Stigmergic Economy And Large-Scale, Decentralized Networks - Matan Field (720p_30fps_H264-192kbit_AAC)

And I remember him fondly for the dinner we had yesterday and I remember he told me he is a a theoretical physicists and he is now working with blockchain the technology behind bitcoin and he told me that string theory is smaller than blockchain so we will have to hear more about this one Ok, thank you Cosimo So, first of all let me thank Network society and specially David Orban for inviting me here It's really exciting and inspiring to see all those networks that are now rising and growing, and extending and spreading and all speaking the same language similarl agendas, Ideas that actually making me optimistic about being able to revolutionize the way we will live here on the planet Great stuff I changed a little bit my title to adapt this specific congress So, it's about Stigmergic Network society So really how I see the paradigm of Network society works and I will talk about my vision of that Network Society a quick observation from nature and a little bit of history of mankind some of the new means to deploy this kind of paradigm and some examples so i will start with the vision for that we will need to engage with a little bit of immagination So imagine a world completely governed by networks self organizing networks imagine facebook network but without facebook inc imagine people renting appartements with no Airbnb in the middle imagin "right" applications with no Uber or Lyft millions of people ensuring each other spontaneously and maybe millions of people building startups together so this is the world that I would like to see So for that, let's take one step backward and look in Nature for a phenomenon called Stigmergy isn't it beautiful So Stigmergy is indirect spontaneous coordination of agents or in Wikipedia is a mechanism of indirect coordination between agents or actions the principle is that a trace left in the environment by an action, stimulates the performance of a next action by the same or a different agent in that way subsequent actions tends to reinforce and build on each other leading spontaneaous emergence of coherent apparentrly systematic activity Stygmergy is a form of self organization it produces complex, seemingly intelligent structures without need of any planning control or even direct communication between the agents such it supports efficient collaboration between extremely simple agents, who lack any memory, intelligence or even individual awereness So, how would we implement that logic to billions of people so what is Stigmergy? basically different agents in a network can emit by indirect i mean it is signal based there is no direct interaction between agents there is a signal mediating actions of agents so agents emit signals and then agents decisions are also biased by the same signals and the Bias is designed to induce synchronicity of actions to induce cooperation now, by design I mean that nature have designed this for millions of years to induce cooperation but now when we are going to talk about people the question is: How do we want to design those rules in order to induce cooperation and no agent has control of anything but notice one thing that this Stigmergy as it appears in Nature is unstable to self aware, exploiting agents so in that sense it is semi resilient still an agent if it was self aware could basically hack the system for it's own profit so Stigmargy in Nature works for very simple agents who lack self awereness So then the question: How do we improve Stigmargy paradigm for people? so this is one example in Nature of Ants this is of course just an algorithm or simulation in run on a computer and in fact a very known problem of finding the optimal minimal path between two point on any topology is known to be solved algorithmically in best way by just mimicking the algorithm of Ants which is very simple this is just an example of finding the shortest path this is another frame from a real simulation where you can see that you starting basically with a random walk of Ants and with some extra rules this is the stigmargy and you see how eventually the Ants converge on the shortest path and this is just for the expression in fact, Stigmargy is the only large scale coordination known in Nature it fits very well, simple algorithmic agents they are basically fully automated in a way already in their DNA and because of that, there is no need for incentive or trust because a simple agent do not try to hack the system So in that sense cooperation is the winning strategy but actually is almost the winning strategy you can actually have better so to say selfish strategy for more complex agents as I said is unstable to self aware agents so how do we apply those ideas to people So let's discuss a little bit the evolution of human large scale coordination so it started with centralized architecture then it evolved to federated architecture then to distributed architecture but with local consensus and finally i woul like to tell you more about distributed architecture with global consensus and you can use these notions for for information distribution, for value distribution or for governance Ok, so Rigid coordination you know very well it's a linear process of information it can take very rapid decisions but it can only efficiently work very static conditions it's not adaptable to rapid changes in the environment it's highly vulnerable to rapid changes and also highly vulnerable to critical mistakes like you have one agent, taking the decision there is high odds it will take wrong decision while as if you have millions of nodes taking decisions some of them will take the right decisions So federated coordination for example democracy and the internet is Semi parallel but is very sensitive to corruption because a lot of power is aggregated in some nodes and still it's not adapting fast enough in fact todays democracies cannot anymore adapt to the technological changes that take place so rapidly there is an even better architecture is the Market or p2p coordination so it's a subjective distribution wich means that it only optimizes according to p2p interaction so it optimizes localy the best example is just pure capitalism so it does optimize locally but there is no global optimization there is no global coordination so that drives competition, it leads to twists and eventually to severe inequalities and as you all know it's non sustainable I would like to build a world with Stigmeric coordination where there is a distribution of power a distribution of decision making, a distribution of value which is inter subjective and we call that DCO Decentralized Collaborative organizations This is the real network society that i have in mind and it has the feature of Decentralization, Scalability Resilience, Macro diversity, Micro homogeneyti and most importantly and that's the big difference from Stigmergic Nature it has an incentive for cooperation and that is the additional piece you need to add for people, you need to have a built in incentive to cooperate and not to corrupt the system So with this additional incentive model Cooperation truly becomes the winning strategy and that's I think the world we all want to live at So what will be the means for that About a decade ago people invented Distributed Hash Table or DHT the canonical example is BitTorrent and it's just a distributed architecture for file sharing It is distributed but it cannot carry value so there is no global consensus buil in this protocol so you can distributedly share informations you cannot make transactions on this protocol So you cannot build an economy with this architecture but that is the first I would say block that stuck and some example are BitTorrent IPFS is a new example and today there is even more advanced one called Synereo which also incorporate social network features into it and then about six years ago some people unknown people have invented the blockchain which is a distributed consensus protocol today the blockchain is much more than that it's actually a distributed computer but the important point is that you have a consensus protocol which means that the whole network can hold objective consensus about a state So imagine that: we all have a computer that we all run together each of us can give input to the software and none of us can corrupt the software and this will have dramatic effect on the ability to create global coordination, global economy and Stigmergy However the blockchains only based on algorithmically verifiable input so the only way to interact with the decentralized network is via input which is algorithmically verifiable such us the mechanisms that are called proof of work proof of stake, proof of storage, proof of CPU and so on.. proof of movement as well the problem is that most of decentralized interaction you can think of requires also human interaction for example let's say that we are now a thousand of people in the room and you want to write a piece of code together this already takes place in the opensource community right? Linux for example the problem is how do you distribute the value of that cocreation? Wikipedia for example Today for Wikipedia there are two Well it's amazing but still there are two issues Firstly, it's based only on voluntary actions like people cannot actually work all of their time and make revenues of that or salaries and Secondly it is not really decentralized It's based on administrated architecture What will happen if you have completely decentralized Wikipedia? that people can actually partecipate and get value from? How much more scalable would that be? or how he would look like if instead of that you have the same network that build some opensource technology and so on but the point is that people are interacted with a network by giving value in terms of code, design and what not and the question is how do you distribute the value of that cocreation among collaborators so you have to rely on input which is not algororithmically verifiable but it can only be assisted by other humans and that's when it enters the BACKFEED protocol so is a distributed consensus protocol based on human input which basically means that it incorporates pear to pear evaluation mechanics decentralized reputation system decentralized value distribution system and last incentive model for early cooperation I'm saying early just to emphasize that built in the model the early you are partecipating in the network the more value will get reflecting the higher risk you are taking Because, the largest problem of the decentralized networks and actually general networks as well is creating the critical mass of participance before the network actually caring real world value So in order to incentivize people for early partecipation before there is operation of value basically people who partecipate the earlier get more shares in a way in that creation and tha indeed makes collaboration the winning strategy so in a sense the backfeed protocol is the DNA for decentralized collaborative organisations or if you want is a DNA for Distributed Network society so that's how the blockchain stack looks like there's the DHT, the blockchain, the backfeed protocol on top of that it will be a collaborative platform and eventually a visual browser immagine a browser Today we are browsing the internet or Wikipedia or anything, via links and clicks but immagine that beyond this database it exist a graph So the internet is a graph Wikipedia is a graph every database that we know today is a graph but why do we browse it through links? cannot we just browse it through the graph itself So you imagine a visual browser where you can actually travel the network from the graph, seeing the graph zooming in, zooming out and so on So I'm claiming that this would actually be a necessary tool for partecipation such gigantic network "this is under development yeah" Where? Where? there are various groups around the world that are doing something similar to that and we are basically taking that knowledge and code and gradually incorporate that in a decentralized world but for example some projects of this kind one of them is called "risi" and another one is called kumu another one is called "napper" So all those projects came in the same year I think many people in the same time realized the necessity of that So this will be like the blockchain stack this can be like ethereum BitCoin and backfeed makes the protocol there is application there and then the platform or the visual browser layer and it is very analogous to the internet stack that we all know in a sense we are building a new kind of internet Imagine an Internet which is built in collaborative UNCLEAR"So built in that in the Ethernet no zone in the network will cooperate" and build things together So very quickly I won't get into much details but very quickly i just say that the protocol works in a way that contributors can make any sort of contribution then any other node in the network can evaluate and assess the quality of that contribution each voice is basically weighted by the reputation of each node so each network has a different reputation score and a different token score the token score is like the finantial equity of that network and the reputation score is kind of like the influential equity of that network the important point here is that it is built in a way such that those people who are are shown to be alligned with the network their reputation increases where those people who are misaligned with the network their reputation decreases this means that when you are voting on some others contribution you are actually putting you reputation on a stake which basically makes a feedback look that drives the system into alignment however those who are misaligned with the network at any time can just fork it like you fork a code on GitHub they can fork the network, fork the reputation score fork the token score and make a new network in that way in a Natural selection evolution you make network of networks which basically, everything is opensource this is a model for opensource technology so basically they all share the things that they agree about and in a sense compete only about the things they disagree about and basically probe all the space of possibilities in a sistematic way so it's again diversity at the macroscopic level and homogeneity in the microscopic in network level Let me just skip that and if I can go back later into more details if you are interested the way token are being distributed and reputation as well so what are the tokens good for? basically the decentralized collaboration is making something could be decentralizing insurance decentralizing startup, decentralizing energy grid or what not and then there are those contributors above that have gained tokens in the early days and they can use them in order to facilitate the new service or product of that network but also there are many other users that never gained tokens and they need to gain those tokens in order to use the network and they can do that in two ways either, just purchase the tokens from the contributors directly in the open market place decentralized market place or they can just purchase them directly from the network which is kind of a decentralized reserve fund and then the cash that is going into the reserve fund I mean the tokens of the contributors can be redeemed against that fund so it's basically autonomous economy which is designed to have a stabilized token and so on and that gives the tokens that have been distribuited in the early days it gives them real world value once the project succeed So it goes like that so initially it has no value and then people in the succes of that network and there is demand for the tokens and then the value is increasing at some point here the community starts providing some goods or services to the public and in order to facilitate or enjoy those goods or services you need to have those tokens and then the tokens value is increasing up until it stabilizes on some value that's how this economy is designed let me almost end with some examples that are actually under development these days So for example you can imagine decentralized ride sharing network So it's a gigantic network of pears driving over the place where a rider and a driver can be matched in complete pear to pear fashion so you don't nead uber or lyft to match those they can actually the software that is running tha matching algorithm is actually sitting on a blockchain so you don't need any centralized server to match the driver and the rider and then the transaction is also directly p2p between the two nodes imagine decentralized journalism we are actually now working with two companies to implement that like kind of reddit but completely decentralized anyone can upload content anyone can edit, suggest editions of others content Right, but whent it kicks in the reputation score so basically it means that many networks you can have one huge database of content but then many networks that looks at this database from different angles with different reputation score So maybe your content will be higly valuable in one network and less valuable in other networks so each of these networks is self monitoring the quality of the content Right Right, so basically so basically if in some network there is a content A that will not be approved or endorced in that network that content will just disappear from the view of that network but if another network actually endorce that content it will be viewable in that network so basically it's a self monitoring mechanism In fact, you know one thing people ask How do you monitor spam? in such networks because you know in a centralized you know we already have evaluation ranking systems, i mean Google pagerank is a system, Quora and we have many others Amazon So why just not using that on a decentralized architecture? that's one question, the second question is How do you monitor spamming? like bad behaviour on decentralized architecture because it is so hard on a centralized architecture the answer is that, and it's for the two questions if you just take centralized evaluation systems and put them on decentralized topology they are miserably failing and the reason is because you can easily hack the systems and spam it so you need to have complete no notion of violation system, a complete no notion of reputation and so on and that's also the answer for the second question in fact the answer is that is actually easier to control spamming on a decentralized network because you don't have one monitoring authority each network has n nodes it has n monitoring authorities so yeah it's much easier to control spamming Yeah I will finish in one or two slides So you can imagine decentralizing insurance network like bilions of people insuring completely p2p each other with this technology that i'm telling you about there is no reason for insurance companies I mean once these cases are deployed Insurance companies will not be able to suggest anything better than that it will be more fair, it will be more aligned nodes on these networks will be more aligned it will be cheaper and it will scale centralized companies will simply won't be able to compete with that you can immagine decentralized makers decentralized startups What would happen if millions of people actually build a startup together? How this will look like? decentralize VC but in one room a decentralized environmental movement actually related to the previous talk we are in touchg with several organizations and I would like to get in touch with the last speaker as well about making a large scale decentralized environmental movement and of course decentralized social network So just before finishing just to tell you that there are many beautiful people that are responsable for this dream to become true and this is (unclear) not updated (unclear) all the time but these guys here are most of them are full time dedicated to this project and let me just end with one last sentence So basically, i just want to tell you Welcome aboard to the transition that is going to take place for humanity Be prepared and most importantly don't worry because nothing is under control Thank you Thank you Matan we still have like two minutes for questions like really short questions really short answers and then we will have to go forward First thank you very much for the great presentation I have a question reguarding the blockchain How do you look at the blockchain growth versus computing storage, computing power versus availability for decentralization So this indeed an excellent question I mean scalability is the largest problem or the biggest problem for blockchain is called scalability problem and i think that is the only problem left in order to complete that technology So this is inprogress I can tell you what the solutions look like it's not yet completed but what are the ways people are taking it's a little bit technical though but the idea is that you don't need you don't need to register everything on the blockchains you can register a lot of things on other typologies like DHT and just in a wat zip them and just register the zip on the blockchain that's one direction and the other direction is that the point is that Why blockchain is so expensive? in terms of computing power it goes together with the consensus protocol the consensus protocol works because all the nodes in a network are repeating the same actions that's what brings the consensus but on the other hand it's very expensive because all nodes of the network are doing the same actions, the same computation So the way to kind of resolve that tension is by having not all nodes of the network observing each and every transaction but then doing that in a complicated complexe manners so that hackers cannot kind of track the changes of typologies that look and observe the network and basically attack it So actually we have dealt with the same problems in the backfeed protocol and resolved them in the same way So this is kind of the direction but now you need to turn those words in algorithms and functions and machine much more complicated but very smart people are working on this in these very days Unfortunately we do not have time for anymore questions I'm sure Matan will be available for answering those and chatting about it later on Thank you Matan for the speach and let's hear it