33The Entrepreneurial State and the Platform Economy

Moreover, it is important to understand that even if antitrust authorities were empowered to

break up companies such as Google and Facebook, that would not eliminate the data

extraction and monetisation that lie at the heart of their business models.

Creating competition among a bunch of mini-Facebooks would not weed out such practices,

and may even entrench them further as companies race to the bottom to extract the most

value for their paying customers. The very nature of platform competition is that it works best when the platforms are big—buyers want to go where there are lots of sellers, and sellers want to go where there are lots of buyers—so the most competitive and innovative marketplace will inevitably be a large platform, which under competition is going to be a private or public firm. If this is nationalized, then you lose the benefits of incentives to innovate in the platform and create contestability. So, the most competitive and consumer welfare maximizing market structure will tend toward a large monopolistic platform. Hipster antitrust policy applied to this context in an unsophisticated form will harm social welfare, even if it is politically popular due to the opportunities for populist big-tech–bashing and multinational corporation tax shakedowns.

Mazzucato is wrong about industrial R&D applied to digital platform innovation, but she is not wrong to recognize that antitrust is not the answer. However, Kahn is wrong about the social welfare implications of antitrust directed at big firms, due to the fact that these policy reform targets will inescapably target platforms. Breaking up platforms may well be good retail (i.e., populist) politics, but it will definitely harm innovation and consumer welfare because it does not ameliorate the need for platforms; it just leaves them less efficient and less effective.

We are currently in an era undergoing a deep historical transition from an industrial economy to a digital economy. This is likely to be as profound and disruptive as the transition from the feudal to the industrial economy that occurred several centuries ago (although more recently in some economies). The transition from an industrial to a digital economy has many manifestations, including the rise of digital and computer capital and of intangible value added; the growth of software (“software eats the world,” wrote Marc Andreessen (2011)); servicization (X-as-aservice); integration (e.g., dev-ops) and full-stack design; shifts in the value of particular skills and types of jobs; the growth of data as a resource; and the emergence of new cross-cutting layers in the economy such as infosec, cybersecurity, and identity. But the other major shift that the transition from an industrial to a digital economy brings is a shift from the comparative efficiency of administrative hierarchy (whether corporate or government) toward the comparative efficiency of platforms (or protocols), made possible by the ability of digital platforms (and protocols) to automate many administrative functions into software-embedded rules. The most advanced form of this evolution is currently the Web3 environments of blockchain (Berg et al., 2019), with protocol money (cryptocurrencies), protocol contracts (smart contracts), which then enable decentralized markets (DEXes), decentralized finance (so-called DeFi), digital assets (tokens, including so-called NFTs), and decentralized autonomous organizations (so-called DAOs) as a network of S. Davidson and J. Potts

34distributed protocol infrastructure. Some of these networks, platforms, and protocols can be extremely large. The Bitcoin and Ethereum blockchains, for instance, are at the scale of small countries in terms of value, or number of users, which is also true of other large digital platform companies such as Facebook and Amazon, which are familiar corporate entities and are both companies. But they are also platforms, in effect marketplaces that provide infrastructure for a local and specific economy, matching buyers and sellers, providing rules and governance services, managing identity, providing security, and even experimenting with offering their own money (e.g., Facebook is currently experimenting with a private money called Diem, developed from an earlier experiment called Libra). While it is technically correct to call these large companies, they are also, to the extent that they provide much of their own economic infrastructure, a type of small economy. The layer 1 protocols of Bitcoin and Ethereum, as with the next generation of layer 1 blockchains such as Cosmos and Agoric, are arguably better understood as startup digital economies, rather than as firms per se. Indeed, they will often form a company or foundation to