"Let's Run The Experiment": A conversation with Chris Dixon about DAOs and the future of organizations online
The Web was supposed to usher in a renaissance of new ways of organizing people and creative talent. But mostly we just got giant public corporations. Could the blockchain offer us a way out?
You’re reading a post from my Adjacent Possible newsletter. Sign up to receive more essays and conversations like this one in your inbox.
This is part two of my conversation with Chris Dixon, general partner at the Silicon Valley venture firm, Andreessen Horowitz, and one of the most persuasive explainers of the various emerging technologies generally called Web3: cryptocurrencies, blockchain, tokens, NFTs, and now DAOs. You can read part one here.
It makes sense that you would cite Axie Infinity as a leading example of how the Web3 technologies are actually starting to work at scale. As I wrote on Twitter today, linking to your original piece, I’ve long felt that game worlds are seedbeds of new kinds of relationships — social and economic — that eventually make their way to the non-gaming world. Because their participants are often young and tech-savvy and blissfully unversed in all the ways the tech is “supposed” to be used, and because the pleasure in games comes in large part from making up new rules, games tend to produce far more experiments than real-world spaces do.
And that’s a good thing. We need more experiments! One of the key qualities that got me so interested in the early Web was this sense that the medium was going to enable all kinds of new experiments in how we organize ourselves into communities or collectives—or corporations. And maybe more importantly, new forms that didn’t quite belong in any of those buckets were going to emerge as well.
Some of that emergence did happen. Wikipedia, for instance, is one of the most astonishing triumphs of the past twenty years, and no collaborative enterprise—as far as I know—has ever been organized at such a vast scale, with that particular architecture, with such world-changing success. But truthfully, a lot of our attention online has gravitated to one increasingly dominant economic model: an advertiser-driven platform maintained by a venture-backed startup that eventually becomes a publicly traded company or is acquired by one. There have been some important tweaks to that model recently: “benefit corporations”, like Kickstarter or Warby Parker, come to mind. But as someone who was hoping to see a great awakening of organizational forms, it’s been a little depressing to see us gravitate back to the same old blueprint that has ruled, in Silicon Valley at least, for more than fifty years now.
All of which is to say: there is a part of me that is instantly taken by the premise that the blockchain has opened up a new door in the adjacent possible of how we form communities, via what are called DAOs, or decentralized, autonomous organizations. You mentioned Venice in your introductory post announcing the A16Z investment in Friends With Benefits, maybe the most prominent DAO in existence currently; in my book Emergence, published twenty years ago (which is insane btw), I talked about the guilds of Renaissance Florence, true peer-to-peer organizations that have had a longer history than joint-stock public companies, if not quite as lucrative. So on some level, you have me at hello with DAOs.
But I guess I’m a little blurry on the details. Friends With Benefits has a manifesto where they describe themselves as an organization of “creators, rebels, artists, thinkers, and doers.” That sounds great! (I would like to be all those things, though maybe I wouldn’t say it out loud if I was.) And there’s an admirable statement of principles that tries to lay down a set of norms for the community, which is encouraging to see at the start. But I need your help explaining what it’s actually going to do.
One reason this is personally interesting to me is that I sense a comparable network happening here at Substack, where a huge amount of the value created is flowing to the creators. Now, it's true Substack authors don't participate in the enterprise value of Substack, Inc, but theoretically that could be part of the package through more traditional means: i.e., successful Substack authors get some equity in the company, along with the lion's share of subscription revenue.
I’m all for peer-to-peer collectives, but what is preventing people from implementing these kinds of org/governance structures off the blockchain? We have a commercial royalty system in the music business that distributes the value created across networks of hundreds of thousands of people, if you include all the folks getting two cents a year for a recording their dad played on in 1975. That’s not really a technical problem, in my mind. It could be fixed by just changing where the money goes, which we could just decide to do, without any new technology at all. Is that naive? I’m sure it is somehow.
Basically I’m sold that the world needs decentralized autonomous organizations. Why do we need the blockchain to make them reality?
Let’s talk about the current state of the internet. It’s been 28 years since the launch of Mosaic and the consumer web. I agree that Wikipedia is an astonishing triumph but of the top 20 websites it’s the only one not owned by a large corporation. I can’t think of a single example of an internet company giving meaningful equity to users (maybe there is an exception I don’t know about, but if so it’s certainly a rare exception). There have been tens of thousands of internet companies funded over these 28 years (our firm alone has funded many hundreds). In my experience most of the companies are run by highly creative, idealistic people. Many of them started off with products that were optimized for the needs of the users and community and almost all of them ended up optimized for the needs of the company.
My conclusion after observing and participating in this over roughly 20 years is that it’s not a problem with the people but a problem with the model. The core issue is there is a fundamental misalignment between the needs of a corporation and the needs of a network. Corporate-owned networks follow a predictable life cycle. When they start out, they try to attract users and 3rd-party complements like software developers and creative people. As they grow in popularity and move up the adoption S-curve, their power over users and 3rd parties steadily grows.
When they hit the top of the S-curve, their relationships with network participants change from positive-sum to zero-sum. The easiest way to continue growing involves extracting money and data from users and complements. The most famous historical examples of this are Microsoft vs. Netscape, Google vs. Yelp, Facebook vs. Zynga, and Twitter vs. its 3rd-party clients, but it happens at a smaller scale everyday, with almost every company that operates a network. (I wrote an essay a few years ago that goes into more detail on the topic).
A company can try to resist this logic but it’s extremely difficult. The reality is that if your goal is to optimize revenue to the company, shutting down APIs, algorithmically manipulating users, harvesting personal information for better ad targeting, reducing revenue shares, and all the things people don’t like about internet companies is the optimal strategy. If you don’t behave this way, your competitors will, and they’ll make more money, the stock market will reward them with a higher market cap, they’ll attract the best talent, will reinvest more money back into the product and datacenters and marketing and so forth, and they’ll win. Wikipedia is a miraculous exception where it turned out the original product from 2001 was the winning product and needed almost zero improvements or investment, just some yearly donations for operating costs. It’s not the norm, and I think the simplest way to prove this is to look at lists of top websites and ask yourself after 28 years and tens of thousands of startups whether this is just a failure to run the right experiments or in fact a problem with the system.
You asked whether you could accomplish the goals of blockchain proponents with legal innovations instead. The current best practice after all this time is to govern the relationship between users and internet services with very long, legalistic privacy policies and terms of service which no one ever reads, nor probably could understand if they did, and the service provider can change on a whim (you probably get emails from internet services when legal agreements change which I’m guessing you never read). This is the best practice — what the most forward thinking companies still do. I suppose you could try to do something simpler, and add equity grants into the mix. People have tried. I started a security company once where one goal was to explain these legal agreements to users in simple ways.
The core issue is again the model. When you have an internet service that serves hundreds of millions or billions of people, there is a huge asymmetry of power and knowledge between the service provider and individual users. To fix this, you’d want a system that provides a way for users to organize and act collectively. You’d want users to receive a straightforward digital representation of their economic and governance rights. You’d want a system that has unambiguous rules along with mechanisms to enforce those rules. And you’d want this all to be available globally and accessible to anyone. Sounds useful, right? Well, the system I’ve described is basically a blockchain. We’ve run the experiment for many years without blockchains and now it’s time, I believe, to run some experiments using them.
Now let me address your question about DAOs. The word DAO is probably overloaded at this point, but in short refers to a new kind of organization that uses blockchains in the way I alluded to above. In some loose sense open source software projects ranging from Linux to Ethereum are DAOs. They make decisions about software and network upgrades via roaming groups of users and developers on various websites discussing and arguing over things. It’s chaotic but mostly works. These systems use “off-chain” governance in which network participants “vote” by deciding whether and how to upgrade the software they are running.
A few years ago, people started asking whether these governance systems could be improved by using “on-chain” governance, which basically means giving network participants a more formal way to vote using tokens. Probably the first and most important on-chain DAO is called MakerDAO. MakerDAO is a complex system for generating what are known as stablecoins (it’s a fascinating system — here’s the whitepaper that describes it). I won’t go into the details but the key idea is the system has a governance token which lets the holders propose and vote on decisions including system upgrades. Maker has weekly community calls and a rotating group of core members who do the heavy lifting. It has generally worked quite well both as a product and as a community. Maker’s model has become the standard in an area known as decentralized finance (DeFi). Uniswap, Compound, and Aave are other examples of popular protocols that employ that DAO model. In my view this experiment has been a success so far, although it’s still very early.
DAOs can also be used to fund public goods on the internet. An interesting recent example is the Ethereum Name Service (ENS), which is essentially a blockchain-based name service analogous to DNS that maps human-friendly names (“cdixon”) to human-unfriendly names (IP addresses, blockchain public keys, etc). ENS has been around since 2017 and has been widely adopted in the Ethereum community. Last week, ENS launched a token which they airdropped to users, which means 25% of the tokens went to people who had actively used the system in the past. The token is used for governing the system (similar to the role ICANN plays with DNS) and down the road could also participate in its economics. Another 50% of the tokens went to the ENS DAO treasury, which the community can use to fund products and services to grow the network. At today’s prices, the treasury has over $2.5 billion. This is partly driven by speculative token prices, and we might be in a frothy market, but even if you give the treasury a 90% haircut for good measure, it’s still one of the most well-funded public goods projects in internet history.
Compare this to how other internet public goods are funded and governed. OpenSSL is open source software that powers much of the encryption on the internet. A few years ago there was a major bug, which highlighted the fact that the organization was woefully underfunded, with only two full-time employees and a budget of under $1M per year. Basically with internet public goods you either find a corporate sponsor who views you as strategic (Linux is supported by big tech companies who view it as a strategic counterweight to Windows) or languish in obscurity, begging for money.
In terms of governance, many people don’t know this but the organization that controls all the “dot org” domains was almost bought by a private equity firm last year, which would have given them the ability to change prices, deplatform organizations, and more. I expect you’ll see more drama around DNS and other public internet resources as the internet gets increasingly politically contentious and geographically fragmented. With ENS, by contrast, the governance is baked into code, which is enforced by the Ethereum blockchain. There are definitely ways this governance can go off the rails, and the community needs to thoughtfully evolve the model over time, but I think it’s a big improvement over the old model of governance.
So I would argue that DAOs have shown strong promise for governing and funding technical projects. An interesting question people are wrestling with now is how to extend the DAO model to less technical, more mainstream applications. FWB which you mention is kind of a social club for technologists. There are a bunch of DAOs organized around buying digital assets. (PleasrDAO recently bought the Wu-Tang album Once Upon a Time in Shaolin). We’ve seen other DAOs that are focusing on e-sports, real estate, fashion, media, venture capital, open source software, and pretty much any topic that attracts passionate internet communities. The DAO structure lets these communities build treasuries, develop governing rules, and act collectively. Members can be anywhere in the world. New DAOs can be spun up in minutes. If those communities instead tried to form traditional companies, it would take weeks and require lots of paperwork. Many of the prospective members would likely need to be excluded due to variations in local laws and other reasons. There would be thorny challenges around governance and enforcing rules.
The other appealing thing about DAOs — and what makes them superior to the web2 model I discussed above — is they remove the misalignment between the company and the network by eliminating the company altogether and shifting the economic and governance rights to the network. (For tax purposes there will likely need to be legal entities, at least in the US, but these will have a limited, administrative role). A properly constructed DAO does this by memorializing the rules around economics and governance from the very start, using smart contracts on blockchains like Ethereum. As the networks grow, and move up the adoption S-curve, they might be drawn toward consolidating power but they’ll be prevented from doing so by the blockchain-enforced rules. We like to say instead of “don’t be evil” it’s “can’t be evil.” Instead of relying on human nature improving, we are baking the rules into the code.
Of course this all depends on designing these systems correctly, and there will likely be failures and scams and all sorts of other bad things, as there are with all new technologies. But overall I think web3 generally and DAOs more specifically can help course correct the internet back to its original, idealistic vision: power and money pushed to the edges, networks growing and flourishing together, a level playing field for talent anywhere in the world, a thriving creative middle class, and a generally diverse and interesting place. It’s not guaranteed, but let’s run the experiment. We ran the other experiment for 28 years and it didn’t work.
You’re reading a post from my Adjacent Possible newsletter. Sign up to receive more essays like this one in your inbox.
There are two paid subscription tiers to Adjacent Possible: for $5/month or $50/year, you’ll have access to new original essays from me, including a special series on designing a workflow for thinking, and you can participate in conversations with some of the smartest people I know. This fall’s lineup includes the legendary Stewart Brand, Immortal Life of Henrietta Lacks author Rebecca Skloot, Think Again author Adam Grant, New York Times columnist Ezra Klein, and Rationality author Steven Pinker.
There’s also an “Ideal Reader” tier for $120/year: for that you get all the advantages of the paid subscription but I will also send you a signed, personalized copy of each new book I write, starting with my latest book, Extra Life.