The knowledge blockchain
I fell into my mission by accident — it certainly seemed like an accident that Wikipedia took off as successfully as it did. It might have been what we wanted and planned for, but that it did take off was as much a surprise to me as anyone. Anyway, Wikipedia’s success led to a discovery about myself, that — at least when it comes to Internet projects — more than anything, I want to develop communities in which people come together to improve the world’s knowledge.
I started building online groups on a smaller scale in the mid-1990s, with academic discussion groups. Such groups are actually how Jimmy Wales (the other co-founder of Wikipedia) and I got to know each other. I still very much believe in the ideal of collecting “the sum total of human knowledge.” That is, I think, a very worthwhile mission.
But I also believe we can do much better than Wikipedia.
I’m a philosopher at heart, though I haven’t taught at a university for over ten years; it has never been a big goal of mine to get rich, famous, or powerful. I have never been fired by ordinary ambition and avarice; I’m easily the poorest founder of a top-ten website. I’ve been offered many jobs in the centers of power, which I’ve turned down. I live in an exurb of Columbus, Ohio, quite deliberately avoiding the expensive, competitive coasts. That’s just not my bag — no offense to those for whom it is their bag.
But I still like to build things; I have lots of ideas. One overriding idea drives all the others: we can create fantastic knowledge productions when we get together online in new configurations. We still don’t know what the best such configurations are. The greatest are yet to come.
While I’m a devoted fan of knowledge (my Ph.D. specialization was epistemology), I’m also a philosophical skeptic. Knowledge is hard, not easy; it requires difficult study. This is a fact, but most information projects online have still not learned it.
Sure, the Internet has been a boon to learners. You can learn so much from the collective contributions to online encyclopedias, YouTube, Q&A sites, blogs, online courses, and so forth. For example, I started studying Ruby in 2016, and it has been easier than learning Perl was in 2002, thanks to FreeCodeCamp, Codecademy, Stack Overflow, and zillions of bloggers.
At the same time, the ease and spurious attractiveness of quick information has made more substantive knowledge less attractive. Substantive knowledge is complex and therefore more difficult to obtain and hold and less likely to flatter our simplistic assumptions.
Again, we have more reading and viewing material than we’ve ever had before, donated by creators around the world. But the quality of both our reading material and our understanding seems to have declined.
It’s time to improve the quality of our knowledge. But how?
What’s wrong with Wikipedia?
A lot has been written about the early days of Wikipedia, and a lot of people are confused about why I am critical of something I helped to create. They think they understand, but they really don’t. So let’s clarify a few things.
I am not a fan of the old Nupedia editorial system. It was broken. As Jimmy Wales said back in 2001: “Larry had the idea to use Wiki software for a separate project.” Wikipedia was my idea, so, yes — I do and did understand it. I was probably the first to understand what was going on at any sort of deep level, in no small part because policies I promoted are part of why it worked.
I don’t want experts “in control.” As I’ve said many times before, that was not the idea behind Citizendium, another wiki encyclopedia I started. I simply think experts should be given a moderate, low-key role, one commensurate with their moderate value to projects like Wikipedia. I think collaborative projects should be governed democratically — I always have.
I don’t want articles and edits to have to be approved before being made visible to all. I was an immediate and easy convert to the open source philosophy of “release early, release often.” My enthusiasm for that idea is why I proposed to create a wiki encyclopedia the very evening that I first learned about wikis.
None of that explains my disappointment with Wikipedia. It’s very, very simple:
Why? We could get into the weeds here. For example, Wikipedia doesn’t engage women nearly as much as men. That is a problem. If as many women as men edited Wikipedia, it would boast about 22,000 regular voting contributors, instead of the 12,000 it now has. Another detail is that Wikipedia deletes far too many articles and just doesn’t go deeply enough into many topics important to scholars. The “deletionists” won after I left; I’ve always been an “inclusionist.”
But I started the damn thing, so naturally I don’t think in terms of incremental changes. I think of the big missed opportunities and the dramatic improvements:
- Forget tens of thousands — why doesn’t Wikipedia have millions of regular voting contributors?
- Why does Wikipedia have just one article on each topic? With the massive amount of traffic it gets today, in 2018, surely there could be many competing articles. There could be competition. Are Wikipedians afraid of competition?
- Why aren’t Wikipedia articles, and versions of articles, rated by experts and by the general public? Why aren’t those ratings posted publicly? Doesn’t Wikipedia trust the general public to rate its content?
- Why isn’t it much easier to contribute — and in lots of different ways? Surely we can make it much easier. Why do we have to use this shoddy, 1990s editorial interface?
- And why is Wikipedia so insular, often driving away new contributors over minor issues and rarely polling its readers about what they want? One gets the sense that Wikipedians are writing for themselves, not for us.
Wikipedia is useful, but it’s just not good enough. It’s much too small (yes, small), poorly-written, low-quality, and undetailed. The root reason for Wikipedia’s mediocrity is that it is insular: it has failed to engage fully with the whole of humanity.
There is a better way.
Why don’t more people contribute to Wikipedia?
I don’t know the precise number, but considering that there were 3.7 million Americans with doctorates recently, there must be tens of millions of people with doctorates worldwide. But you don’t need an advanced degree to write a good encyclopedia article on many topics. There are 45 million Americans with Bachelor’s degrees; it is reasonable to assume there are hundreds of millions of people with college degrees around the world.
And how many of them regularly sum up and detail the best of their knowledge on the most-read encyclopedia on the planet, the fifth most popular website? Only 12,000 people have done enough work (and that’s not much) to be able to cast a vote on the English Wikipedia’s administrators.
That is a huge problem, pregnant with possibility: How do we get vastly more people working on encyclopedia articles? Doubtless, millions of people have contributed to Wikipedia (there are 33 million user accounts). But only a tiny fraction of those people are still at work.
So why don’t more people contribute to Wikipedia?
If you just ask them, maybe the most commonly given reason is — as people often tell me — that it just looks hard to contribute to. It’s a chore for many non-technical people to learn even a simple markup language. Worse, most people don’t have the patience to read through a lot of policies.
But that’s not even the half of it. A common wry observation about Wikipedia is that it’s tremendously useful, but you don’t want to see how the sausage is made. The Wikipedia community can be wonderfully generous with its time, but it is also dysfunctional; it has a habit of driving people away.
Local know-it-alls undo perfectly good edits, citing Wikipedia’s badly-understood, labyrinthine rules with an alphabet soup of acronyms only a bureaucrat could love.
Narcissistic amateurs who have read one book on a subject demand that their gorgeous prose be taken with as much seriousness as that of experts who have spent their lives writing about a subject.
Opinionated editors closely monitor articles and bark at passers-by who encroach on their turf without first stroking their egos.
Ideologues with pull and seniority twist Wikipedia’s neutrality policy so that only their view is considered “neutral,” and if you complain without paying proper obeisance, they will enlist their allies to mete out mob justice, Lord of the Flies-style.
And if, as inevitably happens in such a place, some people get impatient and upset at such unfair treatment, they must tolerate the passive-aggressive condescension of the basement-dwellers who inform them, apparently with no awareness of the ironies involved, that courtesy is an absolute requirement. If the upset person is not immediately booted for expressing their ire too vigorously, they must endure a process for adjudicating disputes that makes it even clearer that the inmates are running the asylum.
Wikipedia’s culture isn’t for everyone, that’s for sure. But you know what? Whatever its merits, it suits some people just fine, and again, the community produces useful work. Wikipedia’s culture does appeal to certain people, and if one did not have to negotiate with it in order to contribute to the leading encyclopedia online, I wouldn’t have a problem with it. Let a thousand flowers bloom.
Maybe we need more than one community
What is my point? I do have one — actually, a series of them.
First, Wikipedia is just one type of online community, which appeals to a fairly narrow (geeky, combative male) demographic. And, importantly, it doesn’t appeal to many other demographics.
Second, the number of people potentially motivated to contribute to some encyclopedia project or other absolutely dwarfs Wikipedia’s actual contributor base, probably by two or three orders of magnitude. Millions of active contributors are waiting in the wings, I think.
In that case, you might say, why not start a project that isn’t so dependent upon any particular community demographic? Maybe the problem is that many very different people are required to negotiate on one and the same article. So why not let different people submit competing articles about the same topic?
As a matter of fact, in 2000, we considered just that idea in our earliest discussions about Nupedia, the project that gave birth to Wikipedia the following year. We abandoned the “competing articles” notion early on, however, because we didn’t think there would be enough interest in writing multiple, competing articles on the same topics.
Eight years later, Google actually tried out the idea, with Knol. Knol failed, partly because Wikipedia was still in the steepest part of its growth curve and sucked the air out of any competing projects. But it was also partly because Knol still had one single community, with one set of rules, and one group of people who voted articles up and down. Knol’s writers were still writing for one organization, Google, and — I remember trying it out and having this thought — if you don’t come out on top according to the single Knol community, then what’s the point? And who wants to write to benefit Google, anyway?
This train of thought led me to a conclusion a few years ago: Maybe, just maybe, there is no single community that could possibly satisfy all the would-be encyclopedists out there.
This insight seems obvious, in retrospect. Knowledgeable writers come in many varieties. It is unlikely that any one community could appeal to them all. But even that isn’t strong enough; surely, you can’t start an encyclopedia-writing community and secure the interest of more than a fraction of all the available writers. It just isn’t possible. If you want to leverage the brainpower of humanity, you have to do it some other way.
The problems of online encyclopedias have exercised me for a long time. It began when I started working on Nupedia in January, 2000. By February, 2002, the bursting of the Internet bubble put me out of a job. About a year later, I made a more permanent split with Wikipedia, because Jimmy Wales refused to rein in the problem contributors whose bad behavior was poisoning the community.
But I wasn’t done with encyclopedias. I’ve started and advised otherreferenceandeducationalsites since then; I call myself an “online knowledge organizer.” I’ve also always felt partly responsible for Wikipedia’s ills, so I became one of Wikipedia’s critics. Always, in the back of my mind, I’ve wondered how to improve on the Wikipedia model.
In 2015, an idea hit me: Why not entirely decentralize the task of creating an encyclopedia? The focus of the problem would be to collect the best articles on every subject already online and encourage more to be written. Rather than require a collaboration on one article, anyone could submit an article about anything — hosted anywhere — to this project’s database. Wikipedia itself could be part of the collection.
But that much, being basically a search engine, wasn’t the whole idea. The really interesting part of the idea is that articles would be rated by the public, and raters would also tag themselves with their identities, affiliations, and credentials — and be able to endorse each other LinkedIn-style. A decentralized system would enable anyone to use the data about ratings (and raters) creatively. I tentatively called it “GreaterWiki.”
Unlike Knol, GreaterWiki would not constitute a community; unlike Wikipedia, its contributors would not be developing one article article per topic; and unlike any previous encyclopedia project, no one company would benefit. People could contribute to any number of encyclopedia projects, or just submit articles individually, and the articles would be collected in one place. In other words, all the data would be collected into a single, decentralized, open content (Creative Commons) database that many people could build on and use in many ways.
This basic idea — which I’ll elaborate on more later — is extremely powerful. You could show the best articles according to different demographics: experts with endorsements, French socialists, programmers, women, Christians, Muslims, etc. Users could recalculate article ratings according to different weightings of various identities and qualifications, in whatever way they pleased.
Let a thousand flowers bloom, indeed!
How I joined Everipedia
Not being a professional programmer, I couldn’t build GreaterWiki myself, and it would be very difficult and time-consuming to motivate others to build such a system — especially since there’s such a huge demand for programmers. So I more or less dropped the idea. “Some day,” I thought.
In the spring of 2016, I got to talking to the founders of Everipedia.org, which was Wikipedia’s latest competitor and easily the most interesting to come along since Citizendium. They wanted me to join them, and I wanted some programmers to build GreaterWiki. So I flew out to L.A. and met them.
Everipedia tackles some of Wikipedia’s problems head-on. It has a WYSIWYG interface that makes it a lot easier to contribute, hiding the arcane code of the Wikipedia software (MediaWiki) that makes non-technical users run away screaming. The reference system makes it much easier to add (and rank) sources, and even provides a built-in way for people with verified accounts to cite themselves.
Everipedia is also much more welcoming to newcomers. You can create an article about anything you like — even about yourself. An inclusionist at heart, that appeals to me. The world needs more encyclopedia articles, not fewer. Everipedia is built on that idea. I also liked the fact that the Everipedia crew is young, diverse in ideology and ethnicity, hard-working, and fun.
So when we met, I was impressed by what Everipedia was doing, and they liked the GreaterWiki idea. But no deal emerged.
But I reflected that there would be a problem even if I did created this website: whatever website organized this decentralized network would have a community, too, and that would naturally drive away many people. What there really needs to be, I thought, was not a website but an open protocol like RSS or HTTP itself. But I knew even less about how to formulate such a protocol and get it widely adopted.
Then one of Everipedia’s co-founders and developers, Sam Kazemian, called me last September (2017) with exciting news: Everipedia was seriously thinking of moving its contents to the blockchain.
I had a rough idea of what blockchains were, but Sam expatiated on the concept and then suggested that the blockchain is a perfect fit for the protocols we would need to make something like GreaterWiki work. I thought it over; I saw that it solved the protocol problem I had no hope of solving; I already knew the guys; they made a good offer. So I was hooked. I joined the effort.
What would an encyclopedia blockchain be like?
Many people still don’t know what blockchains are, which is OK; it’s complicated. Let me give it a go. Even for those who do, I still need to explain what an encyclopedia blockchain might look like.
The typical sort of explanation of blockchain is mind-numbing. I won’t blame you if you can’t get through the following paragraphs.
A blockchain is a data ledger, comprised of “blocks” of data, each of which has a cryptographically hashed header that points to the previous block (which is why it’s a chain of blocks, and why it can’t be edited; so it’s called immutable). The blockchain is copied in toto on different nodes located in many different places, and owned by many different entities (so it’s decentralized, a little like peer-to-peer file sharing).
The entire blockchain is made into a store of value — meaning that certain blocks (depending on the protocol defining the blockchain) can be bought or sold as tokens or coins. Adding a purely financial block simply records a newly-mined coin or else the transfer of some quantity of coins from one wallet (controlled by a person) to another.
In the case of Everipedia’s blockchain, newly-proposed articles (and other content) will, if accepted to the blockchain, unlock some quantity of new coinage. Whether a proposed new block is added to the blockchain at all is something that can be calculated by all the nodes according to a commonly-agreed protocol. The first node to make this calculation and confirm the addition receives a small portion of the value (coin) produced by adding the new block.
Perhaps the most important aspects of the blockchain concept you need to understand, for present purposes, are these: (1) it’s a store of data (like a database), (2) it’s decentralized, and (3) it is a store of real-world value.
It’s also useful to say what an encyclopedia blockchain isn’t:
- It isn’t an encyclopedia. It contains encyclopedic content, but more than that. It’s more like a peer-to-peer database.
- It isn’t human-readable. I mean, you can view a blockchain’s feed, but it will look like code.
- It doesn’t live on one server. It’s spread around an arbitrarily large number of “nodes,” kind of like (but not exactly like) a peer-to-peer file sharing network. This decentralization is absolutely essential to the benefits of the blockchain.
- It isn’t owned by Everipedia. It’s being started by Everipedia, but it’s owned by whoever contributes to it.
- Everipedia isn’t building the underlying blockchain technology. The encyclopedia blockchain’s blocks will actually be part of the EOS blockchain, which is engineered for projects like this.
- It isn’t controlled or managed by Everipedia. Its governance is determined by whoever has certain tokens (or coins) in the system, as calculated by the protocols that define the blockchain.
- Most of the content that the blockchain stores isn’t proprietary. The tokens certainly are, but those are different from the content. The content will be Creative Commons licensed.
Finally, a blockchain encyclopedia solves a problem that I wasn’t even trying to solve: allowing people to earn value from the work they put in. Cataloguing knowledge is labor-intensive and, for the most knowledgeable people — who are regularly pushing the frontiers of knowledge, rather than explaining the basics — it can be rather tedious.
This is one reason why Nupedia and Citizendium didn’t take off, and why Wikipedia still isn’t comparable in quality to the best specialist encyclopedias. Professionals generally require money, or else some other sort of professional compensation such as plaudits from fellow experts, if they are to craft difficult-to-write articles that draw on their broad expertise.
That’s all I’ll say about the blockchain architecture.
How Everipedia will move to the blockchain
“So it’s another encyclopedia project,” you might think. “It’s unlikely to be able to compete with Wikipedia, right? Let’s be realistic.”
The premise of this reaction is incorrect. An encyclopedia blockchain would not be just another encyclopedia, and the ways the blockchain differs from traditional encyclopedias are precisely why it has a real chance to compete with Wikipedia, as I will explain.
Everipedia will launch this blockchain — which we’re calling the Everipedia Network — in two phases. First, we will make all of our current articles, which includes the English language Wikipedia (as edited by Everipedia) plus about a million more, available on the blockchain. The blockchain will have three modules: an article module, a governance module, and a token module.
This will already be a significant development, because starting from the base of a vibrant, growing community, it adds an incentive for users to be compensated with tokens for their work. Remember, Everipedia incorporates all the articles in the English language Wikipedia. You will now be able to unlock a store of value — the network’s “IQ” tokens — for working on those articles. Maybe more importantly, tokens will empower users to join in the democratic governance of the network. Unlike Wikipedia, encyclopedists will be able to govern an encyclopedia protocol democratically. In time, a charter for this nascent group of token holders will be developed.
A second phase—which, realistically, won’t be fully developed until 2019—also has the potential to be revolutionary, taking us well beyond Wikipedia and Everipedia. At this stage, we’ll add:
- Competing articles. Suppose you think you can do better than Wikipedia’s (or Everipedia’s) article. You’ll be able to upload a competing article to the blockchain, either from Everipedia or from other sources.
- Other encyclopedias. We will enable other encyclopedias to add their content to the same blockchain and go head-to-head with Everipedia (and thus, also, with Wikipedia) in the ratings.
- Ratings. The public (individually and in groups) will be able to rate all of the articles on the blockchain. The rating data will be fair (tied to carefully-verified real world identities) and open, and completely independent of blockchain participation and ownership.
- Links to articles not on the blockchain, also for purposes of rating. Even if an encyclopedia doesn’t want to put its articles on the blockchain itself, we will still be able to link to them and rank them. In this case, while their content wouldn’t be part of the blockchain and thus couldn’t be compensated, a link to their content would and could.
- Metadata: user tags. To ensure that a wide variety of views, including especially well-informed expert views, are represented, we will enable users to volunteer information about themselves, such as credentials, gender, race, ethnicity, nationality, religion, and political views. Also, we will support users and organizations being able to rate each other’s expertise, much as LinkedIn does.
- Metadata: organizations. In addition, we will enable organizations to identify which users are members. This will enable those organizations to curate their own specific content. By “organization” we mean something quite broad: academic associations, professional organizations, universities, departments, government bureaus, clubs, and ad hoc groups. As far as the blockchain is concerned, no organization, no matter how distinguished or authoritative, will be privileged over others. So, for example, while an American Philosophical Association group might determine who its members are, it would not be able to control membership in the philosophy category.
- Varieties of article rating filters. With the aforementioned metadata, collected as part of a neutral, open protocol, users will have the data needed to ascertain the ranking of articles according to various groups. While the Everipedia Network will not itself publish aggregate data, websites (like Everipedia.org, as only one example) and apps that use individual rating data will make it possible able to see the top-rated article for the topics “capitalism” and “socialism” according to Democrats, Republicans, men, women, professional economists, CEOs, activists, etc. Weightings of ratings by different combinations of groups (e.g., 50% Republican and 50% Democrat) might be used to determine the most “neutral” articles.
- Much more. I can’t possibly cover all of the features of the network here. See our white paper for phase one and this (Dec. 2017) proposal for phase two.
Why think Everipedia will grow?
Let me begin by explaining why I’m bullish about the future growth of this community. The “GreaterWiki” notion behind this plan envisions a group of participants and supporters that would grow much larger than Wikipedia’s. Very well—why think it would?
Not limited to one community. While Everipedia.org will start this blockchain network, it won’t be limited to any one website or community. We are already in talks with a number of reference publishers about publishing to the Everipedia Network. In the end, there will be many different ways to put content onto the blockchain.
This handily solves the problem that has kept participation in Wikipedia relatively small (considering its traffic), and which also prevented projects like Knol and Citizendium from taking off. To participate, you won’t have negotiate with any single community, like Wikipedia. You just have to follow a certain neutral technical protocol, much like HTTP or TCP/IP. It’s possible that you won’t even have to publish via a community at all.
All of this has exciting consequences. Let’s spell them out.
A global project to rate and rank competing statements of knowledge. Imagine a time, not too far in the future, at which the Network had aggregated all the world’s encyclopedia articles (or at least links to them) on one blockchain. Imagine the Network has a protocol for registering ratings, by uniquely identified users, of those articles. For the first time, there is a credible review system for Wikipedia articles — alongside reviews of articles on the same topics from many other sources. That means a system that rates and ranks all of the world’s claims to knowledge (in the form of encyclopedia articles).
It’s not a stretch to say that the mere existence of this system would be intensely interesting to many intellectuals the world over. It would be especially interesting to writers. If you wrote the top-rated article according to some group of your peers — what a career perk that would be in so many fields. Even if Google continues to unfairly place Wikipedia articles at the top of its search results, it will be meaningful to writers for their work to be rated above Wikipedia.
Compensation for work that was previously volunteer-only. Tokenizing encyclopedia articles means you’ll unlock tokens for writing. This is what Theodor Forselius, Everipedia CEO, calls an “incentive layer” for the blockchain. This is one reason why it matters that this project is being started on the blockchain. It is no accident that it is being developed by an energetic technical team, with significant funding and enthusiastic support by the blockchain community.
Everipedia and the blockchain community will evangelize the project: “You can actually be compensated for your work — and become co-owner of the resource — simply by adding it to the blockchain.” For the first time, work on Wikipedia articles, which was previously volunteer-only and uncompensated, can create a return. Of course, submitting articles through other sources will also give you the possibility of being compensated.
Truly broad-based and open. Like Everipedia, the Network will impose no restrictions on encyclopedia topics; there will not be a blockchain-wide notability policy, for example. There will be no need, in fact, to negotiate any controversial blockchain-wide editorial policies. Controversial policies shouldn’t be part of the protocol at all, because if they were, the blockchain would need a stringent, selective editorial process—which would inevitably make it centralized and thus susceptible to various kinds of bias.
But this doesn’t mean that the network will be full of nothing but garbage and spam. There will be minimal requirements, such as that articles submitted must contain a minimum number of words, that the content does not constitute copyright infringement, that the content is an attempt to introduce some general topic in complete grammatical sentences in a recognizable language, etc.
There are other reasons to think people will get excited:
- While many people appreciate Wikipedia, a lot of others are deeply disappointed with it. This project is a credible and significant opportunity to create something better.
- It will be the first complete encyclopedia on the blockchain.
- The blockchain will also permit exclusive professional groups to share their member lists and article ratings. This will not only provide a fair way for experts to organize their online editorial activities, it will also give a voice to experts who are not yet at the top of their professions. This is apt to generate excitement among graduate students, post-docs, and assistant professors.
These various advantages, when taken together and implemented, will be huge news. The basic narrative is compelling: “Who can compete with Wikipedia? All the rest of us, taking ownership of a neutral, decentralized protocol.” I’m confident that this will prove to be a very popular plan.
How will this system result in higher-quality articles?
Let’s answer another potential criticism. Why think that this system would result in higher-quality articles than Wikipedia’s — which aren’t that bad, or so it is widely thought. Wikipedia’s defenders say that many hands make light work; open collaboration turns out to be a surprisingly robust way to produce fair quality.
That is all easy to admit. Wikipedia articles are, however, frequently amateurish — just ask anyone who is an expert in a field, especially outside of technical fields, to give a serious review of the Wikipedia articles about their areas of specialization. Also, a problem that seems to worsen as the years go by is the fact that some Wikipedia articles are quite out-of-date.
Quality matters. When it comes to serious scholarship and education, Wikipedia really isn’t good enough. That’s why I have, occasionally, subscribed to Encyclopedia Britannica for my two sons: its quality is much better and more consistent than Wikipedia’s, and when it comes to education, I want my children to have the best.
So why think the Everipedia network would be an improvement over Wikipedia?
Let’s begin with some interesting, but perhaps less consequential reasons:
- Giving people an incentive to participate in the form of tokens, which have both monetary value and confer a right to govern the network , will motivate users to do their best.
- Users will have to stake a small amount in order to make contributions to the blockchain. We expect this will prevent the most egregious problems, such as vandalism and attacks on the network.
It’s very possible that financial and governance incentives, together with the benefits of staking, will be enough to make Everipedia significantly better-quality than Wikipedia.
I think the competition inherent in the second stage of the Everipedia Network — together with the ratings system — will more effectively secure the highest quality. But is that really right? How might this competition work?
The following speculations are admittedly based on an assumption that there would be many people motivated to rate articles; I’ve already explained why I think the community will grow.
First, note that academic and other traditional reference publishers will be able to join the network (and we have heard from a few that are very interested). Even if they are not part of the network, we will be able to link to their content and rate it, thereby preparing the “lists” for a head-to-head knowledge joust (as it were) among all writers and publishers.
Imagine, then, a modern platform that makes it easy to surface every encyclopedia article available; whether or not you can get past a paywall is another matter. The articles are ranked according to average rating by default. But you don’t have to use the default. You can (and doubtless you’ll be interested to try out) what the ranking would look like if you give special weight to ratings by liberals, or by conservatives, or by experts in your field. Suppose you’re the first person in your profession—say, Wonkology—to view and rate a certain page of articles. Power! You can determine what the initial ranking will be!
If you are the tenth or hundredth wonkologist to weigh in on wonkological articles, you will naturally be interested to compare your own judgment to that of your peers. In all likelihood, you will have some disagreements. You might well be motivated, then, to do your small part to nudge the ranking toward the correct views, by supplying your own ratings.
Then think of this: if you are a specialist in frippernomics, and you judge all of the articles to be awful, you will be motivated to earn the glory of a top-ranked article. You and other frippernomics experts will compete to outdo each other until there really is an article about which most people are willing to concede they can’t do any better.
The result? Extremely high-quality articles will be written about every topic in every field—at least as long as there is adequate competition.
These observations will surely be true not just of academic, technical, and other professional topics, but also of hobby and fan topics. It’s easy to imagine geeks writing, improving, debating, and rating many competing articles about New York City, Mordor, downhill skiing, Buffy Summers, Beyoncé, the 1909 VDB penny, your high school and college, Yosemite National Park, and in short everything that animates our interest.
Because there will not be just one set of ratings, but different sets based on different characteristics of raters, there could be competing rankings. You might not have a prayer of writing the top-ranked article about freedom of speech, because your views on the topic are out of the mainstream; but, as a libertarian (or philosopher or German), you might have a chance to take the top spot among libertarians (or philosophers or Germans).
Quality may be measured in terms of completeness, accuracy, adequate citations, correct mechanics, and readable style — at least. It seems clear that competition could maximize quality in all of these metrics. But what about neutrality? If different identity groups, nationalities, academic fields, etc., are enabled to write competing articles, won’t that mean there will be a number of competing “bubbles”?
Having rating data associated with different biases will make it possible to approximately a middle-of-the-road opinion, if not, precisely, a neutral one. If you wanted an all-American rating that was neutral with respect to party and ideology, that could be easily arranged by weighting the ratings appropriately.
Another possibility — not yet decided for sure — is that we might ask people to rate articles according to a variety of qualities, among which would be neutrality. While you might disagree with your ideological foes about what is neutral, the canceling effects of different groups, weighted in different amounts, will arguably produce a more defensible opinion.
The goal: ever-improving, brilliantly-written articles available universally via search
Lastly, I want to discuss some goals I have in mind for this network. To be clear, I do not set the company goals for Everipedia, and obviously I cannot speak to goals of future partners or the community. There must, in fact, be a plurality of goals. That is, if the network is decentralized, that means there can and (given human nature) must be many visions for the future, since human beings are naturally animated by many sometimes conflicting values.
So I can say only this: It is the following vision that motivates me to stay involved. But see if it inspires you too.
At one level, probably the main goal of the project is that humanity will be able to discover consistently reliable, ever-improving, and brilliantly written encyclopedia articles via search results. Right now, Google usually surfaces Wikipedia articles at the top of its encyclopedia-type searches — and as I said, that is certainly not the best that humanity can do.
If you were able to write a better article than Wikipedia’s, it would be nice if you had a real shot at gaining the coveted top spot on the Google search. Right now, that is not the case. Many Wikipedia articles rank more highly than better articles from other sources. If you want to contribute to that top-ranked Wikipedia article, good luck: the article’s self-appointed watchers jealously guard their prerogatives — frequently barking, as I said, at unwanted interlopers, regardless of qualifications. Nevertheless, Google shows users the Wikipedia article so often because Wikipedia has become the best-known brand of encyclopedia.
Google’s algorithm is obviously imperfect. It needs more meaningful human inputs, together with some way for better and updated articles to leapfrog to the top, if they are better than the other articles now available.
If a blockchain system surfaces the best articles using credible rating data, that would be precisely the sort of human input Google (and DuckDuckGo and Bing) might use. It’s possible that Google will tap into the blockchain’s API directly. If that happens, you might no longer have to negotiate with Wikipedia to work on the top-ranked article, because that spot wouldn’t be unfairly reserved (as it were) for Wikipedia any longer.
If the Everipedia Network finds and credibly rates all the world’s encyclopedia articles, it is likely that Google would help itself to what would be, after all, a treasure trove of valuable data. But if Google eschews the blockchain, I can imagine users learning to eschew Google if they just want to find the basic facts on some topic. Such users would just use any number of other front ends for this new “World Brain” (including Everipedia).
One way or another, the mere availability of a rich, rated, and growing network of encyclopedic content would, I think, inevitably change the way we learn new things via search engines.
A bit of speculative fiction
The real consequences of any radically new technology are notoriously difficult to predict. But let me share the vision I have for this network in the form of some “speculative fiction” — written in 2028, after ten years of development:
Within the first few years of its existence, Everipedia got its own content on the blockchain, invited people to share ownership and management of the Network, and started rating articles. Soon after, it supported multiple articles on the same topic, some uploaded via Everipedia.org, some via other sources. Other websites and apps started using the blockchain soon after. Everipedia, its staff and broader network, found encyclopedia articles everywhere, both free and behind paywalls, encouraging more and more ratings, which were compensated by IQ tokens, as were articles and edits.
Beginning essentially as a fork of Wikipedia, the Everipedia Network rapidly grew. The opportunity to rate Wikipedia articles — and then to find (and upload new) articles that were better — turned out to be extremely popular and cathartic for academics. After three years, there were Network groups for many major academic and professional associations, competing furiously (for the first time ever) to write the very best summations of topics in their specializations. Similarly, many civic, religious, political, hobby, and fan groups sprang up. For anyone who wanted to absorb the local knowledge of San Francisco, or fan lore about The Lord of the Rings, or the latest about a zillion other topics — there was now no substitute.
Wikipedia, stuck in its ways, got on the bandwagon late, and by the time it was posting its own versions of articles to the blockchain directly, different Wikipedia forks, led by Everipedia, had already invented ways to choose the best version of Wikipedia-sourced articles. If you wanted to collaborate with others to have a chance at writing the best and most popular article about some topic, you no longer had to negotiate with Wikipedians.
Many people had migrated to Everipedia, which was friendlier and more open. In terms of quality, Wikipedia versions of articles were not always the best; Everipedia versions were frequently better-rated. Wikipedia became just another community contributing to the world’s knowledge.
Google started using the blockchain data after four years. They too were late to the party, because by then, people had gotten used to using Everipedia, DuckDuckGo (an early adopter), Britannica, and other apps that offered a wide variety of front ends for the blockchain.
Today, it is generally agreed that the Network’s impact is equivalent to the invention of the Internet itself. Who knew how important it would be that we create a decentralized knowledge marketplace, a protocol that would enable us to easily find the best of human knowledge, and to fill in the gaps? In 2018, the year the Network went online, there were a few encyclopedias, dominated by Wikipedia, representing a few viewpoints on their topics (usually biased and often mediocre in research and writing). Since it was so laborious to find and judge different portrayals of a topic, you basically just took whatever Google gave you, which was the Wikipedia article.
How things have changed. For schoolchildren today, it is hard to imagine a world in which top experts, individually and in groups, do not go head-to-head to share summations of what is known, which are ever improving, always up to the minute, brilliantly written. At every needed reading level, too. Remember struggling to finding age-appropriate encyclopedia articles for your children’s reports? Educators and parents cheered as the educational riches heaped up.
In 2022, support was added for a long-dreamed feature: confirming that particular claims were supported by particular sources. While this might sound like a minor detail, it now lies at the heart of our ability to achieve limited consensus across groups. Different users (representing the Network’s many different groupings) rate sources and agree, or disagree, about how far different claims are supported by those sources.
Do you remember when “fake news” was a thing? It’s not that people don’t believe false things anymore; it’s just that they’re now so easy to refute with the help of some of the best possible resources that nobody bothers to publish easily-refutable stuff anymore. Maybe more importantly, everybody knows where to go to find answers about what is widely agreed and what is actually disputed (across politics, religion, nationalities, etc.), and just how much expert support there is for various claims. Some new apps, drawing on the data, even auto-generate and auto-rate articles based on the trove of facts and their ratings.
The Everipedia Network also accomplished something that no one really expected an encyclopedia to do back in 2018: it made it possible for us to understand each other. If you want to understand the a certain traditional Sunni take on sharia law or jihad, lo and behold, it is there. It is drawn not from some mullah’s pronouncements but based on the ratings of thousands of distinguished Muslim scholars. You can compare it side-by-side with a moderate Muslim view as well as of sociologists of religion who are not Muslim.
That’s just one example. Scholars report that comparing the top-ranked articles according to different groups is absolutely fascinating and has already resulted in many breakthroughs and fascinating cross-cultural conferences. Competing positions on every controversy now has their own loci — position papers, as it were — voted to the top by experts who take those positions. Debates before the Network came online were like shouting at each other in a crowded bar; it was hard to hear, and it was hard to understand each other properly. Refuting straw man positions was quite normal.
Then we started competing to write the best explanations of every point of view about every question, and the signal-to-noise ratio shot through the roof. Across nations, religions, ideologies — across every way that we are divided — we understand each other so much better than we ever did. The quality of debate, of human cross-cultural understanding, and, dare we say it, of knowledge and rationality have grown enormously in a very short period of time.
In 2018, one of the most significant problems was the division of the body politic in many Western countries — but especially the United States — due to failure to understand and take each other seriously. We are awestruck at how much things changed and how quickly since then. If you engage in political debates with your friends, how many times did you say the following, as the Network was coming online? “Here, look at this article that I found on the Everipedia Network. This is pretty much what I think. You are attacking a straw position. If you want to know what I think, just read this.” And then your friend replies, “OK then, read this, which I also found on the Network. You don’t understand me, either.” And then, suddenly, the whole debate is elevated.
It’s not like we couldn’t cite articles to support our views, back in 2018 (and did we ever! There were hot takes galore!). It’s that it was basically impossible to find an article that the other side would take seriously. This has changed. Top-rated articles we cite at each other today are written by experts (or collaborations informed by experts), are brilliantly written, and if they make mistakes, they aren’t simple mistakes.
That’s how, some of us now believe, the Network has elevated discourse: it is now so much more difficult to lie about what our opponents believe. We learn more about each others’ views and engage at a much deeper level than people did in 2018.
Disappointingly, but unsurprisingly, not everybody approves of this development. As many have trenchantly observed, it’s becoming delightfully easy to spot the intolerant dogmatists and ideologues in our midst: they drop hints that they hate the Network. They seem offended that any other viewpoints are even permitted. Of course, they’re not above using articles they approve of, but they seem to hate that other views are so beautifully described as well.
They don’t put it that way, for fear of seeming anti-intellectual. They say other views are “dishonest” or “systematically wrong” or “racist” or “heresy.” They rarely come out and say it, but they seem to wish that only their own views were so clearly and brilliantly expressed. The easy availability of contrary viewpoints to impressionable students and the general public is a grave threat, as far as they’re concerned, to the success of their ideology, religion, or philosophy.
Probably, such ideologues should be worried. The Network is teaching the world that the truth is complex and cannot be arrived at via propaganda and censorship. China and Iran have become some of the Network’s loudest critics, and while they have tried hard to censor it, they can’t, so far, thanks to blockchain technology. The blockchain and IPFS are spread around so many nodes, and so many websites and apps use it, that attempts to censor it amounts to an unwinnable game of whack-a-mole.
In 2018, no one would have said that the Internet had enlightened the world. To be sure, there was a lot more information easily available then than there was in, say, 1998; but the quality of information was appalling. And to be sure, the Internet had already taught the world a great deal by 2018, making learning of all sorts more readily available and raising the standard of living almost everywhere. So that’s nothing new.
It’s still probably going too far to say that the Network has enlightened the world, per se. What is clear is that with so many authoritative sources, and from so many different viewpoints, that it is much easier for teachers to prepare authoritative resources, for textbooks to be both factual and balanced, for journalism to be fact-checked, for friends arguing on Facebook to gather facts for their debates, etc.
Making so much high-quality information instantly available is a huge achievement. But another achievement is the propagation of some philosophical lessons that, if learned properly, might really enlighten us:
- Knowledge is amazingly difficult; we cannot reduce important truths to slogans, and we should not let ourselves be persuaded easily.
- The quality of information is crucial to education, research, journalism, and yes, even daily life — much more important than we knew a decade ago. We were wrong to be so impressed by quantity of information in the early Internet and to neglect quality.
- Editors, peer reviewers, and other gatekeepers are just not as effective as a system that allows us easily to poll a broad assortment of people, and especially experts.
- The way to motivate enough people to trust and participate in such a poll is to guarantee that the system is decentralized, neutral, and impossible for some single authority to co-opt — in other words, it needs to be a technical protocol.
- The thoughts of our opponents on every side are more coherent, and capable of being expressed more intelligently, than we ever knew possible.
- Maybe, now that we can more easily learn the depth and reasonableness of our opponents’ beliefs, humanity isn’t doomed after all.