At the start of this year, Twitter and Facebook banned then-President Donald Trump from their platforms. Google, Apple, and Amazon kicked Parler, a right-wing social media app, off their platforms. 2021 gave clear proof yet that tech companies control online society more than any government, constituency, or constitution. Elizabeth Warren has led a chorus calling for a breakup of big-tech, while the big tech companies themselves race to pour billions into the ‘metaverse’ to establish their dominance on the next major platform.
But on a complex issue more in the spotlight than ever, most commentary is missing a simple framing: internet-based platforms are natural monopolies. If our prescriptions miss this diagnosis and treat the symptoms rather than the underlying cause, we’re doomed to recycle the ailment. If we recognize the situation, we can design the right governance for online networks to keep private and public interests (sufficiently) aligned.
A natural monopoly is an economic term for
a monopoly in an industry in which high infrastructural costs and other barriers to entry relative to the size of the market give the largest supplier in an industry, often the first supplier in a market, an overwhelming advantage over a potential competitor (Wikipedia).
This has historically applied mostly to large-scale physical infrastructure projects with high upfront costs, returns to scale, and network effects: public utilities, railroads, ISPs. But natural monopolies aren't limited to physical infrastructure networks. Setting up an online platform may have much lower upfront costs than a railroad. But network effects online can grow faster and more global - and so achieve dominance in a similar way.
The exponentially climbing value of networks (powered by Reed's law as well as Metcalfe's) means there is a natural tendency towards a winner-take-all natural monopoly in online networks. Winners are protected from competition by their strong, ever-growing network effects.
Specifically, online networks run on both code (which define the services, algorithms, products, etc) and on data - the information that populates the apps and websites we use every day. It's the value of this information - our social graphs, user data, information and interests, browsing history, etc. - that gives large players such a huge advantage over any competition and effectively locks users into their products and services.
In the analysis of big-tech platforms, a common mantra of libertarians, technologists, and home-office-chair economists has been “if you don't like it, go elsewhere." This line of reasoning argues that these are private companies in a free market; if you don’t like how Twitter operates, go use another social network. This stance doesn’t hold in natural monopolies. As with delivering tap water, the natural (economically efficient) number of service providers for any given network online is 1. There is no ‘elsewhere’ to turn to in monopolistic markets because the market conditions ensure there won't be viable competitors.
Our largest tech companies aren't natural monopolies in random vertical markets. The web is our digital society, and these platforms operate as our public square (Twitter and Facebook), our public utilities (Google, Apple), and our public infrastructure (AWS). Anyone excluded from these platforms will be unable to participate on the web. They'll have no voice and little chance to participate, compete and thrive. Right now, big tech companies have the power to disenfranchise any participant or perspective from our digital society. That makes them the government of our digital society. And it is a dictatorship.
Economists and policymakers going back to John Stuart Mill have known that natural monopolies can't be left alone in a free market because they have too much power, no competition, and the ability to extract from the public without checks. Some intervention or public-interest governance is necessary to prevent any network owner from abusing their uncompetitive position.
Throughout most of history, the only realistic way to apply this public-interest governance to private companies was through regulation or privatization by state governments. This nearly always leads to inefficiency, often to capture or corruption, and usually isolation from the valuable advances and innovations that competitive markets provide. Government action, when necessary, comes with a high cost.
Another option is to ‘break up big tech’ to re-establish a competitive market. But in markets driven by strong network effects, the tendency towards monopoly will simply lead to reconsolidation and recycling this pattern with ever-greater inefficiency and confusion. Others argue for forced interoperability, similar to regulating telcos - forcing big tech companies to open access to their underlying networks (databases). This is possible but would require more technical nuance and foresight than most regulation can supply.
We now have a new alternative: distributed ownership and governance of high-value networks -- powered by cryptography, blockchains and public data systems
At the heart of a better model is a simple idea: separate the network layer from the application layer. Today, platforms bundle both their product & service with their database and network. In the future, applications can be proprietary and fragmented, but the data that populate them (social graphs, user accounts, information) can be shared and natively interoperable across them.
A physical-world analogy makes it pretty obvious that this is how things should work.
Today's "Web2" is held back by a few companies' control over these critically valuable assets. Another example: ‘if building houses were like building software, you'd tell your contractor you want the wall blue, then they'd tell you they’ll check whether there's enough demand to paint every homes’ walls blue.’ We should be able to have many versions of software, modified at the app layer with our own needs and preferences. But because all data is gatekept in a rigid, siloed, company-controlled way, we have very little flexibility in our online experiences.
We've tolerated the platform-owned model until now because it's been the only viable option. The data, social connections, and user accounts have to be stored somewhere. Most users can't do it themselves. So companies ran servers and built databases to do this. Then big companies started offering their infrastructure, software tools, and some starter data to smaller ones (e.g, FB Connect). In return, they asked for access to the small company's data. Suddenly a few big companies had all the data about everything happening across the web. And all the network effects that come with that.
New technologies have made an alternative viable: we can store the data on shared, open, distributed networks. We've had some of the core peer-to-peer storage building blocks for a while. But with many authors and participants writing and reading data to the same data network, challenging problems emerge. Among them: how can you trust data on the network and know that it hasn't been tampered with? New technologies are addressing this:
Cryptographic keys (and IDs) let anyone sign, transact, and participate in these networks, enabling direct management of information (I control my data directly) and the ability for network participants to trust (because they can audit) the network
Distributed, content-addressed storageoffers reliable storage and retrieval based on the content rather than its physical location (URL)
Tokens, or crypto assets, let network designers build incentives directly into the network, helping ensure it stays secure and aligned to the interests of the participants while rewarding contributors to the network (and its value) economically
These technologies enable a "Web3" that is far more interactive and fluid than today's web because many apps can operate on the same core data. Instead of a company having to choose Notion or Evernote for their productivity suite, each person can choose their own tools and still collaborate on the documents and data. You could use Twitter, I could use Chirper - because I prefer their feed algorithm and moderation policies - but we could still like, RT, and comment on the same messages threads.
When the common layer is the protocol, not the platform, innovation isn’t throttled by a single corporate gatekeeper that decides what can happen on the platform or not.
This model's magic is that it gives us more diversity and more network effects at the same time. The app layer can diverge — many different filters, views, interfaces, and services — while the underlying data and network for all are shared. Network effects grow faster than ever. Apps and services work together seamlessly because they don't require complex integrations - they are operating on the same data layer directly.
In the presence of dominant shared network effects, incentives change too. "Composing" on existing products is far easier than ever because of the shared data - whether assets on a blockchain like Ethereum, or other data on a network like Ceramic - so entrepreneurs can create useful products and services far faster. And for users, switching costs between products are fall dramatically because data comes with them.
This means there will be more economic value in creating useful products and services than trying to build a huge, proprietary user base for its network effects. The quality of the products and services we use will take off. Networks will be far more powerful but now shared by all. Apps will be individually far less powerful than today’s platforms, but collectively far more powerful.
In Web3 we won't wait for "10x better” products — we'll have constant marginal improvements that compound way beyond 10x. With public networks, the web will be competitive again.