 Our next session is a panel discussion on collaborative research and Carva is the moderator for this one, so I will just give her the floor now. Thanks, George. Welcome to our panel discussion on collaborative research. We're looking forward to a lively discussion about how we can work together to break through bottlenecks and advance the state of consensus research. One of the larger goals of consensus lab and I think from this workshop, we can say of the extended research ecosystem is to empower consensus research breakthroughs. And this panel discussion is centered around the question. How can we build a robust and effective decentralized collaboration network for consensus research. What are our first steps, and I'd like to introduce our panel. First, we have Christian Tasha of the University of Band Christian is a professor of computer science, where at band where he leads the cryptology and data security research group. And he's held visiting positions at MIT and a BFL. He's also worked for IBM research and Zurich, and he has a background in cryptography. He's interested in all aspects of security and distributed systems, and especially cryptographic protocols consistency consensus blockchains and cloud computing security. We have next, Zarko Milosevic of informal systems. Zarko is chief scientist and leads research efforts at informal. He believes that specification and implementation should be in love and strongly encourages techies to garden informal systems is working toward a vision of an open source ecosystem of cooperatively owned and governed distributed organizations running on reliable distributed systems to bring verifiability to distributed systems and organizations. This is Don son of Oasis labs don is the CEO and founder of Oasis labs and a professor in the Department of electrical engineering and computer science at the University of California Berkeley. Don is a researcher and creator of security and privacy technologies, specifically in the areas of software networking distributed systems applied cryptography blockchain smart contracts and machine learning. The protocol labs is focused on building the next generation of privacy technology for the real world. And finally, we have our very own George swage of protocol labs. George is a research program editor here at protocol labs with the background in communications sensor networks distributed robotics and marine technology. In a previous slide George served as a patent examiner with the European patent office, where he focused on human computer interaction and computer construction. George is inordinately fond of ants at PL research he uses his background to create support plans for discovery and invention over PL's longest time horizon. So now that we know who everybody is. I'll ask our panel to need themselves. And I think I will just a frame discussion. I'll say that during this talk yesterday, Marco raised an interesting question. As a systems engineer tasks with building a decentralized computing infrastructure from scratch. How would you start, and I'd like to reframe that as a thought experiment for building a decentralized collaboration network around consensus research. If you were engineering a decentralized computing infrastructure for consensus research, building a collaboration network. What would your first move be. How do you begin to assemble your team. And so I'll open that I'll open the floor to our panel. And maybe seed the question with what does a decentralized collaboration look like for consensus research. This is sort of a trivial answer but I think a good place to start this to organize consensus days I mean that that is the main conclusion of this effort right. I mean, our goal with with our main goal with organizing this event was really to bootstrap a collaboration network right and I think we are. I mean, this is a very promising start and I think we are on track to meet it, but but of course I mean it's it's it's also just it's it's a very small part of the puzzle, but bringing the community together bringing all of the people in this room together is is a great place to start. I actually wanted to say that Zarko here works for a company, it's not even a company I'm not sure it's informal systems yeah and humans tend to organize and sort of companies are like organizations and informal systems is also an organization and maybe Zarko you could you could explain how that works because it seems to be different from other systems, namely from formal systems. Actually, we are sort of on informal and I would say Cosmos side, trying to, to like organize the cross work teams which can work on the same project, and one of them is standard consensus engine, and, and somehow surprisingly, the, you know, start thinking about you know growing and bootstrapping team, the technical concern are not necessarily like the, you know, the first thing you're thinking about that that's that somehow that it's a bit surprising because I guess that would be the first idea. In our current mental model we are trying to think more about like users we are satisfying how to essentially incentivize properly the team, and, and how to essentially be able to grow and scale properly. And then there are also these technical aspects of really how we can be able to have like technical infrastructure in place, so we can go from idea to the future in a much shorter cycle. But like the. So, so this is something which kind of, I would kind of like to, to, to like have some chat about like how you guys are thinking about this, but like from from our experience on Cosmos and tendering the one of the biggest was really like attracting and keeping good people talents around for long enough, because being in a core team was like amazing opportunity to go through and then you know like scale do your own stuff. Unfortunately, the core team were not receiving enough like value in that sense like that we can really keep people for a long time, and having a high, you know retention rate and having like a lot of location in the core team, I think is not really in anyone interest. And so what we are trying to think about is really like, and we had a chat with Mark also on those kind of topics, how we can in general be able to to provide really value to core contributors and give them possibility to do, you know, the work they are best at and don't necessarily need you know to leave or change a lot. And then they're able then doing the work they're best that be able to have like upside exposure and be able to also like benefit from the success of the, let's say upstream projects. But don't you have to steer this in a certain way also. I mean, I mean that's what maybe I understand what you're saying is, you are steering this, developers who came up independently with some things now you're organizing and towards the common goal. Also through financial incentives in a way. Because that's something what I have seen in the, what we have seen on the network level in a way as well before there was, before there were any of the coins, they were peer to peer systems, yeah, okay, you can turn it on they can turn it off, it didn't matter. In some way, unless a specific group came together with a goal, and then you would participate in file swapping for a goal. And I think that's a, that's for me an element also then with the addition of financial incentives it became useful for me to work in this context because I have to relate to my other life as well. So that's, that's an answer for me. And that's exactly what I was saying about incentivization that like that that is a part which was not necessarily present. And so the, I think that that's something we need to change. So like, I think that that we need to be more aware of like, you know, technology we are depending on and people we depend on, and thinking also how we can sort of pay back and not just take the stuff. I don't see my feeling is that there is not enough of that kind of thinking. And I think that it's in best interest of everyone to have a more of, you know, like really taking care about like the people who are delivering value to you. And financial, you know, compensation is definitely and you know we have tokens now so we have like tools available so we can make, make this fun. Yeah, these are great discussions I have a couple points to add to this. I think here we all agree that it's a really important problem and a really fruitful direction to really explore how we can build systems decentralized systems in the decentralized way. I think, so there are a couple things I would like to elaborate on this. So one, I think you're right this is a really fairly, you know, Greenfield, even though there has been a lot of, of course past history and experience on developing open source practice in general. And so there's a lot of exploration and new research and study that's needed to be actually done on this particular topic. So a couple specific things that you know we are doing that I'm happy to talk about is so one. So as many of you may have heard, we have been organizing, we have been doing a DeFi MOOC. Essentially it's a free, massive open online course. And for anyone online can join decentralized finance. And so far for the class we have thousands of students over 30 countries globally. And interesting aspect for the class is, so one is also offered at Berkeley, where students at Berkeley can take for credits. And we have close to like 300 students on Berkeley campus and then we have thousands of students online concurrently taking the same class. So an important component for the class is essentially, besides doing like quizzes and labs is also class projects, which is open ended with essentially research aspects and components. So, so for the Berkeley students, then there is a process how the students can form groups usually it's about like five, three to five students per group, and then they figure out what research topic they want to do and then they write the project proposal and then milestone and then final report and so on. So for the online students, and we actually have been essentially experimenting trying to figure out what's the best thing to do there because the online students are like Berkeley students they know each other the online students they actually first need to get to know each other and figure out what they want to do and also there's a question whether the online students can collaborate with the Berkeley students. So I think this, this is a great opportunity for anyone who is interested to also you know join the process we are actually working out the process as we speak. We help the online students to get to know each other organizing two groups, and we also help propose project ideas for the online students and to actually my hope is that if throughout this process, we can have this online students join and then collaborate together and in the end to results in the research paper. There will be are develop some tools that brings value to the community, they'll be amazing so I think that's a great experiment for this decentralized research. So so that's one aspect is essentially exploring the domain by actually doing it and I hope that at the end of the class will have some valuable lessons that we can share since we are really in the process of doing that as we speak. The other aspect is also at Berkeley where starting a new campus wide initiative focus on decentralized systems called the RDI responsible decentralized intelligence center. And the part of the center we have essentially many focus on many different questions and research topics on different aspects of decentralized systems. And one of the questions is essentially how we actually built these decentralized systems in a decentralized way. We also have the right incentives and also I think one it's important, you know, because we want to enable the system being built in decentralized ways and also I think even just for the decentralized systems itself, it's important that it reduces essentially the code, monoculture, and also reduces the power of code and so on. So I think as I mentioned about incentives I do think that's a very important aspect. And what had the actual, you know, proper design space, how to think about these incentives, how are people are going to be rewarded for their contribution for example one way you can think about is, you can count like the components of the code that people contributed to this, you can count in real world, how often, like that components have been used so that you can measure, just like a miners getting rewards you can say that, instead of miners just just miners getting rewards for running the nodes, maybe every time when the system is being run like you can look at the work component of the system is being used and whoever contribute to that components will also get some rewards for example. And then of course then there's a question maybe some contribution among conceptual and also right like this components maybe someone wrote it first and others changed it later and so on how do you do the incentives that way so of course it's a lot of, there are a lot of open questions. And this is some of the, I think research topics will be really interesting to explore. Yeah, really great example of scaling an open decentralized program, the idea of decentralized matchmaking where students are internally motivated to find each other and collaborate together on topics of interest. And I'd like to point out there's a couple of questions from the chat. You mentioned incentives. And there's a lot of interest in what are the appropriate incentive mechanisms for forming collaborations and then for rewarding participants and collaborations and research groups for their efforts. And I'd really like to get the panel's ideas on what those might look like and how they might differ from the incentives that operate in a traditional university environment. Obviously, that's something that we at PL have also been been thinking about the, and by the way to Don's points, the one of the projects that we supported, or that we're supporting is source cred which is basically it tries to, it's a page rank like approach to mapping the to mapping and incentivizing contributions and this may include code, but also, you know, participation in designs or just discussions in a community, right. And I mean, it's not, it's very far from a solved problem, but I do think it's, it's, it's going in some good directions. That's how I read that incentives are important. I would say that we are sometimes overly focused on incentives, if I can put it that way. I mean, bad incentives are terrible and there are many of those in the current system, without making an appeal to to in its human goodness. I do think that we sometimes overemphasize the importance of incentives in behavior. And obviously I mean there's there's lots of of previous work on the open source world right, including some that that that we are self supported but the fact is there are people working alone on foundational projects for the Internet. We have basically no incentive right and that shouldn't be the case. They should be rewarded by for their contributions but but they do exist right so so people will do this out of their own volition. Is that fair probably not how can we how can we help create a better system that that actually not just incentivizes but supports the work that people will do otherwise. And to I guess the question that that was asked in the chat I mean my view would be that grants do definitely help. We do need better ways to give out grants, they as it stands it's still overly bureaucratic even outside of the traditional system. It's it's slow, and it's also relies a lot on reputation, which is probably not ideal. I think, you know, things like bounties will be in our future. I mean, there's there's definitely space for for doing things in more radical ways and we were recently discussing the. I think it's called Vita though which is about that that is focusing on on research in the life sciences. And that has a model which is not exactly my favorite it's heavily reliance on on patents and patent revenue. It's not fine for for pharma. It's not fine for for what we're doing here, but but there are definitely, you know, doll like mechanisms or smart contracts that that that models that that could be useful in this context. And they're motivated by interesting problems. Yeah, so they're interested by by cute things yeah they want to have something that makes them shine also among the peers like they also are here. And also, and all among those who are not interested in a sign in commercial success. So so that's because they're interested in building the knowledge further. And then a project comes with interesting problems, usually coupled with freedom, academic freedom in some way, because the scientists tend to think that they think that they know for themselves what's most interesting. Then scientists are motivated also. At the end of today scientists also have to eat breakfast and then that's true. I agree with Christian here. I think that that I would just said that for for like the projects it's also very important that we have like a clear vision, and like a clear roles in this, you know, open source decentralized projects. And this this also like in addition to incentives which are, you know, in a form of tokens or or any sort of like this financial reward or compensation is important because like in the team where there are clear responsibilities at a role level, and there's a clear vision of the project, it's also easier to contribute. And people could also be be like willing to take more, let's say risk in being involved, if there is a clear trajectory or path like how you can really make the contribution. And we also like when in informal we have people who sort of like we're attracted and leaving academia to work on a project because I guess they believe that we take would sort of apply their knowledge understanding it kind of real world, and so kind of real problems. So for me like when we talk about incentives. I think that this sort of vision and the clarity in terms of like organizational structure need to be paired with like this sort of financial gains you're having and also as people like working on a shiny project and also, let's say, only in draw, you know, contributing to the public good. I think that there is no harm in like rewarding them, if the project turns out to be like fundamental building block for like you know multi billion dollar story. So the, so in that case, like, I think these things are not in conflict. So, even if you're not like necessarily asking for it, like, I think that we should reward people who contribute or have mechanism to reward people who contributed to like, you know, building blocks, which are very important for the success of the project. And yes, I agree very much with what Christine and as echo mentioned, we all agree that this is an important goal and they really figure out how to make it work I think it's a, it's a hugely complex problem right so one. I think, when we talk about incentives, there are many different forms of incentives and people are motivated by different things. So of course there is the financial incentives and there's also the credit incentive. And also, ideally in the ideal world, we hope that everybody is motivated to work together to make the world a better place to bring greater societal impact and and so on. I think if somehow we can really help achieve that that'll be that'll be amazing. And also really help these different incentives to to work together. And also I think, in some sense, here we are also dealing with some just traditional or some fundamental like organizational problems, and also even some of the human psychology and so on. So everybody knows about the tragedy, the tragedy of the commons. So oftentimes when you have a decentralized team and trying to write to work towards the same goal, but then, since it's not centralized top down structure, then for certain right, like, different tasks that needs to be done everybody thinks that others should do it, then in this case, naturally, then maybe things don't get done. And also just as an example like people talk about climate change everybody knows that it's really important, but when you know each person talking about taking a little sacrifice, changing your own behavior to help the society to achieve the goals then oftentimes that's from what we have seen that's also been difficult to I guess my point is that the problem here is very deep. And also brings out actually more issues in how righteous in general how to people work together not just about building, you know decentralized systems I think the progress that we make here can have big, you know, ramble efficiencies and huge impacts in other parts of our society as well I think it's a great topic to to work on and really explore how we can do better and that's why I said I think it fundamental actually get to how to really see human nature and how we can actually better organize as, you know, a human race together and achieve some of this better bigger goals together as decentralized and as individuals and so on. That's an exciting topic. Thank you both here and in the chat I'm sorry I spoke over Georgian, please. No, no, no, that's not a problem I was just gonna agree with on right I mean my point was never that we should not reward people people should be rewarded by for what they do right. My problem is that in many cases or my point is that in many cases this may actually exacerbate the tragedy of the comments right I mean in the well known case in open sources open SSL right which was extremely foundational and was being funded by no one. And sometimes the incentives are such that it makes more sense to create a shiny new thing than it makes to to to actually work on those foundational services that are broadly useful to the community right and so it is important to find out how to out to route some of the financial outcomes to to to that foundational work rather than than just to to the latest shiny, you know, blockchain or cryptocurrency or whatever. Speaking of shiny attractors here in the chat there's been some attention to the idea of having attractive problems, really interesting goals that that attracts people to work on these problems. And I'm wondering I'd like to ask the panel, is it necessary to have a single goal state or a vision of the future that we're all coalescing toward, or can we accommodate multiple goals and how do you as George said how do you route people toward the appropriate goal, rather than toward some shine your tractor that may not be the highest and best use of their time in a decentralized fashion. It's just like in cosmos we are, we are sort of quite decentralized. So there is no like single organization informally is one of one of them. And so there are, even like within single organization, there are sort of challenges between teams, when you need to work together and so like, when you are opening this further, like, it doesn't get simpler. And so what we are, we're trying to sort of manage this by recently using something called workflow mode, which is, I think initially proposed by by go on. We are a company from verification space. And so the their idea is to have like clear roles and clear responsibilities, and essentially even in this decentralized space I think we need to have like clarity like who is leading the effort. And so it's not like, you know, although we are decentralized, the, we are not all equal in terms of like, you know, responsibilities. And so I think that we need to have some structure in place where people can can naturally organically take some of these roles, and to be able to establish this workflow relationship with like other leads from other projects. And so I think that that we on the topic level. We need this kind of clarity or alignment between like different, let's say groups or orcs if you want to do something together, because we have like, I was having a lot of these experiences recently. We have a call with some other, you know, company or or the, you know, ecosystem representatives, and we're super aligned. We want to work together, we share the values, but then everyone is like super busy with their own stuff. And like, and nothing really happened there. There's no really clear joint call and people who will essentially lead the effort, and then try to, you know, take care about like resourcing it finding people who can work on it and taking care of this. It seems it's not really happening. So that the sort of my experience so I think that that it's very hard to have this organic only so I think that someone need to take care about this at the end, and although being decentralized is kind of cool and you're a contributor I think at the end of the day we need someone who is sort of responsible and orchestrating the efforts. I wanted to take your question into another dimension, namely seem to suggest that you could, there could be some central idea planning or some central things around like if there's a marketplace for ideas first yeah. If you build an idea is or a formalization of a problem is usually half of the solution. Yeah. So if somebody can clearly articulate the problem they wanted to solve, then they have already nailed it down that they have to build something and so in that sense this wouldn't work so well because it would be difficult in the first place, just to articulate this this this research problems like this. In a clear in a very clear way that is definitely that would be measurable if it comes to a solution, and it will also look a bit like in the times where people are running for gold on the countryside if somebody puts out the map and it says where he or she suspected the next gold gold mines to be found yeah. So that is for me a bit difficult to say how to do this planning centrally. I think what we couldn't say what I can say at this time is sharing a vision sharing the common vision. That's the most important thing that we're going to go to make it more clear in technical steps, but we know we have to break it down into technical steps but that's a difficult part, because only did that is work. I think I agree with the Christian said, I think also, when you look at problems and ideas that there are, you know, many different types right so for example in math. I think this has worked really well. Right like the last for my theorem and these examples you can write a really succinct statements that even like six year olds can understand by takes mathematicians, you know, decades or even centuries to try to figure out how to solve. In math there are clear that there are many examples of clearly specifies problems and long standing problems, where then the community, and try to work on over a long period of time, and try to make progress, and so on. So that's in math and I think he may be in physics that has been similar. You know, theories and hypothesis and problems and so on. So like if the reticence proposed some hypothesis and the different experiment experiment, experimentalists, they would then go on to design different experiments to prove or disprove certain theories and so on. So I think when it comes to becoming a science and systems given the complexity in a diverse nature of the problems, like Christine said, sometimes right even clearly identify as specifying the problem itself is already a big contribution. Although there I'm hopeful that with decentralized process, potentially that we could one day have that essentially done in the decentralized process too. I mean even just I think in thinking about our own process oftentimes for people we have like a few co authors and in general, it's a small group of people in advancing ideas of each other and continue to refine and the problem. It takes a period of time for the idea to to to germinate and then to and get crystallized and so on. And in general, I think so far typical hard works is that you have a small group of people that already know each other really well so you can bounce ideas of each other in a very efficient way and so on. And I'm hopeful that with the with the technology actually this could be done at a greater scale so for example like with the covert I think like many of us in the past we've had in person discussions with our collaborators, but now we often just do it online. And like oftentimes with my student collaborators we right we have Google Docs and we write on things and we make comments and then it seems to be actually fairly effective with the technology so I'm actually very hopeful that even this kind of process that we can scale to the bigger scale with more decentralized participation. I'm thinking like this kind of process if we just make for example the Google Doc or whatever open to the public that others can, if they're interested they can contribute they can come and they can revise things and so on that could be an interesting process and that's also some process away plan to experiment to explore in like the Divino class that I mentioned and also in some other like decentralized collaboration setting that I mentioned so so yes I think I'm hopeful that with technology, a lot of these things can get much easier and make it easier for us to collaborate in a decentralized fashion. I think technology obviously helps a lot right but there are still like a human aspect to this thing that did not change that much over time right. I mean I was listening to Christian and I was being reminded of something that is also a frequent discussion for me, which is the traditional technology roadmap for semiconductors. So the semiconductor industry roadmap right and this happened. I mean this was first created in the 90s, I guess, and it was a 15 year outlook into the future of the industry right and the goal was clear the goal was, there are a lot of pieces in this puzzle as a foundry you depend on a lot of manufacturers of different parts right and so all of these people need to be aligned if you're moving to a 10 nanometer process at one point you need all the equipment from the different phases and the different suppliers to be there at that same time right and so the industry came together and and in every few years published an updated 15 year roadmap for where things would be and this was a consensus exercise by the industry right now obviously there is a big difference here the semiconductor industry is extremely compact, it's you know a few large players at the actual semiconductor stage and a few upstream suppliers. So it's much easier to do this exercise but then again they were doing it without the aid of all the technology that we have today. So I do wonder whether this is something that we would never see happen as a broad consensus in the in the blockchain or the distributed systems space more generally. One who asked there how do we facilitate building strong relationships over the internet. I think like one like you were saying, by extending a network like a graph yeah by one one one to one relations and then growing it like this not by a start apology because the center of the star would be overloaded. And it can also not be by complete graph where everybody talks to everybody because that's not scalable. That doesn't work. So it has to be going with a hierarchy but also buy it by personal trust in some way. That's a prerequisite for in, I would say for for trustful and effective collaboration. It's related also to this university activities and I fully also agree there. So we are, I would say in general, quite under resourced so we are like definitely lacking a lot of, you know, people able to lead various efforts. And so I think that that this is this is something where there it shouldn't be so hard to collaborate across like various you know projects and ecosystem how we can, we can be able to attract more people and how we can able to sort of be able to onboard them and so they can then be part of some projects and be able to contribute. So it's really like it's it might be one of the really like the biggest problem you're facing and then also that the blockchain sometimes has like the negative connotation, which like for me coming from like this through the systems. It's it's kind of a bit hard to understand because it was, but like a lot of people are they really don't see a value in blockchain and I think that the whole maybe cryptocurrency and you know that that side the speculative side of thing is just like discouraging people to be part of this. Although we are working really on like core infrastructure projects in most of those, I guess, please people here and, but it's just a big challenge which I was sort of like, not taking so seriously but it seems to be also something we need to take care about. And being able to be educated people and being able to you know spread more word about like what exactly problems you're solving, and be able like to contribute I think we need like more stronger like kind of, you know, some kind of internship so or like a collaboration project also between maybe teams and also with the universities also so that we can we can grow new leaders basically. We've discussed potential decentralized governance structures incentive mechanisms for building collaborations, how do we build trust to decentralize fashion how to organize knowledge work. So I think we have a lot of interesting things to think about as we work on building the consensus research collaborations of the future. And I'd like to thank all the participants of the panel for a very interesting discussion, very stimulating discussion and encourage everybody to check the slack and to continue to participate around the discussion maybe form some decentralized collaborations. Secretly by slack, and thanks again for attending this symposium. Thank you, Carla, and thanks to all of our panelists.