Breaking up Google ain’t so hard to do

Given the multiple competition cases launched against Google by regulators in the last 10 days (in the US and Italy), and just over a month to go until the European Commission publishes its proposal for an EU Digital Markets Act, what might an ideal outcome look like for consumers in the search engine market?

Politico’s sub-editors came up with the catchy Why breaking up (Google) is so hard to do headline for a detailed article explaining why — actually — the converse is true. Unlike the telecommunications firm cabling and other hardware infrastructure the US and EU spent the 1980s/90s privatising, regulating and/or breaking up, Google’s software and data could be trivially “cloned” (in economist argot). The cost of servers to run it is not trivial — but then Google and its cloud services competitors have done an excellent job in reducing the capital expenditure required to run large-scale online services.

Two remaining elements of Google’s customer-facing search business have high fixed costs: indexing the entire Web, and the customer acquisition needed to benefit from the learning that comes from hundreds of millions of queries per minute. Google and Microsoft are the only companies that do this ongoing indexing (“web crawling”), which they then license on particular terms to many competitors, including my personal favourite, the privacy-focused DuckDuckGo. (It appears Apple has recently revved up its Applebot crawler to increase Apple’s own Spotlight search capabilities on its devices.)

So, an obvious starting point for competition regulators would be to require the enormously dominant Google to license its index and click-and-query data on fair, reasonable and non-discriminatory (FRAND) basis to smaller competitors, and for the Google/Apple smartphone OS duopoly to introduce fair choice screens for users to select default search engines. This is close to what the UK’s Competition and Markets Authority is demanding, while the US is initially demanding Google stop imposing exclusionary terms in its distribution contracts (potentially with structural remedies to follow.)

The next step would be to require Google’s own use of the index and clickstream to be on the same basis as its competitors. More ambitious and interventionist regulators might look for internal functional separation within Google to enforce this. And the next step would be to demand structural separation of the “natural monopoly” business unit into a separate, heavily regulated, utility, similar to the structure of many gas, electricity and water industries in Europe. Then, multiple competitors — including perhaps several mini-Googles forcibly separated from its parent — would have the best chance to compete on the merits of their products.

This kind of detailed analysis would be needed for other Google businesses where its market power remained an issue. But I think it would be a mistake simply to split Google into search, Android, cloud, adtech, mapping and other lines of business units — many of which would remain monopolies in their own antitrust markets, also tempted to cooperate/collude (implicitly via algorithm, or explicitly, via contract (legally) or backroom deals (illegally)). We already see Google and Apple “co-opetating” this way given the $8-12 BILLION paid by the former to the latter for setting Google as the default search engine across Apple’s device operating systems and browsers.

The same holds for the other members of GAFAM (with Amazon perhaps the most difficult case, which the House of Representatives antitrust subcommittee spent much time analysing in its recent report) — and with the US attempts, which ultimately failed, to split Microsoft into a series of line-of-business “Baby Bills” in the 1990s (as the US succeeded in doing to AT&T and its “Baby Bells” in the 1980s).