Sifchain + BlockScience
Conversation Between Co-founders
This seminar is a part of a series of conversations by the Sifchain team with industry leaders. You can view the video here
BlockScience provides strategy, research, and analytics for economic systems which emerge from the incentives and mechanisms implemented via blockchain networks and smart contracts. The team is founded and directed by Dr. Michael Zargham.
Below is a transcript, edited for readability.
Sifchain
So Zargham, it’s really good to connect and we’ve got a pretty good audience near 76 people. And I see some familiar faces in the chat. This’ll be awesome. I think I want to start just by thanking OmniFlix, which is another really cool Cosmos project that focuses on media distribution, who is helping us. We hope to get a lot more of these out through our networks. And Zargham, thanks for being willing to chat here on a Saturday.
For an intro, I’m Jazear and I am managing Sifchain. We’re kind of having an open discussion and a bit of a preview of what’s to come, because Zargham is a really great guy. I’ll let him introduce himself, but he’s also helping us build out Sifchain by giving us advice and guidance on economics. There’s a lot of cool stuff that we’d like to do. On the flip side, Zargham has a lot of cool stuff that he’s been working on which we want to support as well. So there’s a lot going on here. Anyways, Zargham, why don’t you formally introduce yourself?
Blockscience
Cool. I’m Michael Zargham. I am an engineering research scientist. I work on collective intelligence, dynamic optimization, and multi-agent systems. I’ve been doing that since the early 2000’s. Actually, I studied undergrad at Dartmouth and worked with a researcher named Rezo Fadi Saber, who is an expert in multi-agent control and sensor fusion. I did a PhD at the University of Pennsylvania with Alija Babine continuing to work on dynamic resource allocation policies. Basically, that whole research arc led me eventually by way of some business process automation and real-time decision-making systems, to crypto. And in crypto, I’ve been working on the economics layer. The automation and control aspects of the dynamic resource allocation problems embedded in lots of these open systems are open games where the players make their own decisions. They bring information through markets. Generally, we look at both the governance and economic aspects of these systems as sensor fusion or signal processing problems, where the participants actually bring information into the system and the protocols help consolidate or fuse aspects of those systems.
So I’m thrilled to be here today. It’s an exciting project. And, to Jazear’s point, I also work on Commons Stack and some other projects in the crypto open source world, largely focused on experimental organizations and governance — in the presence of these sort of economic primitives and that kind of reintegration of economics and political economics in these social systems.
Sifchain
Cool, cool. Thank you for the background. I think the way I’d like to do this is, kind of steer through — you and I have had a lot of really cool conversations and we’ve always said, hey, wouldn’t it be cool to just broadcast one of those? So I want to have that sort of feel here in that we will go into my mind space and yours. For me, one of the things that I think is really cool about block science is this idea that there are a lot of pieces to it, but I want to dive into the term computer aided governance to anchor us. So from the Sifchain side, I wrote out an article describing that there are a lot of these market failures that occur, where people in the market are not necessarily informed enough to make decisions that are in the best interest of them either individually or collectively, and thus, they may see some adverse consequence.
We list a few examples, for example, Ethereum class APA, double spending multiple times, or one of the DAOs where, like, they actually had it to GX out, where they had more money in the treasury than there was even a circulating market. People just bought all the tickets off market and voted themselves to take the treasury and run. There are all these weird market failures that we see and then computer aided governance is a way to mitigate some of those problems by having the market do what’s in its best interest without necessarily relying on just the human intelligence of the people in the market. So that’s my understanding of it. I’d love for you to describe how you see it.
Blockscience
Yeah, sure. So, I mean, I think the first thing to introduce to talk about computer aided governance is this idea of a governance surface, which is the set of variables — parameters are like decisions that the governance mechanism is empowered to undertake. There might be multiple levels, like some things where you have an explicit vote to change a parameter, but there might be other things where you need to have a procedure for quote, unquote, amending the constitution, literally deploy a new contract or an upgrade or something — open-ended change. Once you start to understand the set of things that governance can do, you can work backwards to ask questions. What’s the shared understanding of the consequences of those changes? For us, at least at BlockScience, one of our emphasis is computer aided governance — that if we have models and analysis and some tools for building a shared understanding of the consequences of changes, you can have a more informed discussion, even amongst people who are making the decisions about subjective trade-offs implicit in changing a parameter or possibly deploying a new contract.
So, this mindset is one of — if it’s a sufficiently complex and computational system that having tooling that allows you to visualize or explore or run counterfactuals helps you, have an informed discourse about the consequences of changes. And an example of this is one of my favorite DAOs that I participate in. We’ve been doing modeling and analysis of changing one of the monetary policy contracts, and one of the earliest community members built out some models to look at the implied consequences on their current treasury. That the proposal to make the changes to the core contracts is now out and being circulated in the community — at least from my perspective, that’s how I think about computer aided governance. Maybe the extreme point of needing this would be an example like MakerDAO, where the parameter set of governance is so big. It’s almost impossible to untangle the consequences and potentially unintended consequences of changing one or more parameters, given the large numbers of collaterals and multiple different parameters on the collateral dimension.
Sifchain
Got it. Yeah. That doesn’t make a lot of sense. To reiterate, basically what you’re saying is you need to think about governance surface, which is, what can you govern and in what ways can you influence the things that you are governing. And then you want to use computers to be able to simulate what would potentially happen if you were to change some parameters. And then you have to, from there, pass that information around to the governors. Also, they can decide based on their understanding of the model, which is obviously more simple than the actual act of governing. This makes sense, is that fair to say?
Blockscience
Yeah, it’s fair to say. We do office hours on Friday afternoons. It’s a working group based DAO. There’s a central treasury and there’s a conviction voting, which is one of the original Common Stack research projects. In a way, One Hive is a part of the same family of organizations, because they’ve been doing a lot of implementation. Even Common Stack’s use of conviction voting was actually built by the people who make up the One Hive DAO. It’s actually kind of cool to see the interrelations between these organizations because they’re distinct, but then even within those organizations, you’ll find working groups and sort of largely, you know, DAO level funding decisions for a working group and then a working group being largely autonomous within that DAO conditioned on the funding they secured from the slightly larger organization. It’s definitely one to look into, but yeah, I’ve been involved, but largely as sort of a designer, advisor type, because they were the original implementers of sub-mechanisms that I designed.
Sifchain
Awesome. You know we’ve been thinking a lot about how to set up a DAO ourselves, which is what we’re calling SifDAO. It seems like there’s some overlap here. SoI think you and I should connect offline about One Hive or other working groups because, the way that we imagine it we really do want to get some high quality thinkers being thoughtful about the future of Sifchain. A lot of people who are in crypto have different time horizons, different understandings of governance, and so forth. So it’s important for us to target communities that are very thoughtful and bring out a lot of value. I also think we’ve been trying to figure out how to handle the funding model, right? Like how is it that you actually allocate resources across groups in an open source way? So we’ve tried to learn from pure open source non- crypto groups like Python. Then, some other DAOs that exist that are very famous like Compound or Maker or YFI, but One Hive and some of these other groups seemed like a good fit at least because we’re already working together and chatting. So, yeah, this is something we should explore more.
Blockscience
Cool. I think the other thing you brought up that I want to highlight is the difference between the stakeholders within the system and their time horizons. Because we talked about a governance surface, which is a system level, I guess, control surface — I’m a control theorist formerly, where you can make a decision and influence something. But an individual participant in the system has a different decision surface. Like a trader or someone who’s adding and removing liquidity to a liquidity pool, they’re essentially exercising a different control plane or a different decision surface. So not only do they have different objectives or different goals, they have different time horizons for their goals. They can be evaluated or interpreted in terms of how they use those different mechanisms. What ends up happening though is, in practice, you can’t disentangle the roles.
I might be a participant in the world of Sifchain and I might be trading on one hand. I might be adding liquidity on another hand, but I may also be participating in governance decisions. So, in a sense, I’m actually exercising three different decision making surfaces, with different sorts of rules associated with them. The goals of those mechanism designs might actually be in conflict at times. For example, between the person who’s adding and removing liquidity versus the person who’s swapping in theory, a swapping person is paying fees that accumulate to liquidity providers. There’s a flow between those two roles. Furthermore, someone who’s setting parameters at the pool level, or even at the system level, might be making decisions about a risk reward tradeoff or a time preference tradeoff. And not everyone who participates in that system might have the same preference, but if it’s at the level of a liquidity pool or at the level of the chain, those decisions have to be some sort of measure of senseless central tendency of the participants.
It’s not obvious what happens through any particular voting mechanism. In a way, you can envisage the voting mechanism in the governance process as a way of attempting to synthesize that spatially and even temporally distributed information about the preferences of the participants in that system.
Sifchain
This makes sense. It brings to mind the idea of a protocol politician, right? There’s this idea that someone can look at the different roles that people are playing, and obviously overlap among them, and then look at the distributions of individual and then collective participation in these groups — and try to synthesize or understand what it is. I’m wondering what you think about this concept, this protocol politician concept.
Blockscience
I feel about it exactly how I feel about the concept of politician, which means that a good faith politician, one who generally goes out and tries to make sense of the world they’re looking at and to act on behalf of the constituents of that system, is awesome. I worry though that by default, the incentives really aren’t set up for that because in a world where the complexity is high and it’s difficult to demonstrate that your point is genuine or that you’re pushing people in the direction that you really believe things are going rather than in the direction that you see is in your best interest as a politician, it’s really hard to disambiguate. It kind of brings us back to this idea of computer aided governance, where what I would like to see these protocol politicians doing is being actual data scientists of the sort that could take the models and tools and data, and actually make a case for the thing that they’re saying and to do so in a scientific way that they’re, like, essentially peer viewing each other and saying, well, I believe these things.
Under those beliefs, we would probably want to do this. We would expect outcomes of this type that — if I assume this, then we might expect that. I think protocol politicians could be wonderful, but I don’t see the first wave as being inherently scientific. They’re mostly starting off as people who have a gut feeling about how the system is going to behave and thus arguing on behalf of what they think that we should do. Often in the interest of, you know, capital accumulation, which is sometimes at odds with long-term sustainability. And so, maybe to close this comment out, my point is it’s in the right direction, but I’d much rather have an evolution of the role politician to something a bit more scientific in nature where the expectation of those protocol politicians is some ethnographic understanding of what the constituency wants and needs. And then the ability to advocate and effectively peer review each other to form a sense-making process that informs voting rather than having something a bit more like memetic warfare, where people are just trying to induce people to believe something, whether or not it’s actually got any scientific backing.
Sifchain
I think that’s very well put. I have a couple of comments. I like the concept for a lot of reasons that I’ve always imagined, like basically doing something where you take a desire from a community and then you bring it to a group of data scientists analyzing our protocol, and then they make suggestions. And they are betting in some way that this makes sense based on their understanding of market behavior, because if they truly believe in their analysis, then they should be willing to put money behind it. But obviously there are always unknowns. And so, they would want to hedge that risk against those unknowns. So, some of them may want to say, hey, I want to have so much that I don’t really want to participate, but I will give you answers, and feel some of them would say, hey, I’m really interested in hedging my positions.
That I’m basically going to be an investor. I might not even tell if I choose to share my thoughts publicly. You are lucky because I’m so focused on alpha. And I think that there’s probably a sweet spot, which I guess was my second point, but I want to turn things over to you soon. Like, there’s probably a sweet spot where you’ve got these protocol politicians who are data backed, maybe they’re being paid in the token. And so, they want you to get rise maybe, but to another extent, they should also just be thinking about their professional reputation, their search for truth.
And those kinds of things. I think, with a lot of the people you work with, a lot of people that we’re bringing on to Sifchain, and just with the way that crypto is evolving, I expect this class to be more common.
Blockscience
Yeah. I think there’s a lot going on there. I’m going to try to break it out a piece at a time. The first thing is, I am a proponent of the sort of plutarchy concept. I think that it is more implicit and less explicit in practice because I see a lot of people kind of staking their time, money, energy, and effort behind what they believe and building that thing. Rather than having explicit markets, I think there are implicit markets that really, you could argue that what happens is, driven by where those, say, long tail high-performer people who are willing to take their time and effort and maybe forgo that FAANG job and just dive in and do something that they really care about and like manifest something new, then to be clear in the crypto domain. Like, you can absolutely profit from that.
Blockscience
If you can manifest the thing that you believe in. I think my comments about this being more implicit than explicit comes in part because when I look at the way prediction markets work, I sometimes feel like they’re not great estimators in a similar way to that. I see the even just like stock markets being not great estimators, they tend to emerge towards you know these selling points, which are people’s beliefs about other people’s beliefs about things. Unless there’s an absolute bit at the end that says this succeeded or failed, it can be really hard to truly, sort of marketize something and like bet on it. It becomes the nuance about how you interpret the end point. It kind of comes back to this problem of oracles and human resolution of, did the thing happen. And there’s something, it just, it kind of always comes back to the subjective judgment at some level, except in really extreme cases.
Blockscience
And as a result, I think the implicit form of it often ends up being more powerful, even if it’s harder to see. Oh, and I’m going to kind of jump past that now, but I bring it up because I, I like, I have a weird, like, mixed feelings about it. Like I really like it in its abstract form. Like I think it’s a good description of how things move forward, but I still struggle with its attempts to be implemented explicitly. I want to jump to one of the other topics you brought up though. Go ahead.
Sifchain
Oh, well, if you want, the last one first, why not? So, yeah, I, I have that weird feeling about plutarchy as well. Like there’s always this thing in the back of my mind that says, you’re basically forcing someone to take a position that like, it’s just annoying, right? Imagine that someone’s like, hey, I’m going to shoot this three-pointer and you’re like, no, I don’t know if you can do that. They’re like, want to put $500,000 on it? Like, it’s just, it’s like, yeah, I’m just making this comment because I have an intuition, I don’t necessarily want to put like any money behind, it just feels there’s an inherent friction. And I’m also reminded of, the last election we had where there was that one Trump, like position on FTX that you can still trade after the election, but before the inauguration. So, it’s just like, there’s, I don’t, I find that people really, there’s a difference between money and utility, right? Utility and economic and microeconomic theory is like the perfect thing that people want.
I call it, I like to say, because I studied economics, like microeconomics in particular, a lot that you have to realize there is a cost in utility for converting money into utility and vice versa, which is, happiness or money or pleasure, or like, hard tasks that are challenging, whatever it is like that’s utility, you always want that. Money is one thing that can be converted into utility, but, the, you can’t necessarily force people to just always constantly engage in monetizing every bit of their brand.
Blockscience
In fact, it’s counterproductive in a lot of settings. A couple thoughts along this, the first is that when we take this framing, we have to remember then that utility functions are unique and private. In fact, even individually, we might not know what they are in some ways our revealed preferences and our actions are better indicators of what we actually care about than any explicit representation. So like in a way, and I’m a math nerd, right? So I’m imagining that there exists a function between sequences of events in the real world and some abstract utility metric, but not only can I not measure my own utility metric, I like don’t know what it is, but I sort of, I respond to it. I’m in this odd gradient descent, even though it’s mine. If we start to realize and like accept that, I don’t even totally know what my own is. I start to realize how ridiculous it is that I might try to impose on someone else what they should do because as I’m going through life, I’m responding to new, exciting ideas. I’m getting frustrated when I have to spend too much of my attention worrying about monetization. Like I personally think of financial, like, accounting as constraints, not so much as objective. In my world, utility is this abstract, weird thing that I can’t totally measure, but I can gradient descend and even sort of strategize over if I think, oh, I really want to accomplish that. Maybe I can go through steps to make that possible. But, the money side of the equation is far less of like accumulate and stack this up and way more of a, is this a constraint blocking me from pursuing the thing that interests me, whether that’s travel, whether that’s, doing a cool new research project, whether it’s collaborating in person with a friend of mine who lives in another country and I, you know whatever. I guess the kind of point I’m trying to get at here, kind of dovetailing back to your point about utility is that, the concept of utility is really powerful, but much like we just talked about, plutarchy trying to force it into an explicit form actually can cause more harm than good. It ties back into even reasoning about complete versus incomplete contracts. Like try to make everything a complete contract things go to shit because you can’t possibly represent everything that matters, a priority you need to like have some space. When we get into computational systems, it’s really tricky to design for the space.
Sifchain
Yeah. You’re right. You’re, you’re very right. I’m gonna to shift gears here. I can, I can give you another question, but I’m curious, is there anything that you want to bring up at this time?
Blockscience
No. I mean, I’m just kinda winging it. I think this is pretty fun. The concepts around plutarchy and around decision making and implicit utilities personally, I think this is really important to understanding the interplay between computational systems and social systems. So, just as far as like viewers, listeners are concerned, I generally recommend that as an area of exploration, not one where there’s an answer per se, but it’s a good place to explore the boundary that the interplay between technological infrastructure and social outcomes, both at the individual level, like what is it that I really want and how does what I do manifest that. And, secondarily at the system level, what is the system trying to accomplish and how do all of our individual actions become manifestations of our local preferences, our local utilities, aggregate or not into some meaningful information about, that distribution of preferences and accomplish some aggregate steering, which ties back into the field of cybernetics, which is not new at all and worth reading about because in a lot of ways we’re manifesting now things that people theorized and aimed to manifest you know 30, 50 years ago.
Sifchain
Yeah. I think, in another sort of group I’m in that’s doing research on, just like these similar kinds of problems. There’s this notion that like everyone has, multiple selves in their minds and you can either say you don’t understand your utility function or it’s really complicated, or another may say that you have multiple and they compete in the different games that you’re playing. I think that introspection can be very hard for some people and it’s, I’d say it’s like, there is effort required. It’s just, like you know any other thing that you can be more or less skilled in it, but regardless like there’s some level of effort. I think that like social networks, in the web era, people are very good at like, trying to understand your models, like your utility functions, and that’s how they do that. That’s basically what ad models are, right?cThey’re like, hey, we understand a thousand different demographic points about you. Therefore we will serve you up this ad, which will speak to something that you may or may not have been aware of. I am very excited for the idea of having a web three version where it’s like, rather than spending a dollar to get an ad in front of your face, I’ll spend a dollar on something that you really need, right? Like, I’ll just send you the dollar and what I have goodwill for you, go, either you spend it yourself or let’s spend it on something that like, because of your revealed preferences that I have from your financial activity, I understand is to your best interest. I’m wondering what you think about that.
Blockscience
Okay. Yeah. There’s a lot packed in there. First thing is like, there’s a degree to which the ad tech instantiation of this is really creepy. So like I would say I’m not necessarily enthusiastic about it as it exists. For two reasons, one the degree to which they can sort of identify the details of a person from the data that is available, purchasable, et cetera, is pretty narrow. Like I’ve even seen someone accidentally target a single person by restricting the stats they use to target small enough, which is kind of crazy. But to your point, I think, there is a lot of power in leveraging that for the person. But also it’s paternalistic because if you look at the way that something like Amazon works in the collaborative filtering algorithms in the background that served this sort of… Okay, you liked these things and other people with features like you like these things, and you’re able to do these, matrix product problems and ultimately, recommend things. Those systems become performative. If you look at, and again, the example was Amazon, you actually see a reduction in the diversity of the things that are available as you funnel the population towards some things that are sort of optimal in this reduced sense. So you actually, highlight, exploit over explore. And, one of the things that I see in web three that is opposite of web two, is that it has a much more explorative modality as opposed to an exploit modality. I mean, there’s actually locally, this exploited modality of take the information you have and make a decision, I guess, to clarify exploit in like sort of computer science terms, explore exploit or two paradigms and gathering information versus leveraging that information to accomplish something, not exploit in the sense of harm someone else. But, in general, I think that the data usage modality in web two is very exploit centric and actually kind of both meanings of the word exploit. Whereas I think web three has a promise of being more exploratory and the extent to which we do see systems capable of exploiting that information… It should be less exploiting of the people whose information is being exploited.
Sifchain
Yeah, I think the way I’m imagining it to simplify things for a web framework is, the ad company has, all of the data Excel has potentially more, but you own shares in the ad company. There’s a much different set of goals that, in a sense that number one, if everyone owned, if it’s a distributed ad company, you could collectively say, this is really annoying. We don’t all want to listen to only the number one person in all the top 200. We want more variety. What’s the thing is that I think that there’s in crypto, if I compare this to 2017 to 2014, 2011, which I’d say are the three, four sort of tranches of crypto markets. I’m really fascinated to see the diversity of things that are being implemented in crypto now versus in a success era. We’ve seen new business models tried and worked in, and there’s been a lot of feelings for the middle that’s come out of left field. My belief, my hope is that if this system is paternalistic, there’s an accelerated gestation process. Eventually, you’re getting people who can go out and build their own crypto enabled economy that does something that’s very different from the… what the models would have otherwise predicted.
Blockscience
There’s a couple of things that are in what you’re saying that I think are really important. The first is the state expansion of the design space, because it’s really a crypto has the programmability of sort of like institutional objects, like, businesses or cooperatives or DAOs or whatever. Like the, we have essentially made a highly programmable institutional infrastructure. As a result, you can spin it up real fast and you can even try it out and discard it as long as you’re not exposing other people to it say financially, but then even in the financial stakeholder category. We have a, there’s a really, a lot less, monopoly, like in the sense that you can opt out of things and into something else. So, one of the problems with the web two infrastructure is that we’re basically all just immersed in it and we can’t get out of it. Whereas in a web three infrastructure, at least in the way that it exists today, you see people like participating in an organization or a new business model as stakeholders, and then leaving if it doesn’t suit them. If you’re familiar with Hirschmann and sort the exit voice and concepts, I think there’s a really important role in exit. This like really explicit highlighting of part of your action spaces, leave, part of your action spaces, join. And, maybe only if you’re a member, do you actually participate in a governance process, but then, not just thinking about what you can do as a member of a particular, entity, whether it’s a, a Dow or a network or whatever, but also the choice is between joining things and leaving things. I think crypto has definitely supercharged this idea that we can evolve rapidly our institutional forms by banking. The, the design space grows the ease of participating in creating and trying out a new thing is been increased. The fluidity with people, with which people sort of exit things and then, or fork those things, or sort of move forward to try new things has been rapidly accelerated.
So I think there’s a lot of carcasses along the way. I think there’s also a lot of potential collateral damage. I mean, I’m not the hugest fan of the 2017 ICO boom, as great as it was for some projects, I would say, it was, overcorrection from the overly exclusive, overly illiquid VC model of funding. So you go to ICO’s, which is like a hyper correction in the other direction. And now we’re like trying to find some points in the middle that are more inclusive and maybe fair in different ways. But that not necessarily having the sort of extreme pump and dump dynamics that we still see. And honestly, they haven’t gone away, but at least my opinion is that, projects and just technologists, as well as, investors and regulators are, thinking about how to at least hopefully mature towards something that is. Maybe on a better parade point between me, again, this highly exclusive, arguably over-regulated, sort of VC investment sort of angel investment side of and involving yourself innovation versus, the I would say like under-regulated, extreme point, which was the ICO boom in 2017.
Sifchain
Yeah. Good points. I have a comment. Then, another sort of insight that I’d love your opinion on. The comment is that… but going back to this idea of these working groups of data scientists, or port protocol, politicians of any kind or helping to decide a policy for a particular cryptocurrency or community of some kind. I have an expectation that the, like those groups should be competing with each other internally in the domain within a similar, a single cryptocurrency. And also obviously with other data science hives in other cryptocurrencies. There’s some sense of regulatory capture, both in the traditional sense in that, if only large institutions can actually afford to follow regulatory, right. Also in like informal sort of , like abuse of the term, which is to say like these institutions to some extent have captured regulators and or regulators have captured the institutions. There’s an understanding that like, there’s some games that you can play with the public, which is collectively like our resource.
Blockscience
I was gonna say that those institutions, the regulators and the large institutions, they have a sort of co-fitness, there’s like a, if not a stable, pseudo, stable equilibrium in their game play, that is sort of at the cost of the remaining population.
Sifchain
Yeah. Yeah. The other thing that like this, kind of, oh, so I was going to say that this is connected to a biology, from formerly, from Coinbase now, doing this whole like network state concept. Where they’re trying to, pointing out that like governments like traditional nation, state governance, our governance is increasingly competing with each other, fighting… See the mayor of Miami, or some of the Texas politicians versus some politicians from a place like California. We’re seeing exit assist from say California to Texas or Miami, because these governments are very forward-looking or Estonia being another example with these digital E-residencies. And this should be helpful and it’s even more in crypto. The thing that I find really fascinating, and I want your opinion on is, with Crypto, you can actually be in multiple domains simultaneously, right? Like I’m fascinated by this idea, which we say is like an omni-chain sort of protocol that you don’t necessarily want a single token to go up. You want your portfolio to go up. And so I imagined that there should be synergies between cryptocurrencies. We have that very deeply with Thorchain and, with a couple other tokens actually. There’s the sense that it’s happening with YFI and sushi and a couple of other projects. I’m wondering what you think about this idea of portfolios and like, the idea that different cryptocurrencies can actually benefit each other financially and then logistically and how that might play into your governance.
Blockscience
Okay. Yeah. So there’s a couple layers there. I’m excited to work through them. I think the first one that came from the beginning of the discussion was your comment about the data science work, the protocol politician work being… I think that it’s actually kind… it’s hybrid cooperative competitive in that you should largely be cooperating on research infrastructure, but then competing in the realm sort of discovering solutions or even debating over, well, I think this assumption, you think that assumption, what are the consequences of the assumptions on the outcomes? So from a very scientific perspective that the open science layer on these needs to be like we work together to build, models and data collection infrastructure, and to tether these things as, infrastructure for science. But then the scientific protocol politician work is going to be really focused on the subjective aspects. You kind of build the objective components together, the machinery, the if then else stuff, but then you should be taking different assumptions or considering different prioritizations or weightings over metrics, or even arguing over which metrics are the ones that are really representative of people’s needs. There’s the decomposition of the technical infrastructure that you use to advance your practice as a group. But within that ecosystem, you would expect to have some obvious disagreements and different priorities. You want to see that be the surface of the political game becomes about not about doing, saying anything that’s not true, but peer reviewing to ensure the objective components are objective and then debating over the subjective components, conditioned on the objective components. So I’ll set that one aside, but you brought it up at the beginning and I wanted to address it. I don’t know if you want to comment before we move back to the second topic.
Sifchain
Well, I, I pretty much agree. I would say that, I would comment that the competition should be ultimately in the pursuit of truth and wellbeing, not necessarily prestige. There are certain issues and, or like selfish profit, right? I see problems with the hedge fund model where you have all these brilliant quants who are doing the same calculations in multiple firms, because they can’t work together because there’s exclusivity and portfolio. Maybe there’s some concept of like an inclusive portfolio in crypto that I think is better for society because it causes less multiplicity of work. There’s other stuff I think, but I’d actually love for you to just finish your point.
Blockscience
Oh, so it’s going to tie it back to the second conversation about the multi-participation. I think this is actually another huge advancement that comes from what I’ll call the, we’ll call it the crypto political economy. The world where you can be a member of many things. So, personally I created Blockscience to create a home for my research and to build a system in which other people I like to collaborate with could be funded and we could learn from each other and we could design new stuff. But then I’m not like uniquely, even as the founder of that entity, I’m not like uniquely tethered to that entity as my sole home. I’m a member of one hive. I was one of the founders of Common Stack. I, we talked before, like I was, originally researching the Balancer project back when Fernando was working as a, a part-time researcher at Blockscience and he and Marcus and I did a bunch of derivations and like eventually led to its own project, which I’m a stakeholder in. And, just like exploring and actually being a member of many things. I think this sort of evidences what I’ll call, the, this shape of cyberspace, like the Discords and the Telegrams and the discourse forums and the GitHub repos and there’s this big digital infrastructure, much of it is still in web two. In addition to the contracts that I hold tokens in or whatever in say Ethereum or, in other blockchain ecosystems, there’s this sort of intersectionality of my identity in cyberspace. When we stepped back into physical space, I can just extend that model to think about my choice of where to live physically. I moved from Philadelphia to California. I lived in California for a few years. I moved to Arizona with my partner, after we got married, so a few years after, but, because she had lived there and her family is there, like this idea that I’m making my decisions and revealing preferences in it. It’s kind of going back to our earlier discussion, but the set of choices I’m making is intersectional. I am defined by my relationships to people and places and things and projects and, actually another nice tether to that is some of the more recent and ongoing research into this executional lens by Primavera, the Philippe who is looking at this inverted model of institutions where you think about them less in terms of roles and rights and more in terms of relationships and, networks, very network science in my mind, way of thinking about institutions. Kind of ties back into the example that you made about people being induced to move that, like, we’re acknowledging that we’re not like some, fifties company, man, whatever, where I live in the same town for my whole life, we’re actually becoming more and more defined by many relationships to many institutions, some of which are, work or digital in nature, some of which are social. I, I feel like that’s a really good way to reason about our individual choice sets says in this graph of relationships between, ourselves and other people and ourselves and other institutions, and we get to make opt in, opt out decisions, really important ones. That sort of in many ways define us, but also temporarily insofar as if we don’t like something, or if there’s a change to the nature of the thing we’re participating in, we have the freedom to sort of move or vote with our feet.
Sifchain
Yeah. Directionally, I agree with a lot of what you said, directionally, what a weird event horizon for me that I’m trying to understand is when I think about DAOs versus, like, hierarchical institutions, there’s a sense. If you’re at the top of a hierarchical institution, you should roughly know, or at least have a really good handle on everything that’s happening with you. Whereas in a networked institution, in a flat DAO structure, there isn’t necessarily that vantage point that anyone has, and no one’s expected to have that role. There may be people with more or less visibility, but you should basically believe in any DAO if you are the guy who started the DAO or whatever things should be happening, that aren’t necessarily that you have no visibility on, but in the cryptocurrency case, that should be like increasing the token price or market cap or whatever. Like it should be increasing the wellbeing of everyone in the group. I think that, like, as I talk about this notion of like… like you have in the physical space, like some kind of a forced, like the only one thing can happen, you can only be in one place at one time, right. As you can be in multiple digital groups at one time, and then your body, however, like operates as a DAO and that your brain works, but like your bottom part does many different things independently. I would like to see, and I think with something like, AI partners, I would like to see something where a human being is not necessarily actively deciding this is good, this is bad, but more like they’ve offloaded 10, 20, 50% of their decisions to AI bots plus people or whatever that are like making decisions for them in more of a DAO of self. You get where I’m going with this.
Blockscience
I want to zoom in on the idea of how we make decisions and just talk about it for a minute, because I think this is really tricky. We have this strong assumption, and it’s really natural to have a strong assumption that we’re optimizing when we make decisions. Something happened to me when I was in grad school, that was really compelling where I switched from thinking of hyper-optimizing, everything that I did to being more go with the flowy because I was studying stochastic optimal control. I suddenly had this moment in working on a proof where I realized that the optimal solution in the high uncertainty setting is often to just go with the flow. Kind of to some extent, you’re like only when you get from where you want to be, do you actually actuate and expend energy to reorient, and then you kind of flow again. And there’s a lot of problems and pretty formal mathematical optimal settings where you return to this almost Chad like, optimal solution where if your energy or attention constrained that… using that resource, that energy or that attention to decide or actuate is actually suboptimal.
We, when we realize this, we run into two problems. One is, that our notion that we can decide the optimal and program, it starts to fade away. Because like, how do you teach a computer to do that? Well, at some level you can do it. If you’re you really able to encode your optimization objective, and you’re really able to encode the constraints that you have and to use literally, stochastic optimal control as a sub-element in the derivation of something like reinforcement learning. So, it’s obviously something that’s pretty popular. Keeping in mind and coming back to this, you need to know what to ask the machine to optimize for. My biggest struggle is in moving towards a more optimal, AI aided behavior setting. I’m like, okay, but then like, how do I figure out what I want it to optimize for? Do I know and okay, but what happens if that changes? How do I know it changes? Do I have to wait until it does something I don’t like to realize I don’t like it? Now I have to figure out what to tell it to optimize for now. There’s like a lot of weird, interesting philosophical questions at the edge of… suppose I wanted a computer to make decisions for me. Because then I have to figure out what I wanted to do to make decisions. Essentially I have to extract that epistemological process and then bake it into a computer and teach the computer to exercise the epistemological process that I would like it to take on my behalf, but it still will only do what I program it to do. Even if that is some optimal response to outcome. Again, boiling back down to how I encoded utility, boiling back into how I encoded the constraints. Like what if I encode a utility function, but I failed to encode a restriction against something that I would consider inappropriate, or, maybe I’m maximizing for revenues, but then there’s a certain degree of exploitative behavior that I wouldn’t want to take. Well, if I want to constrain that, I would have to be able to tell a computer what that thing is in some mathematical way to avoid it, choosing to do that. Like, I’m with you from a very futurist perspective, I love the idea of increasing the extent to which we delegate decision-making to machines, where we tell them, I want you to accomplish these things. I want you to constrain yourself and to not violate these restrictions, and then you decide and we’ll run with it. But the edge of that formulation is always, how do you decide what the objectives are and how do you decide what the constraints are? And the computer can’t really do that part for you.
Sifchain
So, I have an order of magnitude improvement that I should discuss with you more another time. That is, I think coming to give credence to the idea, even though ultimately yes, you can’t get a computer to know every single thing you want. I decided to pop up the chat for a bit. We’ve got some good discussion going on there. Can I ask a question? Oh.
Blockscience
Yeah. Sorry. I just, I wanted to make one last comment on that because it wasn’t, it’s not that I’m not, I’m opposed to using automation and optimization for decision-making. My argument is more that the models matter a lot. You really have to decide and like, scope and say under these circumstances, I’m willing to enable machines to make these decisions. Like, it’s really about drawing the boundaries. It’s less about not doing it or doing it as being careful about the scope of control and like, what is possible within that scope of control that you’re passing to machine decision making. Cause I mean, honestly, literally that’s the whole thing that I do. I work on machine decision making, so I don’t want to be like, poo-pooing it and saying, don’t do it. I just take it with a very serious, a consideration, because I think it’s easy to overdo that like if you stretch a little too far and you don’t totally understand what power you’re giving a machine decision maker, you’re setting yourself up for, maybe outcomes. You never want it. Yeah.
Sifchain
No, I, I agree. I’m telling you there’s a specific implementation we can talk about. Cool. Yeah. Yeah. So, Mark, actually Mark Miller, really awesome guy. Maybe you can have a chat with him at some point. How does the question… what do you mean by go with the flow here? I’d like to maybe give what I interpret, and I’d love to hear yours, as well. For me, I always look at the DAO, this concept of the way, as it’s described in, but I also close this as literally labeled spam risk, so no benefit to answer that call. So, so I, I always think about that. There’s just the way things are, which you can never really know. There’s just a DAO of life. And, for me, like if you think of something like a, the concept of flow, right in psychology and positive psychology, where you’re at the perfect state of challenge and skill, you’re just in the zone… not necessarily thinking about things.
I also think a lot about trans state and states of consciousness and they had this idea that you’re in a state of consciousness such that you’re constantly just building and maybe not self-aware you’re maybe not. You’re in a good spot, like, however you want to slice it. I think that’s very important. There’s a lot that our biologies and our psychologies can teach us about what should be done, but it’s very much something that you meditate and you live life to learn from. But I would also say that I’ve learned that this issimilar in markets, right, when you watch money flow from one source to another, from one DAO, one DEX or one token or whatever it is, you can really see how the market is just like, I almost think of it as like resonating with a piece of information, right? The money is information. That information is very valuable, whatever you will, however you wanna slice it. Like, there is a sense in which flow in markets should be looked at in connection to the concept of psychological flow. That’s sort of how I capture this, but I’m wondering what you would say.
Blockscience
Yeah. I’m going to sandwich this with a really overly practical and then an overly theoretical, answer. The really practical example is that I was an elite level competitive ultimate Frisbee player for a decade. Flow was the main concept that we would use. In fact, the way that you would talk about this was that you try to induce flow in the sense that things are getting moving, and once they stop, it’s a lot harder to get them moving again. If you actually stopped long enough to think about what you were doing, you would lose the flow. You couldn’t possibly retain flow by over analyzing a situation because the very act of stopping to think about what you should do will sacrifice the flow that you had. So you wouldn’t be able to get the benefit of the flow if you were overly analytical about the flow, because by the time you had stopped, it was gone. So, that’s the sort of overly practical example. I also coach some, and it’s always tricky to get people in that mindset in playing and not overthinking or whatever. I’m very much just kind of sport side analogy. The extremely abstract side of this is that like, I work a lot with dynamical systems and representations of systems that evolve, their representations are differential in nature. They say how things change. I brought the concept up in the context of stochastic optimal control. What I really meant there was that the dynamical system was evolving along a direction that although noisy was, it was not clear that actuating or like doing anything would be better than just evolving along the differential equation as it’s going. No computation to determine what to do, the only computation was to determine, not to compute, to do, to try to figure out what to do. You’re just like realizing not to do anything and then not actuating and letting it flow. Except for under circumstances where you know that you’re like, you actually know that you’re moving in a direction, you don’t want to go, then it might be worth figuring out what you should do and then doing it. But like separating the idea of, calculating the best action from the idea that we should stop and try to calculate a best action at all. For me, going with the flow in a more mathematical sense means having some heuristics that say, don’t even bother trying to compute the optimal action, just let the system evolve the way that it’s evolving. You might, it might be worth actually expending the effort computationally or, attention, if you’re a human to actually decide what to do, but it’s like a harsh reg, a regularized there where you’re just like, yeah, well, the system’s relatively uncertain. It’s kind of flowing in the right direction. Let’s just go with it. Only when there’s enough information to cause me to want to analyze an optimal action or, something to do. Then I expend that effort. Then I take that action. I think a lot of people get stuck in, analysis paralysis in precisely the kind of situations where you should just not like just, you just go with it. The tricky thing about this paradigm though, is you don’t want to get stuck in either. You don’t want to never analyze something. Cause sometimes it’s really important to take stock and make a strategic decision, make a plan, or be rigorous about it, about a particular decision. I try to use this concept, and this is why I say I had this sort of moment. I was in grad school doing my master’s, which was a lot of hybrid systems and stochastic, optimal control robotics stuff. I was playing lots of ultimate, like traveling and competing and that sort of synergy between the idea of the mathematics of the dynamical system and the decision-making about the decision-making, resonated with the, both the real life experience on the field and the sort of mathematical formalisms. I just carry that forward now. I try to say like, am I in a paradigm where I should be a system one in the common sense or in system two. Right. There’s a lot of ways, it’s what I like about it is the synthesis across my experiences as a Frisbee player, as a sort of roboticist, and even like in my extracurricular readings. So, it’s kind of cool to see them dovetail.
Sifchain
Awesome. Awesome. Okay, cool. In terms of timing, we’re at about an hour here, I know that what I want to do is be mindful of everyone’s time. I want to, make a comment on what you said and then make a comment to the community. I have another question for you depending on if you have time and then regardless of what you say, I want to give a closing statement. Does that sound good? All right. The comments on what you said is that, I can definitely relate. I just want to comment in a slightly adjacent paradigm. What came to mind in listening to you, is that, a few things… Anyway, I’d say some of like Antifragile, I was there, although there’s some other stuff that I could talk about for Antifragile. But, the bottom line is with, in this book, there are two things to point out. The teacher says move, which like, as a, there’s a translation sort of issue there that you have to really like dig into, if you want to understand the question. I think the question was like, dogs go to heaven? And then the teacher goes like, moo and moo does not mean, no, it more means like, nothing, like, imagine if the teacher said nothing or like, ooh, right. It’s the way that things are broken down, if you study Buddhism and, or you read this book, is that it’s almost like one answer to the question, or like this is irrelevant or whatever. There are many codes that do different things, and it’s basically self-executing code in your body, that determines your genetics and ultimately like your peptides in your proteins and so forth. So, the stock code on is sort of a meta comment that is stop. It doesn’t mean add a new, it doesn’t mean add a new amino acid here, right? There’s like a chain of amino acids that are either being linked together to form a polypeptide and the stop code on does not mean any particular amino acid. It means don’t add any more amino acids. I do believe that if you are actively trying… can solve problems in your mind rigorously, you will come to scenarios in which the correct answer is don’t do anything or do something else. I think that, like, you can be very systematic about coming to this realization, going with the flow. Does it mean, make no decisions or do no analysis, as you said, I believe that you can, with computational rigor, identify these specific points at which you should stop. I believe that this is embedded in like the foundations of information.
Blockscience
I’m going to agree with you. I wouldn’t have necessarily tethered this to the concept of go with the flow. Like when I said go with the flow, I meant it in this just, I meant it in the sense that I just described it. But what you’re describing is something that I also think is very important, but I actually consider this something like minding the epistemic gap. If you pose a question and you determine that it’s ill posed, then you have to repose it. Or if you’re working within the context of a model and you hit the boundary of the model’s valid domain, then you have to step out of it and reframe the model, or define an interface and define a new model. Like the concept that you’re addressing now is super critical, but I actually just associated with good quality systems engineering or model based systems engineering with the philosophical component of, understanding the boundaries of models themselves, all models are wrong, but some are useful, useful to some end in particular, to what end are they useful, or like really testing the boundaries of whatever assumptions or machinery you’re using. Your stopped code is like, I hit the edge. Like if I’m using Newtonian physics and suddenly I’m approaching the speed of light, I throw it away. And I switched to relativistic physics. Why? Because I can no longer use the machinery at present to handle the scenario that I’m in. For me, that’s closer to what you’re describing, it’s that as I, if I’m in a particular context or scope and I pushed to the edge and that scope essentially breaks or fails to fulfill, or I can’t keep moving, then I have to step up and I have to revisit the assumptions I’m making or the tools that I’m using, or, and again, in this computational systems alongside of social systems, I tend to think of this as minding the epistemic gap. Because anything computational in nature requires some modeling. That’s exactly when we need to stop and step up. That’s a much more rigorous, much more sort of engineering concept. Whereas my comment about going with the flow was a bit more like the other direction.
Sifchain
Makes sense? I have this response on subconscious thinking. I think I’ll save it for another time. The thing I wanted to say to the community is, I see a lot of really good attention actually, since we started more people have added, have joined the community. That’s really cool to have joined like this chat I mean, and we have a lot of questions that are just like, about the tokens we’ve listed or our design or other stuff, or Sifchain. I just want to say I’m really grateful, to have so much activity that we had a really strong uptake in the token. I also want to acknowledge that like, as philosophical or whatever as this conversation is ultimately like his argument, Zargham and I are trying to help build Sifchain — me more so than he is like many other projects. Like this is very much, part and parcel with, building the community and building the product. So, I just want to say basically thank you for adding the funds to the community that will help us, have conversations like this. My last question to you Zargham is kind of in that line, which is like, you have the Balancer stuff, you have a lot of crypto companies that you work with. Like, there’s a great sense in which, so much of what you do is done, like with income and support from the trading community, and people who have participated in these ICO’s, I’m just wondering, like, how do you feel about that? How do you like, how do you square that audience, that group of people who’s affected by your work with, the part of your work that we’re discussing, which may not be as relevant for people who are just, trying to day trade?
Blockscience
Yeah, so the short answer is that I believe that web three based financial infrastructure is infrastructure. I come from a line of engineers… my dad’s a civil engineer, a relatively senior civil servant in the New York department of transportation. My grandfather, a structural engineer and businessperson, my uncle an architect. I kind of approach all of this with an engineering mindset where my goal and job is to not just make roads and bridges, right? But to make safer roads and bridges. A lot of the work that I do, whether it’s work on actually doing some research, collaborating with folks from MIT on the ethnography of infrastructure in particular around, DAOs… Where this idea is that if we’re building infrastructure in cyberspace for this economics and political and financial stuff, then, a lot of it for me is about figuring out what does it mean to make that safer and more expressive. And, some of the ideas of what is safety, and what is good in the systemic sense rather than necessarily good in the, I want to do this right now, let me do the thing I want to do, which can often be great in the moment, but it could be risky both at the individual or at the system level. And really kind of understand the interplay between the technological components in the social components and the individual and the system, and the, bringing this paradigm of other large scale cyber-physical systems, even to know power grids transportation systems, et cetera, and asking how these concepts, both in terms of computer aided design, formal methods, governance, and subjective decision-making about things like parameters or upgrades, et cetera. All of that, to me, ties into this at the level of, if we treat it like an infrastructure, we’ve got to take it for all the way from why are we doing this in the first place? And, who is the constituency it’s being done for? How do we make it effective for them as individuals and safe for them as a group and persistent, including in some cases having elegant modes of failure, like if something is going to break down, can it break down in a way that minimizes harm, et cetera. I guess the thing that squares it for me ultimately, is that people are involved in crypto financial systems because they have a reason to opt out of the existing systems and to put their capital in the emerging ones. It means they may have a higher risk tolerance for sure, but it also means that there’s a lot of room for us to improve the rigor and safety and infrastructure of those systems. While at the same time preserving what makes those systems unique from the legacy ones… So there’s a lot there. Like I feel a deep motivation towards that. My work spans economics and governance and control theory and distributed systems and psych and ethnography with lots of collaborators who have taught me many things, precisely because I want to pull that whole engineering picture together from what we can do that’s new and innovative all the way to, how do we make this something that’s safe? And we can think of it as an infrastructure. You can just show up and use it and not worry about so much like, what’s going to happen to you. That doesn’t mean you’re not taking risks, just that you understand the risks that you’re taking. In a financial infrastructure, we’re not removing risks, we’re simply bounding them, making them legible, making it so that the people who are taking risks know what risks they are taking.
Sifchain
Well put, well put. I think we can leave it there. It does make a lot of sense in this method of going with the flow and having multiple random threads of thought, be connected through things like humor or other interesting consciousness, tweaks, and then directly to what you said. I see things very similarly, like we are here to use engineering systems, engineering principles, to build a way for people to move resources from where they are to where they need to be. And I think that the whole world is asking for that. I feel very honored to be able to have this opportunity. With that, Zargham, I want to thank you so much. Let’s chat more. I mean, this is, there’s been an amazing conversation and it seems like the chat is really into it as well. I’m sure that when we post this online, there’ll be a lot of uptake. So for now, thanks again. This has been really great.
SifDAO Tribe
Main Official Channels
Local Language Channels
