 Yeah, hello. So I want to talk about the colony reputation system. Just going to wait for the slides to appear. Here we go. So you may know Colony as a platform for organizing teams, collaborating online, doing work. And that's very much what we're about. It's to use the blockchain to help us work together on projects that we care about. We like to think of ourselves as the people layer of the decentralized application stack, meaning that we're very much a blockchain company. But our view of the world is not something like this, but the blockchain front and center and people sort of as an afterthought of the periphery. It's about using the blockchain as a foundation to help us coordinate and really do projects that we care about, that teams care about, helping them work together. To that end, we want to create a platform for self-governance without a strict hierarchy that's flexible in who controls and who influences what goes on. And for this to work, we have a reputation system that allows us to wait our decisions. It's not about voting per se, but it's about people who put in a lot of work and a lot of effort, earn reputation, and then they have more say in the project at hand. So what is reputation? And reputation is fundamentally something subjective. It's sometimes hard to pin down. And we often talk about reputation in terms of reputation scores. And so these here are not reputations. These are scores that might convey reputation that tell you a little bit of information. So you might like a movie with a high tomato rating. It tells you a very specific thing. And what that means to you is still subjective because you may trust IMDB ratings more than the tomato ratings. And the information in each one of these ratings is very specific. Reputation really doesn't translate. So if you have a high credit rating, which is a type of reputation score, what does that tell you about how good you are at tuning pianos? Very little, right? Reputation requires context, so it has to be more than just a single number. We cannot have. No, it just doesn't translate. So the reputation system that we're working on today is what I want to talk about. It's not just a number. It has to be more complicated. It keeps track of a lot of things. Your activity in the colony is what determines your reputation. So it keeps track of the work you do, but also in which domains or domains are like departments or working groups within a colony. It keeps track of what skills you used, what type of work it was, how much you earned as a proxy for how big the work was. Was it a huge task or a little tiny fix? How well you did your work, what the other people thought about your work, and more and more and more. So it's very complicated. It has to try to keep track of all of these different aspects. So as an example, suppose in our colony there's this task called give a great talk. And in our interface, it might look somewhat like this. So you can see up there, I've been selected to do this task. There's a bounty of 100 tokens for it. And it's been tagged with the public speaking skill, the communications domain, which is our communications department, the DEF CON 3 sub domain, which is our working group for preparing for DEF CON 3. And if I do this task and I do it well, then I'm going to earn 100 tokens in our colony. But I also stand to earn reputation. And I will earn reputation in the skill, public speaking, in the domain communications, in the DEF CON 3 working group, and so on and so on. And how much I earn still depends on how well I do. The 100 tokens give us a proxy for how big the task is. But if you give me a standing ovation at the end of this talk, I can upload a video of that. And then I earn extra reputation points for having done really well. So I'll get back to you on that later. The reputation that I earn helps us in the collective decision making. And what we want, so this system is a civil resistant decision making mechanism. You have to actually do work to earn reputation. It's also meritocratic, not plutocratic. I mean, what we mean is you have to do good work that your peers like to earn reputation. It's not something you can buy. So it's not the person with the most funds who has the most say within a colony on this level. And although I'm not going to go into it too much, I want to emphasize this is not about repeated rounds of voting. It's more an implicit decision making mechanism. And votes are just of a last resort. So it's not bureaucratic. And we use this for almost all day-to-day decision making in a colony. That asterisk, by the way, refers to our last year's talk at DEF CON 2 about token weighted decision making, which still has a place for questions concerning tokens. So token holders must have a say when you're deciding to dilute tokens. But for the day-to-day activity, it's your reputation that counts. So you can get reputation, as I said, by working on tasks, but also doing administrative work, administrating tasks, evaluating other people's work. All of these things could earn you reputation. And also policing the colony. That is, if you see people abusing the system by creating wasteful tasks, if you go around flagging those, that might be valuable. You can earn reputation for doing that. If you abuse that system and somebody else flags your flags and essentially initiates the dispute process, they could also earn reputation. So we're trying to balance all these different actors. And if you participate in the reputation mining procedure, and that's something I will talk about later in this talk. So reputation is an amalgam of all these inputs. And it has to be this complex, or let's call it nuanced, to be useful. So to be truly useful, we wanted to keep track of all these different aspects. So the question is, is this too complicated? And what I mean by too complicated is that on the one hand, we've got meaningful reputation scores, and they're competing with the gas cost. The more complicated the calculation is, the more it's going to cost to interact with it on the Ethereum blockchain. And that might make it less useful. But even more so, we're competing with the gas limit. There's a maximum cap for how complex our reputation update calculation can be. And I'm going to let you know in a secret what we're trying to do far exceeds the gas limit. This is not something we can do on-chain. So does it really have to be on-chain? Does reputation have to be on-chain? The easy answer is, yes, of course it does. If you wanted to have any effect on decision making in the decentralized context, on payouts, anything of that sort, it has to be on-chain because it's consensus-relevant. But our answer is, it doesn't really have to be on-chain. It's not necessarily to do the calculation on-chain. As long as you can prove to the smart contract what your reputation really is, you don't have to do the rest on-chain. And the system, I'm going to describe scales nicely, because no matter how many colonies, domain skills, or any other tweaks we put into the update formula, we're only going to need 32 bytes to be stored on-chain. So the way that works is the following. We have all of these reputations here. So maybe r0 down there might be my reputation in our colony for the public speaking skill, and r3 might be, I mean, these are all the different reputations. If they were on the blockchain, the contract could read them. What we do instead is we hash them. Now, if those hashes are stored on the blockchain, then the contract, of course, would not know what my reputation is. But if I supply my real reputation, r0, in a transaction, the contract could hash it, check it against the stored hash, and then it would also know that what I'm saying is true. And the next step should come as no surprise in this conference where Merkel trees are mentioned five times a day. We hash them pairwise and pairwise and pairwise until we get to a single hash, and that's the only thing we need to store on-chain. And referencing that, anyone can prove what the real reputation is. So the only thing we left to do is figure out a way to calculate the reputation root hash and to submit it to the blockchain. So how does it make it to the blockchain? And that's this reputation mining procedure. So anyone can calculate the correct root hash. Everything that goes into changing reputation happens on-chain, completing a task, administrating a task, taking part in the colony. So calculating it is not that hard, but we need to figure out who can submit it to the chain and make sure that the submission is correct. So the first ingredient is a little bit of game theory. You can earn funds by submitting the hash and then you can get punished for being wrong. And for that to work, we need to be able to prove something wrong. So the key insight is you don't have to be able to do the calculation on the blockchain and figure out what the correct hash is as long as you can prove that a wrong hash is wrong and penalize the submitter of the wrong hash. So suppose you want to take part in reputation mining. You just simply set up your mining farm. Actually, no, you just take your laptop. All you need, it's a small calculation in the off-chain world. It's only hard in the context of Ethereum. You need to stake some tokens. Our colony tokens are needed to become a miner. Then you run the colony mining client. And the colony mining client will monitor the Ethereum blockchain and collect all of the little events that happen that influence your reputation. Apply that to the reputation tree, figure out the correct new root hash and then submit it to the blockchain. Having said that, all miners of course want to be the one to submit it to the blockchain and we don't want to spam the network with too many submissions. So there's this competition going on where at the start of a submission window, very few miners qualify. And as the time goes on, more and more miners qualify until all the submission slots are full and then a new cycle will start. Yeah, and then miners earn tokens. Miners earning tokens is the paradigm in this space. So that's no surprise that happens here as well. And so that's the overview of this mining procedure. And I want to say just a little bit about this CLNY token so that you don't get the wrong idea by thinking this is all it does. Any colony has a token to do its internal accounting with. And the CLNY token is a token of a colony the so-called common colony, which it's a special colony. It's the one that's responsible for managing the colony network. Once we're launched in a fully decentralized way, we don't want to be like us as a colony team want to be responsible for maintaining the network. And definitely we also want to keep that decentralized. And there is a special colony whose role it is to maintain the contracts and so forth. And it's this colony's tokens that are used in the mining procedure. All other colonies pay a small network fee to the common colony for the services provided, the maintenance of the software and also the mining. And if this network is profitable, if the common colony is profitable, people who hold tokens and reputation could expect a payout. So there's an economics involved in this entire network. That's more detailed in our white paper. Okay, so let me talk about what happens if someone tries to cheat. For our reputation mining to work, all we require is a single honest miner, which is good. But if there are several submissions, then all but one of them we assume are wrong. And how do we determine that? So if there's two submissions, both parties have to submit a justification for what they've submitted, which takes the form of another hash, the justification root hash. And these indices down here are all root hashes of a reputation tree. So RH0 down there, that's the root hash we agreed on at the end of the last cycle. So we all agreed on that. And this one down the other end is the one that the miner submitted. The one they claim is the new root hash. And each step from RHI to RHI plus one represents a single update of a single person's reputation applied during the update period. So if you agree on the very left and disagree on the very right, at some point there's a first step where we disagree. And what the contract does, it challenges parties to prove the inclusion of a single specific reputation update step, which is small enough to calculate on chain. And by looking at the Merkle proof, it can determine where the discrepancy is. And after a few iterations, either one party will have to stop responding because we found out that they cheated or we actually determined a mistake in the calculation. So the wrong hash gets discarded and whatever is left in the end is correct. And this is fully automated. So if you run our mining client, you will never have to deal with this. This is all completely automatic. And it also doesn't require a challenge response protocol of different actors. The challenges come from the smart contract and this is fully automated. And that's because the reputation update, while not being able to be done on chain is completely deterministic based on inputs from on chain events. All right, we released our white paper recently in September, which just goes into this whole process in far more detail and also describes the economics of the network. So why did we spend so much time on this? We spend so much time on it because reputation really allows us to fine tune the incentives. We've heard a lot about crypto economic incentive systems and they often involve paying people to do things or penalizing them financially to try to get them to behave in a certain way. But the reputation system also allows us to reward and penalize people in a not immediately financial way and we can really fine tune what we mean by good behavior and bad behavior and tweak that as we see fit. And that's especially important because in this context of trying to avoid voting on everything, we want colonies to be very fluid where for example, if a task is created and I say, I want to do that task, if nobody objects, it will happen. There's no need to ask my team members. So if you think of the teams you're working in, most of the time you are in agreement amongst yourselves and if every little action required a vote, it would be in this headache. So it's designed not to use that. And if somebody goes in and forces a vote all the time, they might actually lose reputation based on abusing the system. If you do it in just those moments where it really counts, the system could pick up on that and you could be rewarded. So because the calculation is off chain, we can make it as complicated as we need to to really fine tune all the incentives of the actors in the colony. And it's much more than we could ever do on chain. And yeah, one of the aspects that we really can't do on chain is the concept of reputation decay. So reputation in a colony is not forever. You earn reputation, that means you then have influence in the colony, but over time that influence decays away. And that's a protection against people early on in the colony amassing massive reputation. They have a lot of influence, but it should not last forever. So when you join a colony later, you can assume that the people with the most influence now are the ones who did the most work in the recent past. And that really evens out the imbalances over time. So we think this is a key feature. So back to the task that I'm working on right now. When I upload the video of your standing ovation, I'll get points for that. For the talk last year at DEF CON 2 with Elena, I will have, I still have some points because that was relatively recent. Jack right here gave a talk at DEF CON 1. His points have completely decayed away. If you want something to really come up here and tell us some more about colony soon. Thank you. So in summary, the points are earned for doing good work and good is something that we can tweak and fine tune to get the incentives right. And the reputation weighs your decisions all throughout the colony, but we try to keep them as local as possible. Again, in the interest of making things as unbureaucratic as possible. Keep them within the small team, within a department and don't always go to the full colony for every little decision. And maybe importantly, I haven't mentioned yet, reputation doesn't translate from one colony to another. And that's because there could be many different colonies and we have no idea what your, there's no way to translate from one to the other. What it means to do good work in one colony might have nothing to do with another. And it relates to the various tokens. So if one colony has great work and this token is worth a lot, the numerical values of the reputation might be quite low elsewhere, they could be completely inflated and meaningless. So it's within a colony, but the reputation mining process updates everyone's reputation in all colonies, in all departments, in all skills. So it's one global system within the colony network. This is our team. Our team has grown quite considerably over the last year and this photo was taken at our last offsite and in fact, it's already out of date. We've added more people since then. It's a great group of people. Really glad part of this team. So I want to leave you with our advertisement for our blog. The blog.colonial.io has a lot of good content that goes beyond just colony. We write about regulatory aspects, token sales, of sharing, of equity agreements in this decentralized space. There's a lot of good content, please do check it out. Find us on Twitter at Join Colony and I'm now ready for that applause, innovation, to prove that I've completed my task. Thank you very much.