The limits of decentralization in knowledge sharing / Abeer Sharma

Transcript

right thanks everyone for joining us here today on the third day I know everyone's pretty tired and no one wants to have anything to do with blockchain stuff anymore so I appreciate you're here to give me all this support and encouragement I'm no one important no need to introduce myself um let's have a show of hands here how many people here have used Wikipedia in their academic work even though they were not supposed to yeah we all have and if you if you say no you're a liar and there's a reason for that right Wikipedia is basically now considered a largely reliable source for secondary research the reason for this is its strong Community Spirit lack of Entry

barriers and emphasis on the collective ownership of content what makes it even more impressive is that it's primarily volunteer run and it's um most of its Revenue comes from donations unlike other top 10 websites however Maybe Wikipedia is not that idyllic as people think uh depending on who you ask there have been allegations of political bias in Wikipedia some saying it's left being a website others saying it's a right-wing website um but there's many ways that this bias might be implemented due to the functional structures of Wikipedia the first is Source deprecation where the

Wikipedia Community prohibits certain websites from being cited as valid references secondly violation of the neutral point of view policy uh based on the article concerned where people will negatively or positively emphasize certain aspects of a topic depending on their political or other ideological affiliations and thirdly is the whole issue of locking Pages which you know happens quite a lot every time the U.S elections are near in the English language Wikipedia what's up another thing while people say that you know Wikipedia is a democratic platform its actual structure is a bit more complex than that I'm not going to go into detailing the entire hierarchy but

you know there's levels of different users with different powers and different Points of Authority which can basically you know control the fate of the website so it's not that open in the first place and of course related to the previous part there's a rule of one percent in Wikipedia where one percent of the editors contribute about 80 of the content and have quite a large say over how the how the articles are structured and the topics they talk about stuff like that because of all these factors there is now a breed of people thinking that maybe blockchain will solve all of our issues and the most famous example here is everypedia who for the past few years

were advertising themselves as the Wikipedia killer the essential idea behind every pedia was that we are going to leverage staking and slashing mechanisms using our native IQ tokens to ensure that you know people coordinate in Fair accessible and open encyclopedia article production so largely what it involved was you stake IQ tokens for 21 days you get some points known as brain power and you can propose edits using those brain power points once you propose an edit you know other people who have stake tokens they vote on whether they want to pass your edit or not and if your edit is passed you get curation rewards the people who

passed your edits also get curation rewards these creation rewards come from token inflation proposed platform fees what is interesting with every pedia was their uh slashing mechanism while generally in proof of stake mechanisms slashing involves a redistribution of tokens from the minority to the majority every pedia instead decide to implement time-based slashing where basically your lockup duration would increase based on how much in the minority if you were if 100 people disagreed with you then your um your stake would be locked up for three months if only 10 people disagree with you okay maybe we'll lock it up for one week something like that

now every pedia looked like a good idea did it work unfortunately it did not work and there was a few issues wrong with every pedia first of all the majority of their submissions they were copy paste job of Wikipedia eventually everypedia ended up deploying a bot which would just scraped Wikipedia and copy paste any new edit that happened on Wikipedia on the everypedia platform there was a lot of misinformation going on you know any original articles which were not on Wikipedia were largely coin spam self-promotional material lots of non-notable Articles conspiracy theories defamatory content and you know eventually like people just stopped

contributing to every pedia they held on to the IQ token but they decide we're just going to use this as some other speculative asset and not contribute now everypedia ever since the past few months just starting last year they have decided to Rebrand as IQ Wiki uh while there's still some plans of staking and slashing mechanisms they seem to be focusing more on the AI hype cycle and uh honestly I wish them all the best let's see if they actually end up improving on the everypedia limitations but here's what went wrong every pedia didn't realize that Wikipedia's key strength was its Community spirit and that is why the donation only model is successful every

Wikipedia does not need to earn Revenue in a traditional way every pedia chose the route of financialization of community interactions and basically or individual self-interest was celebrated rather than the idea of contribution and that's why the IQ token never passed it made it past empty speculation now there have been a bunch of studies in this regards you know studies indicate that Community identity and a desire to volunteer are linked in a virtual Circle you have a stronger Community you're going to want to contribute more in a volunteer spirit and so on it goes on in my virtuous cycle and well I'm no historical social Anthropologist or something there are

several proposals on the function of money in society and studies indicate that money makes cooperation possible when people cannot rely on reputation and kinship and that one of its primary roles is to facilitate temporary cooperation between stage strangers who will never meet again obviously that entire function of money sort of you know grinds against what you typically understand as a community spirit and this also ties down into ownership of content production Within These respective platforms what Wikipedia did was that it depersonalized the editing process by emphasizing Community

ownership of editorial output this together with low transaction costs to participation fostered collaboration now if I go through Wikipedia and I think something is slightly wrong I don't have to be an expert right I can go ahead and edit that no one's going to come back to bite me in the ass and tell me I did something wrong they might reverse my uh edit but that's fine I won't actually be hounded or persecuted for it in every pedia they explicitly Associated edits with individual contributors because that was their selling point right you make a good edit you will be rewarded we will consider you a hero you will be the cool guy in the room but what this did was it resulted in the monetization of

content production which also introduces an element of risk and those and that element of risk is a transaction cost which discourages people because Studies have also indicated that people aren't perfectly rational they tend to they tend to be biased more towards risk aversion than undertaking uh you know neutral assessment of potential risks and Rewards another problem every pedia did not realize the importance of dissent in knowledge sharing because crypto economics taking a slashing mechanisms they are perfect for coordination coordination is not the same thing as cooperation right because um in cooperation it's there's more of a

voluntary nature involved where you want to come to a mutually acceptable conclusion with someone else and you don't want to force them to think the same way that you do in coordination and force coordination there's no room for good faith disagreement this may be a good thing in certain instances such as you know compliance you need to have standardized accounting practices or construction codes it's a good thing that you know if people aren't allowed to be creative with them but there are also instances of human interaction that required more nuanced decision making right a facilitation of diverse perspectives that you don't disagree with and you don't really want to stifle um disagreement in these issues and you

don't want to enforce Conformity because that kills heterogeneity it results in group think and everyone in general is worse off when that happens it's then no surprise that eventually the shelling point for every pedia became Wikipedia because you know if everyone gets rewarded or penalized based on what they do and you know they have to convince everyone else people are just going to go for the most obvious solution which is Wikipedia if it works for Wikipedia it should work for us as well No One's Gonna disagree we won't have these debates But ultimately if you know all you're doing is scraping Wikipedia what was the point of trying to create a Wikipedia killer

and this is related to another new um burgeoning Trend within the decentralized ecosystem known as decentralized Science Now decentralized science is a bit of a hype you know it's basically involves the introduction of blockchain technology in various aspects of the scientific process whether it's data management whether it's funding research so on and so forth what's relevant here is this idea of decentralized peer review so basically a subset of decentralized science people have identified that the peer review process is unfair and inefficient because peer reviewers are have unpaid labor there's quality

control issues you know people might be biased people might not be doing their job as peer reviewers they might be slow and inefficient what did what these design people decided is that we are going to implement staking and slashing mechanisms in the peer review process based either around Financial or reputational tokens or both and while there are many projects with many different uh proposed mechanism designs it all boils down to the same fundamental idea that if I'm peer reviewing your scientific research eventually my output will be evaluated by others they could be the authors they could be co-editors they could be other reviewers or the scientific community at

large and based on their feedback me as a peer reviewer I will be rewarded or penalized accordingly so there's a few issues with this proposal it'll probably not make uh the the peer review process more efficient or fairer and the reason behind this is that scientific progress requires paraisia right perihisia means not just the freedom to speak freely and boldly but the obligation to do so that if it basically encourages dissent and disagreement if you think there's something wrong even though you're not completely confident in what you're saying and the necessity of paralysisia has been in science it has been viewed

throughout history just a few examples of you know commonly of common misconceptions that were at one point the majority um yeah so ignac Samuel wise a Hungarian scientist he was sent to a lunatic asylum by his uh fellow scientists why because he said that doctors should wash their hands before operating on patients that people that you know birthing mothers are dying because you're not washing your hands his fellow scientists did not like that they punished him the guy who came up with the idea that stars at our distant Sons he was burned at the stake the guy who discovered that we have a pulmonary circulation system he

he was burned at the stake his name was Michael cervites now some people might say but we have moved past the whole burning at the stake aspect we're not killing people what's wrong with staking and slashing mechanisms and that is true we might not kill someone when we're penalizing their reputational or financial stake but that goes against the point of science it's that the point of scientific progress is not to create Martyrs right if there's any sort of mechanism that desenses incentivizes paraisia that is not good for science that is not good for knowledge Discovery and this is a problem even with tokenized reputation mechanisms those

that are not based on finances because once you quantify and tokenize a reputation whether through sbts or whatever you start create you start creating an objective standard by which people will be incentivized or disincentivized which again might be great in case of strict compliance is not great when you want to encourage people to be diverse and to disagree with each other so what is the conclusion here basically the conclusion here is probably going to be unpopular in this conference I don't know but the conclusion is that not everything needs to be

decentralized personally I would argue that most things do not need to be so decentralized because the pursuit of quantifiable tokenized incentives compromises fluid natural interactions among people enforce coordination is not always ideal there are situations where you want to encourage people to be wrong right the blockchain hype in general it focuses too much on buzzwords like transparency immutability trustlessness think in your regular lives do you really want to live in a world where everything is transparent everything is immutable and everything is trustless no obviously you don't you don't even want that in your financial lives forget you know everything else that goes on so if

so if this is the selling point of blockchain Technology maybe we should think uh think back for a second and you know wonder whether if we should be integrating this into every single one of our processes and ultimately this ties back into sensible system design what is sensible system design it involves knowing the fundamental culture and interactions a community organization or process requires it's not good enough to say that hey maybe blockchain can solve XYZ the first question you should ask yourselves is should I be using a blockchain to solve these issues in the first place and I believe this is a problem that we see everywhere in all these events whenever

people come up with a new blockchain invention they're basically showing us I built a cool thing they're not really able to explain to us why have I built that cool thing and that's basically the end of my talk how many how much time do I have I try to get it under 20. okay we're done so if you have any questions or answers you can always ask me later [Applause] great hey um thank you very much so um I I understand that you are opposing the idea that boxing should be used

everywhere completely completely agree there I mean I'm all in for for blockchain but I think you are right you know not everything needs to be decentralized on this topic of uh peer review so you are claiming that the design people are wrong because they are probably proposing something like this so I'm uh also I I would not claim I'm part of this side Community but I follow some of the thoughts and I do not think that this kind of solution has been um generally proposed I think what the Desai Community is saying is that the peer review process is flawed that's very similar to what open science says

and that it should be replaced by something else also that the peer review is kind of an experiment and that that experiment has failed because peer review is based on a premise that you should forbid people from publishing bad ideas but the most progress in human history has not been done by forbidding this kind of publish public Pub Publications but rather from publishing something that was super super cool like ideas uh of Albert Einstein and people like it have not

been peer reviewed and what Peter what peer review does is that it uh it's based on luck it's based on chance and it's yeah it's not an efficient process so what I understand from this I is that those people are people are claiming hey we agree that this is not true this is not correct and we should do something something better but there's not a widely accepted solution of staking slashing and stuff like that as you presented uh actually there's about five projects I can name which are you know trying to build staking and slashing peer review mechanisms and they sometimes emerge that with open review I'll probably check in my phone to give you their explicit names later if you

want now if the contention here is that peer review is not necessary for scientific progress and you can name Einstein at all I agree I agree but the point is that you know can we get can we do away with the idea of peer review entirely which which forms that basic first level of quality control within scientific discovery I'm not entirely convinced that we can do it mainly because you know all of us have limited cognitive capacity and there will always be some room for you know relying on the understanding of experts Gatekeepers so to speak so and and that's why these are decentralized peer review proposals are

being um started if you want to share your discoveries without undergoing the peer review process again you're more than welcome to do so but you know the question then comes up is how seriously will people take it I do not think that you know we should get rid of the peer review process entirely I think the solution here is just to um sort of you know encourage A diversity of of different ideas which I think most people here would agree with but the point is that crypto economic mechanisms that emphasize coordination are not doing that are not doing that right because they're info they're trying to make people agree convert upon

one point and especially in the case of knowledge sharing and scientific discovery that is not a great thing to do foreign yes I would like to add all of this um decentralized um science topic so the presentation that you gave um in my opinion it has a very negative View and it has the kind of presents the opinion that the world is good as it is right now um so maybe this is something we should also discuss because for everything that exists there is a potential for

improvement now the thing about and what you mentioned gatekeeping or centralization mechanisms or open science and curated science and one thing that's kind of missed is the role of journals so the reason why we are looking for open science is because the journals are Gatekeepers it's not peer reviewers it's model journals and the editors and most historic examples of like really Innovative breakthrough ideas didn't get published in papers not because other scientists didn't like the idea but because the editors didn't like it they thought it's it wouldn't fit in our journal and that's why they don't even pass it on to the peer review process and I think that open science has a real

great opportunity to improve the system here so that all kinds of ideas can be published and reviewed in a even more decentralized fashion that they are not just one or two chosen reviews but a larger group and the question at hand what we are looking at is is crypto a technology that can help to do this so this is basically the question and the answer will be probably in a couple of years emerge if it will work or not but in my opinion it's definitely worth trying well here's the thing I mean I wouldn't say I'm saying the world is good as it is I am saying that if you don't like the world as it is at least don't make

it worse right um now when it comes to a you know these all these decentralized science proposals the problem is many of these proposals they miss the entire point of what the blockchain is the blockchain is at its core a consensus mechanism right so it's simply saying it's simply including the world decentralized does not rob it of its function as a consensus mechanism now are any of these decentralized science proposal whether through you know implementing the open open peer review process or open publishing actually resolving any issues I would say not necessarily because ultimately you do not have a blockchain

if you are not coordinating people using uh tools such as economic economic or reputational staking or slashing I mean otherwise the alternative for that is having a centralized point of failure which is basically what journals are right that's much of the criticism that journals get so ultimately is blockchain technology solving this problem is so-called decentralization which at its core means coordinating people without that without that centralized point of failure solving this issue I would wager no I think you know the solution the solution would be to you know incense encourage and

incentivize many different sources of these centralized pointer or points of failure and hope for the best and that's what people are trying to do right now you know having competing journals or whatever that is fine you know if one Journal does not agree with you you go to another one ultimately I do not think that you know decentralized coordination mechanisms are the answer for improving the world or the state of science thank you [Music] foreign