Economics of platforms: Implications for cyberinfrastructure

I recommend an interesting paper by Glen Weyl and Alexander White, “Let the Best ‘One’ Win: Policy Lessons from the New Economics of Platforms.”  The abstract summarizes the message:

The primary policy problem in platform markets is usually considered to be excessive lock-in to a potentially inefficient dominant platform. We argue that, once one accounts for sophisticated platform pricing strategies, such concerns are overblown. Instead the greater market failure is excessive fragmentation and insufficient participation. These problems, in turn, call for a very different policy response: aiding winners in taking all, ensuring they and not their copycats profit from success, subsidizing adoption and regulating the resulting “One” dominant firm.

Their analysis resonated with my own experiences with the development of “platforms” of various sorts within the research environment. Don’t we all too often suffer from excessive fragmentation across different “platforms” (think, for example, of our many competing supercomputer centers, workflow systems, identity management systems, etc.), and often bemoan funding policies that seem to reward rather than discourage fragmentation?

The following extracts from their Section 4, “Policy Implications,” expand upon their proposed policy responses, and provide thoughtful discussion of opportunities and challenges. Read these extracts from the perspective of a research infrastructure that you care about. I think it’ll bring you fresh ideas. See the original paper for the full text and references.

Aid winners in taking all without “picking winners”

The most basic implication of our view that over-fragmentation is a leading problem is that public policies should try to aid eventual efficient winners of platform competition in consolidating their dominant position as quickly as possible subject only to the constraint of allowing sufficient “market deliberation” to sort out which platform is in fact best. Such efforts could take various forms and would clearly require careful consideration.

Help the “One” appropriate gains from innovation

[I]t is also important to ensure that a winner can appropriate the social gain associated with creating a (superior to prior art) platform. … [A]bsent such appropriation, superior potential platforms may, like Esperanto, languish in limbo.

Appropriation has two related components: financing initial subsidization and profiting from eventual adoption. Absent more careful modeling of the distortions that may cause failures of financing, it is hard to clearly predict an appropriate policy response as we know that information frictions may lead to second-best optimal, insufficient or even excessive provision of financing for entrepreneurial projects like creating a new platform.

The impacts of imperfect appropriability are clearer: aiding the appropriation of benefits clearly makes necessary ex-ante subsidization easier to finance and the entry of new, efficient technologies more attractive.

Limit profits from fragmenting copycats

If policy should lead towards aiding winners in appropriating maximum gains from their technological improvements, it should lean against aiding low-value, fragmentation- inducing copy-cat entrants from being able to appropriate rents from entry.

Subsidize participation and development

Perhaps the clearest policy prescription, the one achieving the most desired goals with the least conflict, is also the one least frequently applied: direct government subsidies for the development and especially the use of the services of platforms. While development subsidies may help to some extent with raising capital and are generally useful in addressing problems with appropriability, direct subsidies on adoption are likely to achieve more socially desirable ends at once. They can help alleviate market power and often also Spence distortions without risking the fragmentation competition can cause and they increase appropriability. Furthermore so long as such subsidies are not too large and complement, rather than replace, market pricing mechanisms, they do not excessively undermine the information supplied by the market about which platforms are best to adopt. Of course such subsidies may be open to political capture and thus must be approached with caution, as anywhere.

A potential way to design them to also help aid the winner-take-all dynamic and avoid fragmentation, while simultaneously exploiting market signals and potentially avoiding corruption, is to base them on user behavior. Per-person subsidies for adoption could be increasing in the number of other users on the platform according to some pre-determined and more-or-less impartial formula.

Consider natural monopoly regulation

All of these recommendations will strike readers as basically pro-monopolistic, and, in an important sense, that is precisely our argument. Even more than in traditional industries with economies of scale, markets where incompatible platforms compete are natural monopolies. As such, they naturally call for a bouquet of regulation. Given the importance of ensuring the appropriability of benefits from acquiring such a monopoly position, however, it is important that these regulations be designed so as to ensure firms internalize the largest external costs of their actions in ways that are minimally costly to the firms.

Some areas where the interests of platforms and the public may diverge and where regulation can secure public interests at relatively low cost to the platform may be openness and non-discrimination across content. Others might include transparency about some aspects of platform design, such as the ways in which private data are used, analogous to those that the public demands of governments.

The design of a new regulatory infrastructure for platforms … involves a host of trade-offs that require much more research. However, we believe that thinking about platforms in a natural monopoly framework, rather than only in a standard competition policy one, is likely to be a particularly fruitful path going forward.


One comment on “Economics of platforms: Implications for cyberinfrastructure
  1. Daniel S. Katz says:

    You might also want to read Thiel’s “Zero to One”; particularly his discussion on capitalism vs competition in Chapter 4, “The Ideology of Competition.”

Leave a Reply

Your email address will not be published. Required fields are marked *

*