Competition through interoperability

A jigsaw puzzle, where different shaped pieces just fit together

In a recent post, I identified competition as one of the major public policy fights that will shape the future of internet and tech policy (it’s the ‘C’ in my ‘CATS’). But the possible scope and strategy of a future public policy fight over competition take more time and depth to explain.

The story starts in Europe. European regulators are a step ahead of their American counterparts in going after the next evolution of competitive harm in the technology sector: leveraging market power from a platform into a separate, but technically and economically integrated, sector.

European competition authorities recently levied a substantial fine on Google, holding illegal the company’s inclusion of a Google Shopping box (containing specific product results) above organic search results that include competing comparison-shopping services. The European Commission has two other open proceedings concerning Google, for the company’s mobile Android operating system and its AdSense advertising business; from what I’ve read, the Commission seems poised to rule against the company on these as well.

But the story is about far more than Google or Europe. In many ways, the EU’s investigation of Google is a bad example to illustrate the broader dynamic of vertical leveraging, or to help imagine what a better future would be. A better one, for me, is Facebook — specifically, Facebook’s merger with WhatsApp. This was a major, high profile acquisition by a dominant platform company of a rising star in the messaging sector. The business case for the merger made sense, even at the astronomical price; WhatsApp had a revolutionary approach to adding new users, and had built a huge user base in markets where Facebook had room yet to grow. But the potential for harm was just as clear. For one, users who signed up for one service but not the other may not have wanted to have their information shared — same for users with accounts in both services who wanted to keep them separate.

It’s hard to balance when you don’t know how to measure

To be blunt: Antitrust authorities, whether in Europe, the U.S., or elsewhere, don’t have the right tools today to handle a merger like that. Their focus is generally more on horizontal mergers and the impact of mergers on competition within a market segment, something that seems like a non-factor when looking at the two different technology sectors of social networking and messaging. Concepts of potential competition could perhaps be developed, over time, within the scope of existing law and policy — in this case, looking at WhatsApp’s potential as a future horizontal competitor to Facebook. But the differences exhibited in the ever-shifting world of technology, including but not limited to network effects (as Ben Thompson writes at more length in his analysis of the EU’s Google Shopping decision, and as clearly displayed by WhatsApp’s value to Facebook), make productive use of the current antitrust toolkit an impossible stretch in any near term action.

The worry I have isn’t clearly described by traditional metrics either. I’m worried about future innovation in a subset of the vertical stack being obstructed by the practical necessity of functional integration with the central platform(s) anchoring that stack. I’m worried that new innovative services in non-platform markets will be squashed by inferior competitors who get special technical treatment by one or more platform operators, conferring quasi-permanent advantages through tighter technical integration.

Hal Singer put it clearly in his recent Forbes piece: innovation harms are a “blind spot” in antitrust. As he describes it, in theory, the tools and remedies are there; but in practice, the metrics to measure harm are misaligned, and the mechanics to impose change are slow and poorly targeted. So, capturing the full range of harms to future innovation from a Facebook/WhatsApp merger is infeasible; and the stronger remedies on the table, like breaking up Google to address concerns over its integration of search and other services, would in all likelihood do more harm than good.

Without the tools to measure or address the future competition harms of a proposed combination like Facebook and WhatsApp, much of the discussion about this particular merger wasn’t over whether Facebook’s ties with WhatsApp could later jeopardize an innovative competitor to WhatsApp or a competitor to Facebook, or empower the combined entity to stifle other future innovation. It also didn’t capture the full potential of horizontal competition between these two services, a significant issue against the backdrop of fluid and fuzzy distinctions between layers of technology. Instead, concerns centered on privacy issues, particularly whether the two services would share data, signing up the users of each for terms and exposure they hadn’t agreed to when they provided their data to the other service.

More and more, all of these are integrated and sharing data, under the hood; is the result still “open”?

These are valid and significant concerns. But they weren’t enough to stop a merger viewed at the time as pro-competitive. I clearly remember my reaction when the merger moved forward with a condition that the businesses not share data. I knew there was no way that condition would stick — and I wasn’t alone. Combining users and data has too much business value, and the dull teeth generally associated with enforcement can’t compete with that. But public unease with permitting such a merger seemed to have been assuaged by the condition, even though it did indeed fall through as predicted.

A better evaluation of the potential for competitive harm could plausibly have led antitrust authorities to deny the merger outright. But let’s consider a thought experiment, assuming the merger were going to be permitted: What if it had, somehow, been conditioned on the services being integrated in an open way that encouraged transparency and interoperability? For example, as a merger condition, the mechanisms used to exchange data and communicate between Facebook and WhatsApp could have been required to be implemented through transparent, third party accessible application programming interfaces (APIs).

The story often told in mergers like this is that the companies will remain independently operated — they have separate teams working on the engineering, they store user data separately, etc. That’s rarely a permanent vision, because the benefits of integration are too great. If you assume there will be integration between Facebook and WhatsApp over time, including tighter connectivity between the services, sharing various forms of data, and so forth: the question then is whether competitors to WhatsApp have even the possibility of negotiating for the same kinds of access to Facebook that WhatsApp has (and Facebook competitors negotiating for comparable access to WhatsApp). This possibility cannot be taken for granted; the integration can be done in a way that makes those future negotiations impossible, such that undoing it it would require re-architecting and re-engineering the systems in a fundamental, expensive, and uncertain way (see: Microsoft and the European Commission).

But, if Facebook and WhatsApp were required to engineer future integration in a suitably arms-length way, then the possibility of future competition would be preserved. Other partners would be able (at least in theory) to negotiate for the same kinds of benefits, including data sharing, preferential positioning or embedding into user interfaces, etc. And there could be other, non-competition potential benefits as well. Researchers could, in theory, tap into the same interfaces. Maybe privacy advocates could even use them to identify, test, and document just how much information exchange and leakage is happening in practice.

Not a literal depiction of accessing an Application Programming Interface

The combined entity would still be able to realize the positive business benefits of integration — just not to the exclusion of others. And antitrust authorities could be brought in later should the platform operator be unreasonable and anticompetitive in how it offers access to its APIs and underlying data — perhaps under a model like the nondiscrimination conditions Hal Singer suggests in his article.

Imagine if this were a standard expectation for technology mergers where the combination is not merely corporate but also technical, resulting in the potential for exclusive communicating or sharing of data across services. What if Facebook and Instagram had the same level of open integration? Or Google Search and Google Maps? Mergers like these are where the circumstances for future anticompetitive leveraging are created.

The advantage of imposing interoperability as a merger condition is that it has the potential for lasting effect even if the underlying condition comes with an expiration date. The mechanisms for access offered to competitors can be phased out over time and replaced with proprietary successors. But that represents a cost and a change in direction, one that might exceed the benefits associated with it. Furthermore, because of network effects among other factors, it’s highly possible that connecting with now-established business partners would prove beneficial for the company even when not required. And, if those partners are jettisoned, benefits associated with connecting to them could flow to another platform level competitor.

I’m not a competition legal expert, and so I’m not going to claim this could be done today. As we’ve seen many times in both Europe and the U.S., most recently with Intel’s partial victory on appeal of its 2009 fine, antitrust authorities are under specific legal constraints as to the remedies they impose and the analysis they use to justify them. But, perhaps there are some creative approaches to get there with the laws we have today — a question for consideration, at least.

Even still, there are things we could definitely do today to establish this as a norm. Interoperability could, and should, become a conscious point of emphasis for antitrust authorities in other ways — such as future Sherman Act Section 2 guidance in the United States, or future legislative proposals in the European Union. Build interoperability and openness into your systems, and we’ll look favorably on how you use them, because we will consider you to be saying, “No I’m not being exclusive; I’m here to negotiate with competitors so all can benefit.”

A good sign of a successful negotiation

Soft power pressure like this can lead to many good outcomes, without getting into protracted multi-year enforcement proceedings (during which competing businesses wither and die) — provided the tools exist to back the pressure up, and regulators bring enough cases to convince businesses that there’s a bite behind the bark.

Ben Thompson laid out the unique features of competition in the tech sector, and the potential benefits of interoperability and API disclosure against that backdrop, far more ably than I have done, in his 2016 post “Antitrust and aggregation.” He, among others, predicts more and more centralization in this sector. In parallel, more and more startups and policymakers are identifying this as an evolution away from the open, innovative, disruptive internet we’ve had for decades.

This brings us all the way back to the beginning: the European Union and its Google Shopping case, and potential other cases. I believe the EU is signaling that it wants competitors to Google in non-search markets to have the same potential ability to benefit from Google’s search platform as Google’s non-search services do. This is a generalizable principle to other platforms run by major companies that have dominant market shares. If competitors lose the technical possibility of integrating with these platforms, realizing the EU’s vision is at worst impossible and at best Herculean (as with Windows and Server in the earlier cited Microsoft vs Commission example).

The time has come for a shift in antitrust and tech. We’re currently bereft of adequate tools, remedies, and even fundamental economic concepts and standards to understand the harm that’s arising, much less address it. But we won’t be in a position to develop, study, or apply those economic concepts and standards if we don’t work harder to protect technical interoperability at the intersection of these services. Without that, the technologies themselves will be built in a way that forecloses effective technical intersections and competition. Antitrust remedies to force it then will become a much higher lift — a retroactive demand to re-engineer the integration to impose openness. Proactive intervention to protect technical openness is an investment we can and must make now if we’re going to have any realistic hope for a competitive future in the technology sector.

Like what you read? Give Chris Riley a round of applause.

From a quick cheer to a standing ovation, clap to show how much you enjoyed this story.