Monday, 20 April 2015

In Google We Trust

The EU Commission has decided to take on Google for abusing its dominance in Internet search. I fear this can only end in tears, and not just because the commissioner, Margrethe Vestager, was supposedly the inspiration for the leading character in Borgen, Nick Clegg's favourite wet-dream (I mean the TV series about coalition government, not the specific character, obviously). Her roots as a "radical centrist" in Denmark's neoliberal Social Liberal Party (roughly equivalent to the LibDem Orange Bookers) should alert you to the naivety of her understanding of how big business actually works, though you'll be less surprised that she continues the EU's track-record of being pretty clueless about what Google and the other US tech-titans are actually up to.

John Naughton notes that "Google’s current dominance in search is a product of three ingredients: astonishing computing horsepower, distributed globally in huge server farms; smart algorithms; and the possession of colossal amounts of data that can be mined for machine-learning and generate further refinements in Google search. And it’s the combination of these three things that matters". In fact, the basis of its dominance is just the last of these. To coin a phrase: it's the data, stupid. The server farms are merely the necessary infrastructure required to store the data, while Google's key algorithms for at least a decade have been those that parse and index (i.e. organise and interpret) the data, not those that rank it for search results. A large part of my contempt for search engine marketing (SEM) is that the snake-oil distracts from this simple truth.

Google's ability to gets its nose out in front of its search competitors in the late-90s owed everything to the quality of its search ranking algorithms, but it has not depended on them since the early-00s, despite the mythos that has been built around Page Rank (surely more laughable now than KFC's "secret recipe"). It is worth remembering that Google's ascent to dominance owed more to clever positioning than superior technology. It bested Yahoo because the latter was still dependent on a human-curated directory that could not keep up with the explosion of the Web. In contrast, AltaVista's more advanced technology - notably its crawler and natural language queries - proved popular with techies but ultimately turned off the growing audience of "civilians" who couldn't be arsed with Boolean operators. Google quietly adopted the techie smarts (and uncluttered homepage) pioneered by AltaVista, and has since deliberately modelled its geek proposition on the "Easter eggs" of video-game culture - i.e. covert fun for the cognoscenti that doesn't frighten the civvies.

Google relied on synopses, rather than indexing everything on a page, and used links as a proxy for popularity and thus reliability (which they rebranded as "relevance"). In other words, it didn't have the most advanced technology, it simply provided the best user experience: up-to-date, rationally ranked, and with the appearance of objectivity. Google recognised that the quality of search results depended on the relationships between data (links between pages), which meant that it instinctively appreciated the value of the user's own data - i.e. search history and other inferred preferences - as a means of refining search results. Having achieved dominance in search, it was able to translate that into dominance in user preference data (your opinions), which reinforced its dominance in search and thus advertising: a positive feedback loop. The subsequent growth of social media has created a new dataset (your mates' opinions, and your opinion of them) that Google has not been able to similarly dominate, but its control of Android means it can access most of the expanding dataset of mobile user preferences, and social media is now largely mobile (hence "mobilegeddon").

The case being pursued by the EU against Google is focused on the self-interested abuse of rankings, specifically "by systematically favouring its own comparison shopping product in its general search results pages". In other words, they are disputing the independence and accuracy of a species of advertising. For that reason, I suspect their chances of success are no better than those of the ultimately fruitless antitrust investigation by the US Federal Trade Commission between 2010 and 2012. Given the higher political profile of the issue in Europe, centred on the desire to make US tech-companies subject to EU regulation, this may result in some face-saving agreement over cosmetic changes. The parallel investigation into Android stands a better chance of proving anti-competitive practice through OEM lock-ins for search and map apps, much as Microsoft proved vulnerable over the bundling of Internet Explorer, though that precedent suggests the outcome will simply be the opportunity to choose apps at setup.

Historically, the state has routinely promoted monopolies, both because of the revenue potential (e.g. issuing licences in return for up-front cash) and because some large-scale ventures (such as overseas trade) required guaranteed incomes in order to raise the substantial capital needed for success. This still happens today, e.g. the issuing of rail operator franchises and the sale of mobile spectrum bandwith. In the seventeenth and eighteenth centuries, opposition to monopolies usually reflected political factions - i.e. a disagreement over which elites should profit from the commercial powers controlled by the state - hence the focus of attack was often the "old corruption" of the monopolists and their political agents (most famously the impeachment of Warren Hastings in respect of the East India Company).

The industrial revolution and the ideology of laisser-faire marked a political turn against state-sanctioned monopolies, but not against cartels and other "conspiracies against the public" by politically well-connected suppliers (an attitude that lives on in Westminster's dealings with the City). The growing political intolerance of such abuses in the late nineteenth century reflected the parallel growth of national mass-consumer markets (and the growth of large, vertically-integrated businesses, or "corporate trusts", this gave rise to) and the advance of democracy. It was the national dimension (the product of improved transportation and distribution as much as an expanded franchise) that made the issue salient, but the maturity of democracy that determined its political progress (the UK didn't get round to creating a monopolies commission until 1949).

Of course, "democracy" in this historical context really meant the interests of "small businessmen" rather than the mass of the population, though popular campaigns in the late nineteenth and early twentieth centuries would often focus on the assumed consequence of price-gouging. The grand-daddy of antitrust legislation, the US Sherman Act of 1890, was aimed at anti-competitive practises by big business that forced smaller producers and service providers out of the market. The most famous consequence was the breakup of Standard Oil in 1911, though that proved so controversial that large-scale industrial breakups went out of fashion in the US after WW1, a tendency reinforced by the merger-mania of the 1920s and the corporatism of the 1930s and 40s (the later dismantling of AT&T in 1982 was directed at its government monopoly, and was thus a form of privatisation). By the beginning of the twentieth century, the focus of antitrust legislation had shifted from restraint of trade to consumer welfare. This was the logical corollary of social legislation, such as the concern with public health and (in the US) prohibition, with the state taking on the role of protector and arbiter.

In the middle of the twentieth century, the antitrust focus shifted towards achieving "allocative efficiency", which reflected the greater role of the state in economic planning (e.g. mandating specific monopolies or de facto cartels through price-setting, as much in the US as in Europe and Japan). This produced a reaction in the 1970s, as part of the neoliberal turn led by the Chicago School, based around a narrow reinterpretation of consumer welfare. By focusing on consumer effects, neoliberals argued that business intentions were irrelevant. A leading light in this shift was the conservative American jurist Robert Bork, who was famously rejected by the Senate as a Reagan nominee for the Supreme Court. John Naughton refers to Bork's book, The Antitrust Paradox, which suggested that commercial monopolies could be in the public interest if they advanced consumer welfare. What Naughton doesn't mention (though I'm sure he is aware) is that one of the last things Bork published before his death in 2012 was a co-authored (and Google-sponsored) paper specifically defending Google at the time of the FTC investigation.

The assumption behind the EU case is that Google is providing a service (search results) and that its market dominance allows it to abuse its position by favouring its own secondary services (Google Shopping) in those results. A parallel would be brewing companies that own tied houses and oblige landlords to serve specific beers. But there are two problems with this way of thinking. First, all price comparison sites mediate actual product sellers. Foundem (one of the complainants to the EU) cannot guarantee that it delivers lower prices without proving either that it enjoys exclusive discounts from product sellers (or is perversely sacrificing its own margins) or that Google Shopping inflates prices. Neither looks likely, primarily because Google Shopping isn't in the margin-maximisation business: what it wants is user data. On a narrow Borkian interpretation, there looks to be little evidence to prove that consumers are disadvantaged by being pointed towards Google Shopping rather than another price comparison service.

The second problem is that preferential treatment by a communications medium is hardly unusual. A better parallel than a tied pub would be the historic tendency of Rupert Murdoch's newspapers to not only carry adverts for Sky but to plant friendly stories masquerading as editorial (Private Eye used to have a regular feature on these, until they realised it wasn't going to stop), much as Sky News in turn carries complimentary (and complementary) stories that refer to News UK publications. While it might be uncontroversial for the government to insist that tied houses be allowed to sell guest-beers, there seems to be little expectation that The Sun should be obliged to publish puff-pieces about Channel 4 to balance out the puff-pieces about Sky, even in these supposedly radical post-Leveson times.

What this highlights is that the EU Commission is thinking of Google more as a public utility, or a public corporation like the BBC, rather than a commercial service. This might not be wholly irrational, but politically it is a dead end. The EU cannot insist on a charter to govern behaviour and an independent governance structure to monitor compliance, not just because Google is a global business that predominantly operates outside the EU, but because the current political wind is not towards greater collective action in Europe, as both Greece and migrants adrift in the Medierranean can attest. Google is going to remain a private US corporation whose sole raison d'etre is profit, and its advertising revenues will continue to derive from its unique dataset of rich user preferences that no competitor is likely to be in a position to rival for years to come. If Facebook couldn't knock Google off its perch, it's unlikely that a British price comparison website is going to manage it.

If the EU insists that Google does not favour its own service in its search results the company would be perfectly entitled to address the issue by deprioritising all price comparison sites in favour of actual product sellers. The end result, in terms of harvesting user preference data, would be reduced but still acceptable from Google's perspective (because better than the competition), and they might even be able to justify higher advertising fees for the sellers who would divert some of their marketing spend from price comparison sites. As far as Android is concerned, obliging users to choose search and map apps is unlikely to alter the landscape given Google's dominance in these areas (Bing is no Firefox). Frankly, game, set and match.

No comments:

Post a Comment