10,532 --> 01:04:12,572
And that's just a simple method.

806
01:04:14,312 --> 01:04:19,052
There's another method that I'm a big,

807
01:04:19,452 --> 01:04:21,572
and maybe this is too much to get into this time,

808
01:04:21,672 --> 01:04:22,712
is decentralized lists.

809
01:04:23,692 --> 01:04:29,532
I want my web of trust to be able to curate a list for me.

810
01:04:30,032 --> 01:04:38,312
So like right now you've got NIP 51 lists, which is a list that I maintain for myself.

811
01:04:38,612 --> 01:04:42,892
If I make a list of Nostra developers, I can use that using NIP 51.

812
01:04:43,572 --> 01:04:51,292
And if I want to add or remove PubKeys from that list, I have to do it myself.

813
01:04:51,292 --> 01:04:57,012
But what if I want to delegate that list to my web of trust?

814
01:04:57,552 --> 01:05:02,312
Then there's a custom nip that I wrote to do that,

815
01:05:02,412 --> 01:05:04,292
which is as simple as I could write it.

816
01:05:04,952 --> 01:05:08,352
But the idea is that anybody can add,

817
01:05:08,712 --> 01:05:12,972
I could create a new list of, I call it Nostra developers,

818
01:05:13,272 --> 01:05:16,632
and then anybody can add pub keys to that list.

819
01:05:16,632 --> 01:05:26,392
And I simply ignore contributions from anybody whose trust score is not above whatever cutoff I decide to have.

820
01:05:26,792 --> 01:05:32,092
So now I've got a living list of Nostra developers.

821
01:05:32,292 --> 01:05:34,932
It can change at any point in time.

822
01:05:35,272 --> 01:05:37,112
I don't have to do anything.

823
01:05:37,992 --> 01:05:45,412
Maybe I submit a bounty to encourage my web of trust to help me curate this list.

824
01:05:45,412 --> 01:05:48,652
then anyone in the world can benefit from it

825
01:05:48,652 --> 01:05:52,112
assuming I haven't encrypted the list header

826
01:05:52,112 --> 01:05:55,592
and now I've got a list of Nostro developers

827
01:05:55,592 --> 01:06:00,752
so then the next step, let's say I want to know

828
01:06:00,752 --> 01:06:05,492
what are the NIPs that are approved by the community

829
01:06:05,492 --> 01:06:09,432
right now you can go to Nostrohub.io

830
01:06:09,432 --> 01:06:13,832
and you can see upvotes and downvotes on custom NIPs

831
01:06:13,832 --> 01:06:19,632
that are filled where anybody who is not,

832
01:06:20,312 --> 01:06:23,492
it's only real Nostra users based on their baseline

833
01:06:23,492 --> 01:06:26,212
who have a trust score.

834
01:06:26,732 --> 01:06:28,292
But I might decide, you know what,

835
01:06:28,332 --> 01:06:32,432
I don't want a list of NIPS curated by any Nostra user.

836
01:06:32,832 --> 01:06:35,332
Maybe I want a list of NIPS that's only curated

837
01:06:35,332 --> 01:06:36,732
by Nostra developers.

838
01:06:37,292 --> 01:06:40,652
So now I'm chaining together different lists.

839
01:06:40,652 --> 01:06:43,712
My list of real people curates the list

840
01:06:43,712 --> 01:06:45,852
of Nostra developers, the list of Nostra developers

841
01:06:45,852 --> 01:06:47,692
curates the list of approved NIPs.

842
01:06:48,452 --> 01:06:50,612
And we can start, chain things together,

843
01:06:50,952 --> 01:06:54,572
and find whatever it is we're looking for

844
01:06:54,572 --> 01:06:57,792
once these tools get built out and used.

845
01:06:58,872 --> 01:07:01,352
David, I think decentralized lists

846
01:07:01,352 --> 01:07:04,712
need an entire episode in and of themselves.

847
01:07:05,432 --> 01:07:07,312
It's a giant rabbit hole.

848
01:07:07,992 --> 01:07:10,732
But maybe, Derek, let me ask you this.

849
01:07:10,732 --> 01:07:14,192
from a discovery and curation standpoint, right?

850
01:07:14,592 --> 01:07:16,312
You talked about new users.

851
01:07:17,152 --> 01:07:20,032
They come in, and right now, if they're on Primal,

852
01:07:20,132 --> 01:07:23,552
Primal has, whatever, 50 NPUBs that you default follow.

853
01:07:24,372 --> 01:07:27,432
And I don't think that's ideal

854
01:07:27,432 --> 01:07:31,012
because it's Primal telling,

855
01:07:31,452 --> 01:07:33,252
or whichever client it might be, right?

856
01:07:33,332 --> 01:07:35,012
Other clients, I suspect, do this too.

857
01:07:35,012 --> 01:07:36,192
but

858
01:07:36,192 --> 01:07:41,172
there's got to be a way for new users

859
01:07:41,172 --> 01:07:41,872
to discover

860
01:07:41,872 --> 01:07:45,332
whom to follow, the different

861
01:07:45,332 --> 01:07:47,232
topics that are being discussed

862
01:07:47,232 --> 01:07:49,132
on Nostra. They're saying, look, I don't want to hear

863
01:07:49,132 --> 01:07:51,112
about Bitcoin. I want to see if someone's talking about

864
01:07:51,112 --> 01:07:53,092
something else. I want to find out about

865
01:07:53,092 --> 01:07:55,132
gardening. Or an existing

866
01:07:55,132 --> 01:07:56,952
user wants to find good music,

867
01:07:57,032 --> 01:07:59,052
good podcasts. How do you think

868
01:07:59,052 --> 01:07:59,552
about

869
01:07:59,552 --> 01:08:02,992
the discovery and curation on

870
01:08:02,992 --> 01:08:04,992
Nostra? And based on this conversation

871
01:08:04,992 --> 01:08:07,772
Do you see how Web Trust can help with that?

872
01:08:09,352 --> 01:08:29,452
I think that part of the issue with Noster, since we don't have these centralized entities to do content recommendation and user recommendation, that people join Noster and they feel like they're in the void.

873
01:08:29,452 --> 01:08:34,952
I think that we have a user retention problem.

874
01:08:35,312 --> 01:08:38,452
I think everybody would probably agree with that.

875
01:08:39,192 --> 01:08:51,932
So maybe Web of Trust can be used for that based on some of these recommended users.

876
01:08:51,932 --> 01:09:02,092
We look across legacy social media, and we do have recommended users through every legacy media app that has ever existed.

877
01:09:03,432 --> 01:09:13,832
Over a decade ago, I was a suggested user on Google+, and that gained me 1.4 million followers back in the day.

878
01:09:13,952 --> 01:09:16,152
So I understand how it works.

879
01:09:16,152 --> 01:09:24,152
um i think that it's also bad too but maybe we can leverage something with web of trust

880
01:09:24,152 --> 01:09:32,112
and maybe we can not just recommend these suggested user list users but we can

881
01:09:32,112 --> 01:09:37,472
recommend people that are part of their web of trust based on certain types of content

882
01:09:37,472 --> 01:09:43,792
right like it's you would still have a little bit of a recommended user but maybe not so much that

883
01:09:43,792 --> 01:09:49,952
user would be recommended. You would be recommending people from their web of trust based on maybe who

884
01:09:49,952 --> 01:09:54,612
you, if you had a join link, like a referral link or something. I don't know. I'm just a bit

885
01:09:54,612 --> 01:09:59,652
balling here. I know that there's something that can be done. There has to be. We have all this

886
01:09:59,652 --> 01:10:05,072
data in a decentralized manner, and we know we have a problem. We just need somebody smarter than

887
01:10:05,072 --> 01:10:07,832
Derek to figure it all out, put it together. Maybe that's Vitor.

888
01:10:07,832 --> 01:10:23,152
But that's, so actually, Derek, you touched on something that Vitor mentioned a few days back or a few weeks back, Vitor, which is maybe Nostra was not ready for Web of Trust up until now.

889
01:10:23,652 --> 01:10:33,852
Now we finally have enough data, enough user behaviors as expressed in events that are sitting on relays that we could actually do something with it.

890
01:10:33,852 --> 01:10:38,792
Yeah, and the reason for that is essentially data, right?

891
01:10:39,572 --> 01:10:45,972
Until, like, I'd say maybe one year ago, we really didn't have a lot of data.

892
01:10:46,112 --> 01:10:53,732
We had a lot of kind of one posts and Twitter-like interfaces that can only go so far, right?

893
01:10:53,832 --> 01:10:56,892
Nostra was very tied to that type of mindset.

894
01:10:56,892 --> 01:11:21,232
Now that we have so many of the other stuff apps going on and getting their own communities aligned and talked about in different ways, you can start to process this data in very different ways to find the things that each user is interested about beyond just followers.

895
01:11:21,232 --> 01:11:32,032
Because for all of the Twitter-like interfaces on Twitter and like on Amethyst, Primo, and Demons, for instance, it's all about follows.

896
01:11:32,192 --> 01:11:38,532
It's kind of like the whole app is designed on this notion of like there are personalities that you need to follow.

897
01:11:38,712 --> 01:11:42,372
And you have this thing in your feed and you kind of follow them.

898
01:11:42,412 --> 01:11:44,132
It's all one thing.

899
01:11:44,572 --> 01:11:47,252
And you don't have multiple communities or multiple lists.

900
01:11:47,372 --> 01:11:49,912
It's all kind of blended together into one interface.

901
01:11:49,912 --> 01:11:52,912
and that kind of worked against us for quite some time

902
01:11:52,912 --> 01:11:56,192
and Amethyst was a big fault in that

903
01:11:56,192 --> 01:11:58,952
and it's I think now

904
01:11:58,952 --> 01:12:02,452
just the right time to start processing the data

905
01:12:02,452 --> 01:12:06,332
to figure out what do people want to actually see

906
01:12:06,332 --> 01:12:08,052
and more importantly

907
01:12:08,052 --> 01:12:11,932
help them tell the app

908
01:12:11,932 --> 01:12:13,532
and tell Noster, the ecosystem

909
01:12:13,532 --> 01:12:15,632
and hopefully the service providers themselves

910
01:12:15,632 --> 01:12:18,732
what they are interested about

911
01:12:18,732 --> 01:12:32,084
and by how much Because until now we only had you know likes replies maybe zaps but they very limited If you start having more interesting things like dynamic lists that David was talking about

912
01:12:32,164 --> 01:12:36,544
that you can put, create as any list you want and you can put any things in there,

913
01:12:37,164 --> 01:12:39,884
you start to see like, I don't just like music.

914
01:12:40,024 --> 01:12:42,064
I like a certain type of music.

915
01:12:42,244 --> 01:12:47,104
So when you're doing content recommendation to me, I don't want to see just music.

916
01:12:47,284 --> 01:12:48,384
I want to see jazz.

917
01:12:48,384 --> 01:13:00,264
I want to see rock. I don't want to see music. You know what I mean? And I think for a very long time, we were stuck in this notion of like the maximum we can do is to offer music to people.

918
01:13:00,264 --> 01:13:04,524
and we knew that no one likes everything in music.

919
01:13:04,684 --> 01:13:08,224
You have to break this down into classes and categories

920
01:13:08,224 --> 01:13:10,464
and maybe age and maybe other things

921
01:13:10,464 --> 01:13:13,684
that will resonate with a new user

922
01:13:13,684 --> 01:13:16,004
or with a user that is looking for something to be,

923
01:13:16,584 --> 01:13:18,204
or to stay on Oster,

924
01:13:18,224 --> 01:13:20,684
to build a community around in Oster.

925
01:13:20,684 --> 01:13:23,984
And I think we're just starting to have that debate

926
01:13:23,984 --> 01:13:27,184
and the level of data to get to that point

927
01:13:27,184 --> 01:13:28,764
that we can actually process it

928
01:13:28,764 --> 01:13:34,664
and figured out how to do better content recommendation engines based on scores.

929
01:13:34,664 --> 01:13:38,224
One of the things that we talked a lot about scores,

930
01:13:38,984 --> 01:13:48,984
and we kept hammering this topic where each person has just one number for any other person, right?

931
01:13:49,164 --> 01:13:53,944
Like Derek is an 80 to me, David's a 90 to me and so on.

932
01:13:53,944 --> 01:13:58,844
and if we go much deeper than that,

933
01:13:59,024 --> 01:14:00,804
we start to realize that the number

934
01:14:00,804 --> 01:14:03,204
doesn't really make much sense.

935
01:14:03,344 --> 01:14:06,344
We want to go further than the number.

936
01:14:06,504 --> 01:14:07,424
We want to say, you know what,

937
01:14:07,984 --> 01:14:10,124
David is a neurologist.

938
01:14:10,664 --> 01:14:12,744
So for neurology topics,

939
01:14:13,364 --> 01:14:16,544
I want those posts to be higher

940
01:14:16,544 --> 01:14:17,984
because I know he's a doctor.

941
01:14:19,064 --> 01:14:21,404
If he talks about baseball,

942
01:14:21,724 --> 01:14:23,284
I want those things to be a 10.

943
01:14:23,944 --> 01:14:30,644
So the same person has different scores based on what the topic of the conversation is about.

944
01:14:31,224 --> 01:14:37,304
And I think we will get that at some point with this framework of like, okay, well, now we're not just scoring people.

945
01:14:37,304 --> 01:14:46,664
We're scoring what they say each time they say and how closely that associates with what the user actually wants to see.

946
01:14:46,664 --> 01:14:52,684
in such a way that people can't just decide to follow, you know, David for the neurology part

947
01:14:52,684 --> 01:14:57,364
or Vitor for the Noster part, but nobody cares about Vitor, the political commentator.

948
01:14:57,864 --> 01:15:04,124
Or you want to follow Derek for the memes and not for, you know, the AI adventures that he is doing.

949
01:15:04,124 --> 01:15:06,564
So you can actually break it down.

950
01:15:06,864 --> 01:15:11,344
And in that point forward, you don't really have just one score per person,

951
01:15:11,484 --> 01:15:13,384
but you have multiple scores.

952
01:15:13,804 --> 01:15:16,104
And now we are talking, now we are trying to see, okay,

953
01:15:16,104 --> 01:15:19,084
this is really a recommendation engine now.

954
01:15:19,264 --> 01:15:22,724
I can actually nail it down to what exactly you want to see,

955
01:15:23,064 --> 01:15:25,664
the types of things you prefer, what you love, what you don't like,

956
01:15:26,184 --> 01:15:29,584
and offer content that is more aligned to what you want to see.

957
01:15:30,724 --> 01:15:33,244
And if you talk about multiple scores, Vitor,

958
01:15:33,364 --> 01:15:36,844
which I think the way you described it makes complete sense, right?

959
01:15:36,844 --> 01:15:41,904
And we're kind of getting to the contextual level of trust with that.

960
01:15:42,364 --> 01:15:44,444
But as a user experience itself,

961
01:15:44,444 --> 01:15:48,704
you're not suggesting that the clients somehow present

962
01:15:48,704 --> 01:15:51,304
all of these multiple scores to the user, right?

963
01:15:51,324 --> 01:15:55,464
It's more about how the content that they're viewing

964
01:15:55,464 --> 01:15:57,704
is presented to them.

965
01:15:58,044 --> 01:16:00,124
Yeah, I think in the end, the scores won't even be there.

966
01:16:00,484 --> 01:16:02,984
We will find a different way to demonstrate

967
01:16:02,984 --> 01:16:07,504
the trust relationship that you have with this particular post

968
01:16:07,504 --> 01:16:09,064
in some other way.

969
01:16:09,984 --> 01:16:12,824
And people have suggested to not score anybody,

970
01:16:12,824 --> 01:16:17,824
You just do, I don't know, a star system or, you know, a yes or no verification metric and so on.

971
01:16:18,364 --> 01:16:23,564
Nothing is really perfect, but the score in the end might not even stay in the app.

972
01:16:23,704 --> 01:16:37,044
What is the important part is can we let them know, one, can, if you, remember, the scores or anything in Web of Trust, they're not really designed to reinforce your follows.

973
01:16:37,704 --> 01:16:39,124
You already trust their follows.

974
01:16:39,204 --> 01:16:39,924
You already know that.

975
01:16:39,984 --> 01:16:40,804
You already know them.

976
01:16:40,804 --> 01:16:41,824
You know their pictures.

977
01:16:41,824 --> 01:16:45,864
when they show up, you are happy to engage in that post.

978
01:16:46,464 --> 01:16:49,984
This course or anything associated like this course

979
01:16:49,984 --> 01:16:53,244
help you talk to people you don't know,

980
01:16:53,484 --> 01:16:55,284
that you have never seen before on Oster.

981
01:16:55,564 --> 01:16:57,544
And you don't know if this person is a scammer,

982
01:16:57,624 --> 01:16:59,924
is an impersonator, or is a real person,

983
01:17:00,044 --> 01:17:02,144
someone else in the world that you just have never seen.

984
01:17:02,664 --> 01:17:04,404
One of the things that I have realized

985
01:17:04,404 --> 01:17:06,544
while using the new version of Amethyst

986
01:17:06,544 --> 01:17:09,904
is how much more I reply to folks

987
01:17:09,904 --> 01:17:14,464
that have a high score because of my network is following them,

988
01:17:15,184 --> 01:17:16,944
but I have never seen them before.

989
01:17:17,784 --> 01:17:21,424
And the score just gives me that additional confidence that I say,

990
01:17:21,424 --> 01:17:25,804
if I reply, I'm not going to be attacked by this scammer

991
01:17:25,804 --> 01:17:29,844
and trying to steal money from me over and over again

992
01:17:29,844 --> 01:17:33,964
because I know that the person has some higher scores.

993
01:17:34,564 --> 01:17:36,704
So I think that is the notion we want to grab.

994
01:17:36,704 --> 01:17:53,684
Whatever convinces users or translates the trust or the score information into something that the users just need to look at and engage with that user or the new user, that's how we want to get in the end.

995
01:17:53,944 --> 01:17:56,564
The score itself is just one of the ways to get there.

996
01:17:56,904 --> 01:17:58,384
And I'm pretty sure it's an ugly way.

997
01:17:58,444 --> 01:17:59,504
It's a very technical way.

998
01:18:00,044 --> 01:18:04,944
And we may not even see it in the future as soon as we figure something better out.

999
01:18:06,704 --> 01:18:16,664
Yeah, I think all these numbers, there will be times when we want the scores to be displayed prominently and times when we want all the scores to be just working under the hood.

1000
01:18:18,244 --> 01:18:31,264
I think right now I'm glad that we're having clients like Amethyst show the scores because these are new tools that we're introducing to the Nostra ecosystem and we need people to know that the tools exist and what they are.

1001
01:18:31,264 --> 01:18:38,944
but at some point, you know, you don't want to shove a bunch of numbers into your users' faces

1002
01:18:38,944 --> 01:18:41,084
and just, you know, it's just too much.

1003
01:18:41,444 --> 01:18:43,444
It's confusing. People don't know what they mean.

1004
01:18:43,604 --> 01:18:45,604
Right. So we will put them...

1005
01:18:45,604 --> 01:18:51,024
I have people asking me, well, I'm a 67. What does that mean? I'm a 97. What does that mean?

1006
01:18:51,024 --> 01:18:56,044
Yeah. And so it's, so I'm glad people are asking those questions because,

1007
01:18:56,044 --> 01:19:02,104
there's so many moving parts to all this. We can't build all of them at the same time.

1008
01:19:02,844 --> 01:19:10,724
Nostra is developing one in baby steps. It was a couple of years ago that clients were calculating

1009
01:19:10,724 --> 01:19:20,304
personalized trust scores. And now we're at the point of making them portable with service

1010
01:19:20,304 --> 01:19:27,344
providers. And I think contextual is going to be what comes next, but we can't do that until we have

1011
01:19:27,344 --> 01:19:34,224
a relatively healthy, you know, at least have a handful of service providers who are calculating

1012
01:19:34,224 --> 01:19:40,504
scores that are personalized and portable. Then we can start focusing on the contextual scores.

1013
01:19:41,964 --> 01:19:47,624
So it's baby steps and we need to, we can't do them in the wrong order even. You know, I don't

1014
01:19:47,624 --> 01:19:51,764
think we could do contextual scores without having them portable or at least we couldn't

1015
01:19:51,764 --> 01:20:03,104
it wouldn't work effectively to do that yeah uh derek any last question for either vitor or david

1016
01:20:03,104 --> 01:20:07,084
well

1017
01:20:07,084 --> 01:20:10,604
how can

1018
01:20:10,604 --> 01:20:13,984
myself as an individual

1019
01:20:13,984 --> 01:20:14,864
or

1020
01:20:14,864 --> 01:20:18,084
a developer

1021
01:20:18,084 --> 01:20:18,764
provider

1022
01:20:18,764 --> 01:20:20,824
an interested party

1023
01:20:20,824 --> 01:20:24,124
run one of these

1024
01:20:24,124 --> 01:20:26,064
trusted assertions

1025
01:20:26,064 --> 01:20:28,504
one of these web of trust calculators

1026
01:20:28,504 --> 01:20:30,484
what can I do to get started

1027
01:20:30,484 --> 01:20:31,644
soapbox

1028
01:20:31,644 --> 01:20:37,544
Sure. Maybe I'm asking for Soapbox. Maybe I'm asking for, you know, I'm a cool dev dot com.

1029
01:20:37,724 --> 01:20:41,184
I mean, whatever. How can I run one to start playing with this?

1030
01:20:41,224 --> 01:20:55,956
Because you know right now I hearing that there a lot of centralization and there has to be in the beginning because you have to start somewhere But we from the group of the world that doesn like centralization

1031
01:20:56,176 --> 01:21:00,936
So we need more of these providers to start popping up, to start experimenting.

1032
01:21:01,436 --> 01:21:03,236
So how can people contribute?

1033
01:21:04,176 --> 01:21:05,396
How can they run something?

1034
01:21:07,936 --> 01:21:10,296
How can I be my own web of trust provider?

1035
01:21:10,296 --> 01:21:14,816
so today I think if you download Relator

1036
01:21:14,816 --> 01:21:17,456
and run yourself and deploy the server

1037
01:21:17,456 --> 01:21:20,936
it should do everything we just talked about

1038
01:21:20,936 --> 01:21:23,176
and give you the scores you want

1039
01:21:23,176 --> 01:21:26,536
I'm guessing like in a few weeks or so

1040
01:21:26,536 --> 01:21:29,076
we're going to have the new version for Brainstorm

1041
01:21:29,076 --> 01:21:31,176
and that will also be similar

1042
01:21:31,176 --> 01:21:32,276
you deploy it on server

1043
01:21:32,276 --> 01:21:36,896
and Brainstorm will compute the same scores

1044
01:21:36,896 --> 01:21:39,156
that we computed for ourselves

1045
01:21:39,156 --> 01:21:40,336
for our test stations.

1046
01:21:41,076 --> 01:21:42,756
You'll be able to do it by yourself.

1047
01:21:42,856 --> 01:21:44,536
We can change the equations as well, of course.

1048
01:21:46,096 --> 01:21:47,376
And hopefully there is more.

1049
01:21:48,216 --> 01:21:50,836
But for now, I think those will be the only options.

1050
01:21:52,616 --> 01:21:55,896
Yeah, so Brainstorm is going to be open source,

1051
01:21:56,036 --> 01:21:56,976
like Vitor just said.

1052
01:21:57,356 --> 01:21:59,856
So hopefully in a couple of weeks,

1053
01:21:59,976 --> 01:22:05,716
you will be able to download and run Brainstorm software

1054
01:22:05,716 --> 01:22:09,336
and calculate your own basic trust scores.

1055
01:22:10,136 --> 01:22:12,296
I would say before doing that,

1056
01:22:12,996 --> 01:22:16,396
it would be worthwhile to use those personalized trust scores.

1057
01:22:16,396 --> 01:22:20,556
And that would be a great thing for vibe coding sessions

1058
01:22:20,556 --> 01:22:26,376
is to make an app where you can upvote your favorite music,

1059
01:22:27,096 --> 01:22:31,936
but only look at upvotes that are from your trusted web of trust

1060
01:22:31,936 --> 01:22:35,596
using scores that are made by a service provider.

1061
01:22:35,716 --> 01:22:46,496
So I think that would be something you might be able to start doing soon, like right now, is just consuming those scores for various different purposes.

1062
01:22:46,496 --> 01:22:49,276
Yeah, I think that's important is we need use cases.

1063
01:22:49,576 --> 01:22:58,616
We can have scores, but they don't really mean a whole lot because people don't know what they mean and people don't really know how to use them.

1064
01:22:58,616 --> 01:23:06,216
So we need easy ways for people to calculate them and then easy ways for people to use them.

1065
01:23:06,716 --> 01:23:11,296
Otherwise, they're not as important as we feel they should be.

1066
01:23:11,776 --> 01:23:19,536
So if you were going to vibe code an app that uses the scores that are already created, what would you want to use them for?

1067
01:23:20,836 --> 01:23:22,916
Why are you putting me on the spot like this, David?

1068
01:23:22,916 --> 01:23:24,016
I don't know.

1069
01:23:24,676 --> 01:23:25,396
That's true.

1070
01:23:25,396 --> 01:23:29,816
You know, I'll tell you what.

1071
01:23:29,936 --> 01:23:45,196
Let's maybe, since we're talking about vibe coding and that's my current kick, maybe we could do something for showing projects that people built on Shakespeare.

1072
01:23:45,676 --> 01:23:45,756
Right?

1073
01:23:45,756 --> 01:23:51,696
Like the app, similar to Nosterhub, very similar to that for showing like nips and so forth.

1074
01:23:51,696 --> 01:24:01,636
But maybe we'd be showing repositories or apps and the more upvotes from our web of trust, they would be featured and so forth.

1075
01:24:02,876 --> 01:24:04,876
Can I make a suggestion, Derek?

1076
01:24:05,236 --> 01:24:11,916
Maybe you call it It's Alive because one of the –

1077
01:24:11,916 --> 01:24:15,156
Frankenstein was a meme for Shakespeare for a while.

1078
01:24:15,696 --> 01:24:17,916
We were literally calling it that and making memes.

1079
01:24:18,216 --> 01:24:20,776
That kind of works out.

1080
01:24:20,776 --> 01:24:27,716
The reason I bring that up is one of the unfortunate side effects of vibe coding is the abandoned wear, right?

1081
01:24:27,736 --> 01:24:29,416
There's an explosion of abandoned wear.

1082
01:24:29,816 --> 01:24:38,836
And maybe if you can have upvotes or downvotes for apps that are actually alive, being maintained and being used, I think that'll be helpful, right?

1083
01:24:39,096 --> 01:24:47,696
Yeah, that's what a directory of actively maintained and used vibe coded apps.

1084
01:24:47,696 --> 01:24:49,476
I think that might work.

1085
01:24:50,016 --> 01:24:50,116
Yeah.

1086
01:24:50,776 --> 01:24:57,416
So should we schedule a vibe coding session to build this?

1087
01:25:00,156 --> 01:25:00,556
Maybe.

1088
01:25:01,476 --> 01:25:03,356
Derek's got Shakespeare plugged in.

1089
01:25:03,556 --> 01:25:05,656
The audio is going straight to Shakespeare.

1090
01:25:05,656 --> 01:25:06,796
Yeah, Shakespeare's building as it speaks.

1091
01:25:07,796 --> 01:25:09,516
I think that would be fun.

1092
01:25:09,516 --> 01:25:14,436
So I do a weekly vibe coding jam session every Monday night,

1093
01:25:14,516 --> 01:25:16,516
about five and a half hours from now.

1094
01:25:16,556 --> 01:25:17,116
I'll be doing one.

1095
01:25:17,116 --> 01:25:27,456
And then next week and so forth, we do them every week to where people can talk about ideas, things that they're interested in building, working on, get live feedback from people.

1096
01:25:28,116 --> 01:25:42,936
Maybe next week we talk about that as saying is that something that's currently in the discussion realm and maybe start tossing some ideas out there about what people could build, what people could vibe code.

1097
01:25:42,936 --> 01:25:55,836
I'd say if you haven't vibe-coded something like that by Nostra Nights, we can maybe sit down at Bitcoin Park and plan something out.

1098
01:25:56,976 --> 01:26:00,076
Okay, yeah, that's coming up in a couple weeks.

1099
01:26:01,676 --> 01:26:08,236
So before we drop, and we've been going for close to 90 minutes here, Derek, do you want to plug Nostra Nights?

1100
01:26:08,236 --> 01:26:18,396
well sure we i guess we kind of just did nostril nights is february 2nd at bitcoin park in nashville

1101
01:26:18,396 --> 01:26:28,056
tennessee where is a the first of quarterly events where we will go and discuss all of the

1102
01:26:28,056 --> 01:26:36,696
things happening around the purple verse from vibe coding since that is the hottest latest way to

1103
01:26:36,696 --> 01:26:38,656
watch your ideas come to life.

1104
01:26:39,216 --> 01:26:44,436
Toil is content ecosystems, app updates, and so forth like that, just growing the Nostra

1105
01:26:44,436 --> 01:26:44,936
ecosystem.

1106
01:26:45,576 --> 01:26:49,516
The first one that's happening here in a few weeks, we're having Will Kassarin, friend

1107
01:26:49,516 --> 01:26:55,056
of all of us here from Damas, that will be speaking about what's coming next on Damas.

1108
01:26:55,496 --> 01:27:02,016
David will be there talking more about Web of Trust, so come and ask questions, IRL, to

1109
01:27:02,016 --> 01:27:07,936
David and then open mic and myself will be talking about our vision for Nostra

1110
01:27:07,936 --> 01:27:09,476
nights throughout 2026.

1111
01:27:09,476 --> 01:27:12,816
And we will have food from the empanadas food truck.

1112
01:27:12,956 --> 01:27:13,996
I forget the name of it,

1113
01:27:14,056 --> 01:27:15,156
but they're delicious empanadas.

1114
01:27:15,276 --> 01:27:21,916
We've had them in Nashville before and food is covered by our amazing

1115
01:27:21,916 --> 01:27:22,336
sponsors.

1116
01:27:22,896 --> 01:27:24,876
So see you there.

1117
01:27:25,596 --> 01:27:26,676
You heard it here,

1118
01:27:26,716 --> 01:27:27,096
folks.

1119
01:27:27,096 --> 01:27:30,336
if the stellar cast of characters

1120
01:27:30,336 --> 01:27:33,816
and the quality of content isn't reason enough

1121
01:27:33,816 --> 01:27:37,816
for anyone within a three-hour radius of Nashville

1122
01:27:37,816 --> 01:27:40,216
to show up on February 2nd, you said, right, Derek?

1123
01:27:40,516 --> 01:27:42,956
For Nostanites at Bitcoin Park,

1124
01:27:43,956 --> 01:27:47,396
then the free empanadas should get you over that line.

1125
01:27:47,556 --> 01:27:47,996
Come on.

1126
01:27:48,116 --> 01:27:49,636
It is really good food.

1127
01:27:50,456 --> 01:27:51,216
All right, folks.

1128
01:27:51,216 --> 01:27:55,296
Well, that wraps up our very first episode

1129
01:27:55,296 --> 01:27:57,996
of Say What by Plep Chain Radio.

1130
01:27:58,456 --> 01:28:01,756
Remember, folks, this is a value for value production.

1131
01:28:02,356 --> 01:28:04,336
If you are listening to this podcast,

1132
01:28:04,856 --> 01:28:08,756
I hope you're doing it on a podcasting 2.0 app like Fountain.

1133
01:28:08,996 --> 01:28:11,776
And if it is Fountain, I would appreciate it

1134
01:28:11,776 --> 01:28:15,056
if you hit that subscribe button to Plep Chain Radio.

1135
01:28:15,576 --> 01:28:19,716
Your support goes a long way in helping me continue

1136
01:28:19,716 --> 01:28:21,876
to make content like this.

1137
01:28:21,876 --> 01:28:30,716
and for these episodes 50% of your boosts and your streams go to the guests we have three guests here

1138
01:28:30,716 --> 01:28:37,236
so guys I'm sorry but you'll have to split that 50% amongst yourselves but let's make it worth

1139
01:28:37,236 --> 01:28:47,056
their while thank you to my guests Vitor David and Derek Ross and we will see you on say what

1140
01:28:47,056 --> 01:28:48,356
maybe in a couple of weeks.

1141
01:28:50,856 --> 01:28:51,676
Thank you, Avi.

1142
01:28:52,676 --> 01:28:53,516
Yeah, thank you.

1143
01:28:53,576 --> 01:28:54,316
It's been great to be here.