Pescetelli Metagov
Speaker 1: Loud. It's my pleasure to introduce, and I'm gonna butcher your last name, Pescatelli?
Top Keywords
- idea 0.006
- ideas 0.005
- decision 0.005
- mean 0.005
- pods 0.005
- integration 0.004
- data 0.004
- example 0.004
- nikola 0.004
- platform 0.004
- discussion 0.004
- organizations 0.004
Transcript
Loud. It's my pleasure to introduce, and I'm gonna butcher your last name, Pescatelli?
Pescatelli.
Yes. Pescatelli. Okay. God. I don't even remember. Nikola, I think we must have met at that Collective Intelligence kind of meet up slash conference. Yeah. Exactly. We also evidently overlapped at Oxford for a while. Nikola has been at NPI before, and he's currently a professor at the New Jersey Institute of Technology. He's also the cofounder of, let's call it Sai, but it's also I think they called Suru together. I mean Yeah.
Suru is the company. Sai is the platform.
Sai is the platform. Yeah. Makes sense. So hopefully they'll yeah. Nikola will kinda tell us about SAI and why it's you should use it for your for, I guess, we should use it for our own community governance since we all have this happened to be a Zoom Zoom call.
Exactly. Alright.
Nikola, I'll hand it off to you.
Alright. So let me share my screen. Okay. So can you see everyone? Perfect. Alright. So, hi, everyone. I'm Niccolo Pesciatelli. I'm really excited to give this presentation and really, like, gather, you know, all your feedback and, you know, from from from the community. And, you know, like, as as Joshua was saying, you know, see how we can collaborate to really, you know, integrate SCI into the matter of the ecosystem. I'm going to present, you know, this platform, which has been, like, a big project of mine, like, in the last, two years. So the last two years, they, you know, could have been better. This paper actually, is, a paper by a colleague of mine that, makes a a really interesting point, which is, the in modern world, like, all the networks that we have created are really hyperconnected, that the density of the network is really, such that information can spread very, very fast, which is great. But the downside is that also, once things start spreading, it's very difficult to stop them. So this is viruses. This is rumors. This is, you know, conspiracy theories and so on and so forth. I like to think about this in terms of, like, resonance versus integration. So resonance is really copying and pasting, like, an idea and and and information spreading very, very fast. But all the platforms that we have offer very little information integration. So for example, polarized views, there are not very many platforms that help kind of bring people together, discuss disagreements and resolve it. And side note, you know, the brain actually is very good at spreading information very fast, but also integrating different streams of information together. So how do we do it in in decision making? Well, you see nowadays, like, you know, there are a lot of communities from web three organizations, but also offline where the ideal would be to to reach a level of decentralized decision making. But oftentimes, like, these organizations grow to a point where decentralized decision making now becomes very, very difficult. And, typically, what the organization does is to, you know, revert back to old ways of governing, like, hierarchies, delegation, representation, and voting. So this type of decision making does not scale. So how do we achieve that information integration while also preserving a level of deliberation? So why deliberation? Why do we want people to really talk ideas? Well, there are, like, various reasons. One thing is that, you know, to integrate information very quickly, you could, you know, put out, like, a poll, a voting system, and you would probably get to a decision very quickly. The problem there is that people don't change their minds. So once you instead are able to talk ideas, now the research shows that ideas start to mix together. And that has a huge amount of benefits, you know, for innovation, but also because the group now can reduce prioritization. So instead of ideas remaining separated, people tend to, you know, change their mind, change their views, and converge. So I'm going to talk mainly about this problem for DAOs, but actually think broadly in terms of governance offline and online. So very briefly, this is an excerpt from the Messari report that was published this year. It says, we'll need to see a 100 x improvement in information flow and decision support tools. You can govern a global DAO or a subDAO with NFTs or social currencies more easily than you might a global corporation. But that doesn't change the fact that without delegated functions, progress in a DAO can move to a standstill when every micro decisions turns into a proxy vote. Okay. So how do we reach that 10 x improvement? Now current platforms out there are, you know, could be something like chat based discussions like Discord. They tend to be, you know, very linear and so very difficult to scale. Plus, you know, a a consensus is never guaranteed. You can have other sort of text based forums like Colony, Commonwealth, but, again, communications tend to be slow and the language, the text barrier, you know, makes very poor engagement. And then we said, you know, broad based decision making like snapshot. But again, you know, no opportunity for debates, very little opinion integration and poor engagement. Now this is a graph that is, I think, is quite interesting. It shows, like, really how DAO size in number of members really co vary with the total asset under management. But it's sort of like plateaus around here. Right? And that's, you know, at SAI, we we like to think that that's because of this increasing governance difficulties. And we have seen it, like, you know, with constitution DAO. We have seen it with SushiSwap. There are a lot of these recurring problems. So just very briefly, the team. So it's me and my cofounder Georgina Dennis who's trained in public sector management. And then we have a bunch of wonderful people, including Chris here who's who's with us today. And, really, like, we source expertise in crowdsourcing, public sector, AI, Web three, and sort of NLP and and web three as as I said. Alright. So I'm going to talk very briefly about the platform, like the sort of the high level insights, and then I'm going to show it to you maybe, like, five minutes at the end. So SAI stands for people supported intelligence. The core idea of SAI is the pod. So a pod, is a collective sense making unit. So, few people discussing a subset of crowdsource proposals. And the outcome of each pod is a local, decision, a local consensus if you want. And so Saia is really like you can imagine it as a funnel of ideas, a very fast funnel where you enter here, crowdsource proposals. This is like beans here, and the platform helps you, organize them into, like, parallel conversations, with these parallel pods. And he assembles and disassembles pods, iteratively until, one decision, remains at the end. It's very scalable. So we estimate that 10,000 members could converge with default parameters in under two hours. And, again, this is using voice. So, really, people talking to other people. And and this is like sort of the main features. Right? So we said that we use audio, which makes it like more engaging so more people can participate. It's based on collective intelligence research, and I can go more into the details in q and a, but he really has, like, interesting information properties. But mainly, you know, it's scalable. You know, thousands of people can converge very, very quickly because of of of this design. Now I said, like, you know, we we are thinking now in terms of, like, online communities, but, really, once you crack this problem, like, you could have really big impact and implications for a number of sectors. So civic governance is a big one. We are currently engaging with a lot of city councils in London, but also corporates. Like, if you think, like, big brands engaging their distributed stakeholders, you know, sports clubs engaging their funds, and so on and so forth. So currently, these are the pilot sites that we are working with. So there is Bangla Stauz that has agreed to USI in the coming weeks. We have a couple of cities. So Lakeland in Florida and Lambeth council in London and the NHS hospital. Very briefly, like, you know, our tooling features to, you know, this is our pipeline. So currently, we are here. We are looking for more early adopters. So if you know about organizations that would be interested in using SyTE, you know, of course, for free, we are happy to really test it out with you guys. Next is going to be really like integration with existing platforms such as Discord using our API And in sort of, like, mid to long term future, we are going to introduce some gamification elements, for example, reward systems, participation tokens, and so on and so forth. And in the end, once you have collected, like, this big amount of data, we can beef up our sort of data analytics and AI capabilities. And one quick note on that is that the vision that we have is really to have almost like a symbiotic relationship between collective intelligence and artificial intelligence. But the important thing is that it's always people that make the decisions by talking, by integrating information. The AI never tampers with that decision. But what the AI can do once that data is collected is really offer insights back to the people. So summaries, highlights, and so on and so forth. Alright. So let me jump to the demo, which is I think is, like, the part that you guys are interested in. So this is the page of, let's say, a city council. Okay. And as a user, you see, like, you know, all the consultations that they have published. So this is actually a real consultation that happened last year in London around police violence. And we took that and put it on-site and asked real people to discuss it. So what you would do is okay. You're interested in the topic, you go and suggest an idea. You can read more about it, and you can see when the live discussion starts. And you can share your idea. You know, how would you solve this problem? Anything from the crowd? How would you solve police violence in London?
Even more cameras. I love it.
Even more cameras. Awesome. You submit your idea, and then you're asked to record a pitch. So this is basically, you know, articulate with your voice why you say you think that more cameras would help. When you're ready, you know, you submit, and that's your proposal as enter the system. Now on the day of the live discussion, you enter the waiting room, and you see all the other people that are joining. This is just like a big waiting room. And once the discussion starts, you are assigned to a first pod. So these are, like, three people, three other users that you will interact with. And we are here in round one. It works in three stages. So first, you have to review the ideas. So these are real ideas from real people. So I have a few ideas. One of them being ensuring that there is appropriate educate education within that community. I'm not gonna, you know, go through them all because it will be too too long. But after reviewing, you enter the discussion. And now I really like the idea and really talk through it. Chat box of understanding, like, the psychology behind why people, cause violence and things. I think that's one way to kind of, look at it, like, actually understanding, like, the So you have, you know, a bunch of minutes to discuss with your in the future and if we understand. And at the end, the third stage is voting. So after this discussion, what do you think happened? What what which idea is the best? At the end of the voting, there is a a sort of this is like the the consensus that the pod has reached. So this idea went forward. And so you go to the next round, and now that pod is disassembled, and now you are entering a new pod where new ideas will be presented. And remember, this is a funnel. So in very few iterations, what you So is with a consensus decision. Even large groups that would take, like, just three, four hour iteration marks. At the end of this process, what happens is that you have this one idea, you know, remaining. But also what you can do is go back, you know, and round by round see which ideas survive, let's say,
or,
like, were proposed initially, which one that were discarded on every round. Almost like providing, like, a sort of high level ranking of all the ideas that is done by live discussion. I'm gonna leave it here because I'm more interested in your questions, answers, and feedback.
So, first thing I guess I'll go first since I happen to post first, then we have Seth. And people feel free to sort of add in comments, or questions directly in the chat, and I'll kind of or just raise your hands. But, how can we actually start testing this tool in Medigov? Is it something that so this is a closed beta right now, and then we would you're gonna need to create, like, a Medigov community, and then we would be able to test it directly into Slack or in these calls?
So we have a an API, which is, like, basically, like, how, you know, like, when pods are created, like, you know, there is, like, a call that the app does to to this API. So we would take basically, like, to open that up to sort of third parties. And at that point, you know, you guys could, like, either integrate it with the tooling that you currently have. But what we are currently doing, yes, like, it's just try to find as many organizations as possible. We are, like, really in this, like, initial phase of, you know, we are not charging. We just, like, want to see how how the platform works with real communities.
Okay. Awesome. So you you do have an API that's, like, relatively, you know, well set up so we can just connect it and run it if we wanted to?
So currently, it's not open to the outside, so it would take, like, a few weeks for us, like, to to do that. But, yeah, there is, like, an API.
Looking forward to it. Awesome. Imagine talking to people and making decisions. Seth, you wanna go next?
Well, you do an efficient job of answering my questions as they're coming. So I wanna replace the first two with, should I imagine that there are kinda issues that we shouldn't use sci for? They're too boring, complicated, divisive, nuance.
Yeah. Good question. Oh, sorry. Go ahead.
No. I mean, that's the question.
Yeah. Yeah. Yeah. So what we say is that it works best with open ended questions. So if you have, you know, a decision that you have to make, which is already, like, has a limited set of options, probably, like, you wouldn't see the benefits of Sy, because, you know, like, you could send a poll and and you would get your answer. I think Sykes shines when there are open ended problems, complex problems where now the crowdsourcing of ideas now really helps. And the second thing is about scale. You know, if you probably want to use Zoom. Right? Like, it's feasible to have a discussion on Zoom with 10 people. If you have 100, now, you know, like, Sai would probably, like, be the the the the best way.
Thanks. And one other question. I I kind of get you know, some of your other materials have, like, a neighborhood y kind of physical issues focus. And what you're sharing with us today is more of a digital online community focus. And I wanted to know, is this just kind of different basis for different audiences, or, are you kind of seeing one as more strategic, more advantageous than another?
Yeah. I mean, this is probably a good question for for Chris. But we started with the idea of, like, serving civic organizations. So, like, really, like, local communities. And then, you know, like talking to to to Chris here who who who is way deeper than I am in in Web three, we really started making the connection with Web three communities that have this problem, like, you know, really, really, like, now. And I am actually, like, you know, at the moment, like, we are really in the exploration phase. So we just want to see, like, which of the two applications, you know, have maybe, like, more benefits in using it, which organizations would stick around for longer. All these sort of things are things that we want to figure out with these, like, early adopters pilot sites.
Thanks a lot.
So I think actually, Matthew, do you wanna go?
Yeah. Thanks. Really, really interesting stuff. I I have a question, I guess, just about information, both in the kind of the pipeline. One of the things that I noticed is that at the beginning of the pipeline, there's a question about how the the original request for information is made and the conditions that are made on that request. So I saw in the prototype, and maybe this is something that's modifiable depending on the community, that it was a 140 characters or maybe a 140 words and thirty seconds worth of audio. Right? And so if you're looking at a complex problem like police violence or something like that, like, it a lot of the solutions or the necessary considerations are multidimensional and probably require a much longer format for the informational display. And so I guess I'm wondering about, like, is the idea that over time, that kind of dimensional complexity around a problem is constructed collectively, and and how is the how does that integration actually how does that work?
Yeah. Excellent point. So currently, yes, there are, like, these, like, constraints, but it's completely customizable. So if you're an organizations that maybe wants to, you know, engage for longer discussions, You can change anything. You can change, you know, the number of characters, the pitch length, the the pod size, as well as like the length of the discussion. But sort of going forward, like, the, you know, what we want to really create is basically, like, accommodate for these multidimensional problems. So how do we do it? The the main intuition is that when people, you know, have to suggest ideas, the initial proposals, they would probably be, you know, very similar to each other, probably, like, not very deep. But through the discussion, basically, allow for these ideas to kind of, like, mix and maybe, like, ex get expanded. So for let's say, if I'm in a pod with you and you criticize my idea and you say, like, you know, I would vote for it if it had this angle. Now that, you know, I can go and modify that idea so that we can sort of collectively expand it and and and basically add that dimensionality. Currently, it's not implemented because it's probably like a a a design nightmare, but that's the idea for the future.
Max, do you wanna ask your question?
Yeah. Sure. Hey, Nicolo. It's been, like, four years.
Oh, yeah. Know.
The question was, you mentioned it seems like the sorting is random right now, but there's, like, opportunities to do things that are more like you could pair people who are have disagreeing stances or something and make them vote off each other. Have you thought much about that dimension of different types of sorting?
Yeah. That's actually something that has been looked into by political scientists. So how to create these tables for situations like citizen assemblies. It's very early researched, so there is, like, no consensus yet. But that's, for example, something that the AI part could could help sort of see, you know, which which pods are more successful. Are the ones that, you know, when you put people together that think alike, or when there is, like, a level of disagreement that can be resolved. Or versus, for example, I put together people that are, like, on complete polar opposites. So all those things, you know, it will be only after, you know, we collect enough data that we can make make a decision.
So I have a question. And it's probably if I vaguely remember having this similar conversation when we last met, Nikola, but maybe this is a good place to rehash it as well. You know, in some ways, like, especially, you know, with respect to, like, goals about, like, you know, aggregating and sort of integrating information, it sounds like, you know, there are similarities between your platform and, like, Polis. And Colin, you know, gave one of, like, like, I've I think last year, like, gave a message on Polis at the seminar. Could you speak to some of the similarities and differences in how you sort of, like, think, like, in what different settings and sort of what different advantages, you know, your design decisions as part of SAI would do compared to kind of a a tooling like Polis?
Yeah. Yeah. Absolutely. So, by the way, like, I I consider Polis to be, like, really like the the the cutting edge. Like, sorry. It was not in the in in the platforms that I shared, like, in my slides. But there is, like, a a a big difference, which is policy is great for, like, really mapping the idea space, like the opinion space, but you never interact with other people on policy. Like, you never discuss. So, you know, you can be part of, like, a cluster, but, like, it's very difficult, like, just by interact interacting with with police to, you know, change your mind and say, actually, you you convince me. What we do with Sai is actually bringing people together and expects people to be, you know, to be swayed by arguments. I guess, like, that's the main thing. It's, like, just more social, and there is a level of sort of information integration that happens, like, within people rather than at the system level.
But is that a realistic outcome? Like, you you can I mean, there's all these studies where, like, political scientists bring a bunch of people into a room, possibly even for, like, you know, a week? Right? And, usually, like, there's some sort of, like, description at the end that says nobody really changed their minds. So, I mean, they kinda like, oh, we are now kind of have a better understanding of each other. Yeah. It's not that, like, I changed my opinion about, you know, Donald Trump or Hillary Clinton or something like that. So I guess, like, is it possible that, like, you know, on certain kinds of questions, it is more you we are more likely to see, like, opinion change, and those are the things that, like, this is, like, size kind of, like, in set up for? Or Mhmm. I'm kinda curious.
So, like, a couple of things. So the first one is, you know, even if I don't change my mind by I understand the other person's perspective, that's already a win. The second thing is is that, yeah, absolutely. You're right. We have talked, for example, with the the president of the Democratic Society in in The UK that organized this type of events with with real communities. One thing that stuck with us was that when you have these exercises, people come in sort of voting always, like, for the, you know, the the most popular ideas. Like, you know, like, I want, you know, less garbage on the street. Like, I I don't want more in in in the world, but maybe, like, not very feasible. But after the the discussion, after, like, you know, having to argue for or against an idea, what what they have seen is that people tend to converge to the ideas that are more nuanced. They're, like, more in the middle. And the reason is that those ideas are actually the ones that offer a best compromise between people that don't change their mind. So in a way, like, we hope that with our platform, like, those ideas that are a little you know, like, if you look at it, like, from the police map perspective are kind of, like, in the middle. Right? They between cluster. They can sort of compromise between clusters.
So maybe, like, kinda slightly different question or angle is, like, not all like, speak okay. Like, speaking in compromise, for example. Right? Not all decisions necessarily, like okay. I guess what I'm saying is, like, compromise is not necessarily always the best outcome in certain scenarios. Like, for example, like, if you look at, you know, like, standards processes or, like, rough consensus. So this is a process that, like, allows pretty much anybody to kinda, like, stop the whole show and say, like, here's an objection. Right? Mhmm. And it's not based on majority vote. It's, like, in some sense, very a little bit centralized, but also, like, really not. But, obviously, it's kind of designed to sort of create, like, this optimal standard that is not necessarily the result of, like, compromise between competing positions as so much as, like, achieving some sort of, like, along some sort of technical criterion. So now I'm not saying, like, this is this is a setting that, like, Psy will be used for, clearly not. But I can imagine cases where, you know, you want to refine not necessarily just the medium position, but, like, try to identify, ways of getting to, like, truly the optimal thing. Is there are there ways in which do you have any experience in which Sai could be used to facilitate that? Or really, like, I guess I'm just thinking, like, are there ways of, like, configuring Sai? Like, what what are the ways of configuring Sai if I'm in different situations where I, as the designer of this, like, voting system or this, like, sort of discussion system, wanna say, like, oh, maybe, like, I want to, like, you know, not so much emphasize consensus and compromise. Instead, I wanna discover, like, polarities and sort of these kinds of comparing perspectives.
Yeah. That that's I mean, that's a good point. I mean, we definitely, we you know, we're not saying that, you know, it's the silver bullets. Like, I'm sure there are, you know, specific areas or, like, specific types of discussions where, for example, you would not want to even, like, you know, to crowdsource it. Right? Like, you you don't, like, in certain, very technical areas, you probably, like, want to, trust the expert way more than, an aggregate of of votes. But, like, the only thing I can think of, like, you know, to to answer your point is I I mentioned, like, you know, in the slides, you know, in the road map, there is the sort of the gamification and incentive feature. If there is actually a it's sort of like a ground truth. Right? Like, you can show that a decision was provably right or wrong. Now those features, like, what what we want to implement with Sai is now having, like, a almost like a, you know, going back at the discussion and see who voted for that idea. So, like, if an idea was positive, you should be able to reward people proportionally to how early on they supported that idea. But also on the other side, so if an idea emerged as a let's say, like, as the compromise or as the, you know, the output of the system, but you can show that actually, you know, it was a suboptimal decision. You should also go back and reward the people that actually disagreed with it. So, like, offering, like, this kind of signals so that people start learning, you know, how even, like, to spot good decisions early on.
I the idea of gamification here sounds very scary to me. But, Seth, looks like you want so wanna say something?
Yeah. Could I so, I mean, you kind of accepted the a couple premises of Josh's questions. You know, this is a collective you you presented us a collective intelligence platform, kinda open ended brainstorming platform, but you accepted the premise of, like, mind changing, which isn't really, you know, idea searching. Like, I I kinda separate those in different stages. And you also kind of accepted the premise that Polis and Psy, you know, do the same thing. And it seems to me that they're built to just do different parts of a kind of pipeline. And, you know, toward that, I'm just if you let's say you had all the buy in in the world and all the glue in the world, and you could glue together a bunch of these different platforms that seem to cover different stages of the process. Because ideation is just one thing. With implementation, there is decision making and debate, and and I don't even know if there is. I mean, that's kinda a little bit my question. Like, if we had a whole full, you know, fully integrated ecosystem of tools for the whole, like, problem we're trying to solve, I don't even know, like, what the larger problem is that Sykes fits into. Can you kinda give a sense of, like, other tools out there and how you're breaking them down, assigning them to different roles and the kinda larger framework? Do you have any vision for that?
I mean, to be honest, like I mean, like, we have kind of reviewed, like, various, you know, tools existing out there. We started, as I said, like, you know, more with with Civic. And the reason is that, you know, they, at least, like, in The UK and and in Italy, they are mandated by law to actually consult the public. But what they do currently is, you know, like, they use, like, really poor platforms. And so, like, Cy was, you know, initially originated from, like, solving that problem. But, of course, like, yeah, online, like, we have now, like, you know, way more digital, you know, solutions, like including, like, you know, policy and and others. In terms of like, yeah, which ones would be like complimentary or like at what stages, I would see them. I'm not I'm not sure. I haven't like thought too much about it, but definitely like in terms of, you know, just Polis versus Sai, as I said, like Polis could probably like be as just, like, to map out, like, this this idea space. While Cy would probably, like, be more generative. And, like, we need to sort of kind of, like, try to get on the same page. Like, we just need to, like, find a solution that
can
what we call community backed. But I would be, yeah, very curious, like, to hear, you know, from you, like, what you know, where do you see your like, maybe, like, which, you know, organizations and communities would benefit, like, from one, you know, one portion of this pipeline versus the versus SAI?
I mean, yeah, I I I kinda wanna leave that left that question line as, you know Yeah. A good question. Yeah. Yeah. Yeah. Yeah.
A difficult question. Right? I could say. It was my convoluted way to say I I don't have unanswered myself.
I have a couple of questions and thoughts. Just in respect to this kind of this pipeline, I think one thing that's interesting about size, it almost seems as though it's maybe trying to do a a few too many things. Like, I think, like, the idea of, like, it being sort of focused on open ended questions and then this kind of flow through conversation and deliberation kind of gets you at a point where you have, like you're basically doing, like, signal capture. You're, like, narrowing it down into signals that can then be, like, placed into a voting tool. Like, you're sort of, like, searching for senses, like, choice sensing. And then with the tooling for voting, you have the opportunity there, depending on how it's designed, to do the kinds of things that Joshua was talking about of, like, having some sort of, like, complete consensus kind of mechanism where someone can sort of be like, no. Like, I completely reject this proposal, and then it gets sent back to the top of the funnel, and you have to go through, like, that pipeline again. So that's sort of like I could see that being one place where this fits. Because it seems to me anyways that, like, the end goal here is to, like, vote where like and then have, like, decision of that. But I think, like, the ability to see, like, what the voting record was and, like, what the choices that are there that are then available to then vote on is really, like, the kind of proposition for this software.
Yeah. Yeah. I mean I mean, that to to some extent, you know, like, the the options that people vote on are, you know, crowdsource themselves. Like, people are are like the ones, like, proposing them. That I mean, I don't see, like, why in, you know, if you have a different way of creating those initial proposals, like, you couldn't create them, like, with that separate process and then input them into the platform. And maybe, like, also, like, you know, going back to to Seth, I think, question. One thing that it's, for example, currently missing that I can see definitely, like, these other tools being attached to it, you know, would be useful for is with for example, once you have converged to that whatever proposal or idea, now what would you do with it? Right? Because, like, in the case of, for example, civic organizations, you know, they are the ones taking that idea home and sort of implementing it. But for example, with organizations like, you know, a DAO online or other other organizations that don't have this executive branch, you know, how do you now separate that proposal into implementable tasks? And so perhaps, like, your pipelines with Asana, other, like, task management tools, I could see, like, definitely, like, a a a good sort of complementarity there.
Yeah. I also wanted to follow-up on that. The the other thing that I think is interesting is this this step four of the kind of road map, like, this sort of, like, integration of NLP and ML, particularly, like, the NLP because, you know, one thing that I've kind of thought would be interesting from, like, a proposal perspective and kind of in terms of kind of trying to break the kinds of consensus that builds up over time with Vows is to, like this would be, like, one implementation. You know, you have, like, a Discord channel where people can sort of post their, like I like, they're kind of, like, casual ideas for things that, like, the DAO could could do. And then you have some sort of bot that was sort of scraping and pulling from that and then feeding it into something like GPT three and then producing kind of proposals of, like, the subconscious of that channel that sort of, like, using that as its corpus and then generating these proposals that are sort of built off of the collective's ideation, but are the kinds of proposals that they would never potentially produce themselves. And so you get the kinds of these unexpected propositions that are sort of in some sense is kind of it could be it could be tweaked to sort of optimize for the kind of thing that Josh was talking about of, like, kind of producing the polarities, or just kind of producing these, like, abstract ideas that, yeah, like, that would would not ordinarily come up in the kind of context that this is taking place in.
Mhmm.
So I'm I'm curious at, like, like, where you sort of see, like, that element playing in. And in some respects, like, to me, that's more interesting than the gamification stuff. Like, the gamification stuff sort of seems like almost like a kind of, like, financial considerations that, like, the the the software is able to sustain itself through some sort of, like, economic mechanism, and to also, like, keep people participating in this once, like, the the market is saturated for this type of software. And so it feels like a it almost feels like an unnecessary layer, whereas, like, I can imagine, like, this kind of mechanism with the machine learning generating the kind of sustained interest, for interacting with the platform because you're getting unpredictable results. And then, I guess, in addition to that sorry, it's kind of, you know, nested. Like, how are you actually dealing with, like, the the data that people are submitting? Like, how long does it last on the the platform who has access to it? Like, does it is it portable off platform, and how is it kind of how are you kind of accounting for the the biases that come with data management and process?
Yeah. Okay. The yeah. It's it's a multifaceted question. So let me try to do it to unpack it
a bit.
No. No. It's okay. So the the, like, the first thing about, yeah, incentive, your rights yeah. Gamification is is definitely, like, you know, not perhaps, like, the most interesting part. But rather than, like, really, like, just keep people engaging engaged, The main point there is actually to provide them with feedback. So a way for them, like, to learn from from from the past. Like, it's it's a this is like you're creating, like, a like a learning social system. The second thing about NLP, I I mean, I personally love the idea. The only problem with that is that now the AI is part of the decision making process. And so it's very difficult to, you know, if a bad decision comes out or, like, if a proposal that has been proposed by GPT three and not a human gets selected and it turns out to be terrible, it's very difficult to, well, first, understand why the system chose that proposal. But, also, you know, in terms of, like, accountability, it's it's it's it gets a bit dicey. And the third part sorry. What was the third part of your question? Oh, data. Data yeah. So the the we are working with the open data institutes in London to basically understand, you know, how start from very early on, like, good ethical sort of data practices. The consensus that we seem to have reached is that the best way to do it is just, like, give that sort of power and up and decision making power to the organizations themselves. So if the organizations, you know, wants to have their data sort of, you know, fleeting or, like, being removed after every discussion, that should be up to them to decide. Yeah. I I think that's my sort of short answer. And
Great. On the on the second part that I could almost see, like, the GPT generated not being things that, like, people could actually vote on, but more just kind of, like, suggestions or kind of, like, kind of, like, resonances that, like, might spur people to actually put it to voice themselves. And because there's, like, this question that, like, if someone, like, is kind of influenced by the suggestion, then, like, where is the the the agency, there? So it's a little more of a muddled ethical question, but not having the thing that is automatically generated be the thing that's voted on, but just these kinds of, like, external, considerations that, someone could develop if they were so inclined might be another approach there. But Yeah. That's your point. Thoughts.
Yeah. And by the way, yeah, Joshua, I I agree with what you put in the chat. That would be very useful. Sorry, Steph, did you have a question?
I wanna make sure that before I keep asking questions that no one else who hasn't chimed in yet doesn't have something queued up.
Yeah. Fair point.
Well, I'm not seeing anybody. But, yeah, the floor is open. From my part, I'm getting is there a coincidence I'm seeing kind of sociopathy kind of vibes? It's striking to me the small circles kind of focus. And it also jumps out to me. I had a chance to talk to a sociology consultant who came by here, and they were talking a little bit about, like, problems or downsides of sociology, which is that it does struggle a bit with scale. Like, it's great in the low hundreds, but if you have a problem in the thousands or tens of thousands, so it's actually can't handle that. And I admire it. I think it's a pretty clever approach to lots of problems that seems to work for the communities that adopt it. But what jumps out to me is that Sykes kind of seems to be, you know, if if if you hadn't brought it to us, if they had brought it to us as their solution to scaling in sociocracy where you start with 10,000 people and use the same circle mechanism in a sort of random way to to generate ideas that then sociocracy, you know, at its scale of hundreds of people implements, you know, I I I wouldn't have flinched sort of. So it's sort of almost an uncanny kind of connection. Was this just coincidence? Have you been influenced by them at all?
I I actually to be honest, I don't even know, like, what sociocracy is. So, like, at least, if you can, yeah, point me to, like, some references. Yeah.
Yeah. I think so yeah. So they're they're kinda underground because they were proprietary kind of consulting scheme for a while, then some like, replicated and open sourced the the secret sauce to provide more people access. It's also called I think in The UK, it's also known as Holacracy, maybe.
Oh, Holacracy, I'm more familiar with it. Yeah.
Okay. Okay. I think holacracy is one of these forks that doesn't change anything. I think it's like a fork of the or something. I'm not totally sure. But, yeah, let's let's focus on self organizing circles that have a deliberate relationship to each other, that has certain internalized norms to keep things moving, and that has a very, clever approach to change where it's you try to be very pro change. You try to resist the the natural thing, you know, liberal or conservative, but there's a resistance to change often. I I'm making it easy, little three month experiments and so on. Yeah.
Mhmm. Okay. I'll I'll I'll look into it. It seems very, very relevant. Okay.
Yeah. It feels like it's it's it doesn't feel quite like sociocracy because I feel like sociocracy kind of implies a a longer lasting relationship in the community. Right? Whereas this is more directly just like a sort of sortition kind of kind of system. Just happens to be, like, bubbling up kind of sortition rather than all at once.
Yeah. It has the social the socio part of sociocracy, and it's really focused on face to face human contact. And but you're right. This is I kinda see this, like, if they were related, this would be, like, a step zero where you whittle down to get overall strategies for a group, and you whittle down to find a 100, like, really excited people out of 10,000 kind of people who have opinions.
That makes me wonder. Like, if you were approaching this, like, kind of work more modularly, so you have, like, a step in which you're like, okay. I'm I'm gonna create these pods or these bubbles through some sort of, like, sortition mechanism. And then okay. Well okay. You Sy has this kind of voice part. And then there is the, I guess, like, the question of, you know, is the community supposed to last a long time, or is this sort of, like, more like a structured interaction hosted by some other entity? Mhmm.
I mean, I I wonder the time scale kinda relates to the function. Right? You don't need to exist for years to do ideation. You do need to exist for years to get stuff done. And so so Socrates is like the, you know, a circle approach applied to the implementation side, it seems to me.
Would would it be stronger if you what's the right word for this? Like, if you could tie in your idea generation.
I mean,
clearly, these are in some sense, like, usually based in existing communities. Right? You know, like, you're running Scion Bankless. You were signing you wanna run Scion okay. Uniswap doesn't really have much social trust whatsoever. But, like, you know, like like, for the civic organizations that The UK is kinda running for, like, local kind of, like, communities. I imagine that there is some level of, like, community trust and sort of, like, identity there. Right? So you are working with, you know, know, continuous communities in a sense. You just don't know what are the social groupings in there that already exist that you could like, if you knew that, you could import that presumably into your pods. Right? Like, do you think that would be stronger if you could do that?
I mean, that's something that we have been discussing. Right? Like, it's, you know, one of the advantages of using audio, and you can see that, for example, in the in the gaming community, this is something like Chris mentioned to me. You really have, you know, these bonds that now starts to be created, you know, these links between people in that community, which means that you can kind of, like, make the community a little bit tighter. But what happens is that, you know, like, maybe, like, you have this very nice conversations with somebody, you know, from across town that you didn't know. Should we allow them to now form a a a friendship link, for example? Could they, like, one, you know, ask for for their friendship? Or Facebook. Or yeah. Exactly. I mean, the the the negative the downside that I see there is that now you are going to recreate, like, those, you know, homophobic clusters that you see, you know, online on, you know, on Facebook or or Twitter, where people start to, again, bonds with people that, like, think alike. It's to be debated, but absolutely, like, if, you know, one way, for example, that would be very interesting to one thing that would be very interesting to to study is whether, for example, pods that are composed of people that know each other already and have worked in the past work better than, for example, randomly assemb assembled pods. And if we, you know, create that sort of we allow that freedom, now we can sort of start making decisions based on on data, based on evidence. Right? But, yeah, until I and until we have that data, I think this remains a bit, like, you know, a more academic speculation.
Yeah. Fair enough. Fair enough. Yeah. I just it would be just really awesome if even for, like, these small level of experiments, if you could integrate somehow the data, what either through, like, continued past experiences interacting with the same community or by sort of importing external datasets, so that you could sort of, like, enhance your or perhaps, like, sort of work against, like, these, like, this, you know, hollow was it homophilic?
Yeah. Homophilic. Yeah.
Kinds of tendencies at least You
mean, like, data from, you know, like, what are the platforms?
Or
Like, suppose, like, you know, the same organization ran on, like, a policy instance. Right? Or, like, policy has, for example, Facebook integration. Or it was on. Like, suppose this was integrated with as as as an app, or it was you know, there's nothing nothing quite like this on block chains. Well, actually, that's not true. Like, something talks a lot about is, you know, because, like, all the addresses are public, you can see which addresses are participating, which DAOs. And if you run one thing in one DAO, you could sort of technically use that data Mhmm. Or pivot it to sort of see, like, oh, there are, like, these overlaps. What can I do with that?
I mean, it's it's a
little bit more tricky. Probably not, like, the most important feature for, you know, for Sai.
No. No.
No. No. I I see I see the point. Yeah. Yeah. So, like, kinda kinda bootstrapping, like, on the data that the organization already have to Yeah. Like, kind of improve the the the process. Yeah.
Yeah. I think there's also some clever ways of improving, like, the user experience for these communities because it's just like, it's hard enough to it's just hard to start conversations. Right? So whatever you can do to kind of, like, generate that spark or get people kind of more comfortable into the mood, I think that's, like like, at scale, that becomes extremely helpful.
Yep. Yeah. Yeah. And that's why I think, you know, like, integration with with other platforms, with, you know, Discord or any any other thing would be super valuable. Right? Because now people don't have to put too much attention. Like, it can be part of their normal routine. Yeah.
Okay. I have lots of questions. I we actually have two, like, questions in the docket. But it is already at the hour, so I wanna sort of, like we haven't been doing this lately, so I wanna get back on track. I would ask everybody in this call to briefly unmute themselves. And at the count of three, we are going to clap for Niccolo in his excellent presentation and the awesome discussion. So 32, one, unmute. Thanks so much. Thank you. Thank you. Alright. Looking forward to the
next meeting.
And, Nikola, are you in the Slack?
No. But I would love to
I'll send you an invite, and, hopefully, we'll people have additional questions, you can post this in the hashtag seminar channel and be able to sort of continue these conversations.
Absolutely. Alright. Thank you, everyone. Thank you, Joshua.
Thank you. See you, everyone.