Relaxing whilst doing Competition Law is not an Oxymoron

Antitrust Clone? Or Why the Google case is Stronger than the Microsoft Case…

with 13 comments


Despite intense media coverage, little has filtered on the content of the EU Google investigation.

That said, the fragments that leaked in the press led me to draw a puzzling analogy: would the ongoing EU Google case be a rerun of the Microsoft tying cases (2004 and 2009)?

Take a look: in Microsoft, the Redmond giant was accused of using its dominant facility – Windows OS for PC – to preferentially distribute related softwares – WMP and IE – and exclude rivals from those later markets. And this strategy could work effectively because consumers are lazy animals often satisfied with default or automatic settings. For various reasons – which included barriers to searching, choosing, and installing a competing software which could stem from a lack of technical skills or simply because of hassle costs – the Commission explained that users only rarely looked for better alternatives. This phenonemon had been labelled “end-users’ inertia“. For more on this, see my paper with my assistant N. Neyrinck.

Now,  the search discrimination allegations levied against Google look strikingly similar. To frame them simply: Google seems accused of using its dominant facility – its well-known search engine – to preferentially display links to related services – Google maps, Google News, Gmail, Google Finance, Youtube, etc. – and exclude rivals from those markets. Interestingly, a key aspect of the theory of harm seems based on the fact that users disproportionately click on the first links displayed by Google’s search engine, and only rarely click on links that rank lower (see chart below). In other words, search users are also lazy creatures who fail to compare the full range of alternatives displayed on the screen. A study found on the web suggests that 94% of users click on a first page result and less than 6% actually click to the second page and selecting a result displayed there.


But, there’s a key, critical difference between the two cases. In so far as the Microsoft case is concerned, offering pre-installed related softwares was a natural business strategy for MSFT. After all, the OS and related sofwares – which came for free BTW – are complements. The pre-installation of such softwares enriched the functionality of the OS, which was in the consumers’ best  interest. This argument, which has been made time and time again ties in with the traditional shoe&laces/car&radio metaphors. But what is more is that in the browser case, the pre-installation was a necessary evil. Absent a readily available Internet browser, consumers indeed could not access the Internet and in turn download competing software. Now, some may counterargue that the problem stemmed from the fact that the pre-installed softwares were Microsoft’s own products. But how could it be any different? Given the number of complementary softwares running on an OS, and the myriad of alternatives for each software, it would be all to weird, and possibly unworkable, to enjoin dominant software companies to pre-install competing products on their own motion: how to select them? in what range? on what terms?

In the Google case, however, the preferential ranking of Google’s related services is NO natural business strategy. Rather, the allegation concocted by the complainants, and possibly endorsed by the Commission, is that Google artificially fiddles with its algorithm to display links to its own related services on top of search results pages. In the “but for” search engine world – i.e. absent algorithm manipulation – Google’s related services would not systematically over-rank competing services (yet consumers would still enjoy links to complementary services).  Contrary to Microsoft, there is here no clear objective, natural justification to what Google is allegedly doing. This, in my opinion, makes the Google case different from, and possibly a tad bit stronger than, the Microsoft case.

PS: With this background, I feel a sense of compassion for Microsoft’s lawyers.  Since 2004 and 2009, they must be in real trouble,  trying to understand how, and to what extent, complementary softwares can be pre-installed on Windows.

Written by Nicolas Petit

13 December 2012 at 11:29 am

13 Responses

Subscribe to comments with RSS.

  1. As I just discussed over the phone with my learned co-blogger, I fundamentally disagree:

    My knowledge of the case is limited to publicly available material, but in my view the Google investigation is completely different -and much weaker- than the Microsoft case. My disagreement is grounded on too many reasons to summarize them in a brief comment, so I´ll develop them in a forthcoming post (so that the Fair Search guys can jump to my neck) 😉

    In fact, even this narrow aspect of consumer psichology (which I see as insufficient to argue that one case is overall stronger than the other) was way more persuasive in Microsoft (off the top of my head: the difficulty associated to clicking is different than that of downloading -particularly at the time of the US Microsoft cases-; accordingly, the fact that 94% of Google users click on the results showed in the first page can be read as an indication of the accuracy/value of those results (btw, the study relates only to Google; if there is a similar study in relation to -let’s say,Bing- one could compare more easily the accuracy of the results offered by the two…). Also, in this case there is no possible tying-related theory of harm; unlike in Microsoft, consumers get these services for free.

    As to what is a “natural business strategy”, well… search is basically about ranking, and ranking something inherently entails the risk of discrimination-related complaints. Unless I’m lost, no conscious discrimination (or “algorith manipulation”) has been proved.

    I can think of a few other arguments, but work calls me

    Alfonso Lamadrid

    13 December 2012 at 1:05 pm

  2. How is the distinction between natural and artificial business strategies different from the competition on the merits/abusive conduct that is the source of most Article 102-related troubles?

    Second-guessing the motivations behind business conduct can go very wrong. This is the first and most important lesson of the Chicago School, which is sometimes forgotten, if not distorted. This post made me think immediately of Hoffmann-La Roche, where the ECJ assumed that exclusive dealing and loyalty rebates are devices implemented by dominant firms to eliminate competition. We all know this is plain wrong, and we all know the consequences this simplistic assumption has had on the evolution of the law.

    Finally: wasn’t the whole case against Microsoft precisely based on the ‘artificial’ advantage enjoyed by affiliated applications offered on top of the OS?


    13 December 2012 at 1:22 pm

  3. Bagnole, on question 1, I’d say this ties in to the concept of “objective justification”.
    The case, and this post, is not about psychological motives, it’s about fiddling tactics, which change the results given by the algorithm.

    Nicolas Petit

    13 December 2012 at 1:28 pm

    • With these references to behavioural economics, I would have thought that it was indeed all about psychology 😉

      More seriously, I do not see how the alleged artificiality of Google’s conduct makes a difference if we establish an abuse in accordance with an effects-based approach (the objective justification would only come into play at a later stage). If a prima facie abuse is established on the basis of the alleged artificiality of Google’s conduct, this would mean returning to Hoffmann-La Roche. And so much for the Guidance!


      13 December 2012 at 2:52 pm

  4. The antitrust talibans would tell you that the effects-based approach is all but compulsory in abuse cases.

    More seriously, here’s my take: under the Guidance paper, there is abuse if there is anticompetitive conduct + this is not outweighed by objective justification. So here’s my equation: in the case of Google, the theory of harm is that there’s artificial manipulation of algorithm leading to anticompetitive effects (this would need to be proved) + no objective justification. This looks fine under the effects based standpoint no?

    In the Microsoft case, in contrast, there was allegedly anticompetitive pre-installation but (+) an objective justification (the Commission + Court failed to see it). Hence I would not have found abuse.

    Put differently, in my opinion, Google should be one step closer to abuse than Microsoft, because it wont be able to avail itself of an objective justification. But this is subject to proof of anticompetitive effects. That said, with the low standard applicable to proof of anticompetitive effects, I suspect the Commission will not find it difficult to nail Google on algorithm manipulation.

    Nicolas Petit

    13 December 2012 at 4:32 pm

    • Since I wrote this post, I was told that its timing was fortunate. The talk of the town is that Google’s representatives are currently at COMP, talking settlement. True/not true?

      Nicolas Petit

      13 December 2012 at 4:33 pm

  5. If it is established that a given line of conduct leads to foreclosure, I do not see how claiming that it is a ‘natural business strategy’ can be validly invoked as an objective justification (and vice versa). Would it be possible to claim: ‘my strategy eliminates my competitors, OK, but it is a natural one to follow’. I doubt it. This goes back to my earlier point: under an effects-based approach, as I understand it, these considerations become irrelevant.

    Great discussion anyway!


    14 December 2012 at 12:33 pm

  6. I’m not sure I see the distinction between “natural” and “artificial” that you are using.

    The business case in both instances is an enhanced user experience. A user who searches for something resembling an address, for example, clearly might be interested in a map result. Providing one makes searching on Google more useful to search users. That seems pretty “natural” to me.

    More on my thoughts, obviously from an American perspective, on Google and preferences here:

    Adam Miller

    14 December 2012 at 10:01 pm

    • To Adam: correct, but the point is that providing Google map rather than rival maps through algorithm manipulation fools consumers and forecloses competitors. Short of algorithm manipulation, users would still be provided with maps. And competitors could compete on equal footing.
      To Bagnole: sorry, but again, I disagree. The objective justification is a critical component of any effects based approach. The Court confirmed this in Post Danmark.
      Also, you take issue with the words ‘objective justification’, but this is a formal pb. What I refer to in the post, is what’s behind, i.e. technological integration of software complements which increases users functionality. This sort of objective justification/efficiency is completely in line with an effects based analysis.
      Finally, I agree is on the great discussion!


      15 December 2012 at 1:46 pm

  7. Of course the objective justification is an integral aspect of the effects-based approach. And desirably so! I never meant to imply the opposite.

    The point I was trying to make is a different one: in an effects-based world, the objective justification needs to be effects-based. Preconceptions about whether a given line of conduct is natural or artificial are irrelevant at best and harmful at worst (Hoffmann-La Roche and Michelin II are there to remind us of the consequences of relying on preconceptions about the nature of business conduct). These claims will have to be established on a case-by-case basis in light of the features of the market.

    Your point seems sensible and certainly plausible, but no more or no less, may I add, than the opposite claim (ie that favouring Google’s affiliated services improves consumer experience). It is at the very least conceivable that giving prominence to Google’s services (eg a map run by google on top of the search list) makes navigation easier, faster and more intuitive (why force consumers to choose between several links to map services if they are all the same and they are familiar with the leading one?).

    I am not saying that one is more theoretically plausible than the other. My point is a different one. Under an effects-based approach, this theoretical discussion matters (or should matter) very little. It will all depend on concrete evidence established on a case-by-case basis, and not on more or less misguided preconceptions about whether we believe that a given line of conduct is ‘natural’, a plausible source of efficiency gains (which is what I understand you say in your post about Microsoft’s tying, with which I fully agree) or (mother of all evils) an expression of ‘competition on the merits’.


    15 December 2012 at 2:25 pm

  8. Pablo, sure that users benefit from links to other services on the search engine, but why should it prominently be links to Google’s own services?

    Let me explain. In Microsoft, the technological integration defense was strong because MSFT has as its primary purpose/economic activity to manufacture softwares, not to distribute rival softwares. With this background, it had every economic reason to pre-install its own complementary softwares on its OS, and not those of rivals, over which it enjoys no control… Arguing the contrary is akin to forcing dominant firms to buy the products of their rivals and resell them. I suppose this is not where you wanna go.

    And from the users’ side, the story is the same: the failure of the remedy in MSFT I proves that users favour pre-installation over non pre-installation.

    So the idea is: a dominant manufacturer that makes several complementary products should be entitled, at least on the merits, to debunk allegations of unlawful tying on the basis of ‘objective justification/efficiency’ arguments.

    In contrast, the primary purpose/economic activity of a search engine is to index and rank as many links as possible, including links to services provided by rivals. In other words, a search engine is a distribution platform, and Google is a gatekeeper on this segment of the value chain. And you will agree that the indexing should thus be done under transparent, non discriminatory conditions. In turn, this means that there is no obvious economic justification to Google’s prominent ranking of its own related services. In my opinion, this makes any ‘objective justification/efficiency’ argument dubious in so far as Google is concerned. Sure they can claim that it is normal to place links to related services on the search engine – it is actually the essence of a search engine – but it does not follow that those links should be necessarily towards Google’s own services.

    So here the idea is: a dominant company that distributes several competing services has from the outset no objective justification to discriminate in favour of its own related services, much to the contrary.

    Nicolas Petit

    18 December 2012 at 10:11 am

  9. I think you overstate the differences between the two cases. There may be some, but of degree, not of principle. An operating system is in essence as a distribution platform. It is not by chance that the remedies in the second Microsoft decision sought to ensure non-discriminatory access to the operating system. If you look at things from this perspective, then the two cases address very similar issues.

    The question in the Google case would be that of whether the ‘distribution advantage’ (to use the expression found in the 2009 Microsoft decision) enjoyed by Google’s affiliated services is abusive. You imply that a dominant firm should not be forced, as a matter of principle, to buy and resell its rivals’ products. I agree. But then it follows from what you say that there is no reason why a dominant firm should be forced, absent exceptional circumstances, to give access to its distribution platform to rivals’ products. Wasn’t the Bronner case about a distribution network, after all?

    If you are not obliged under Article 102 TFEU to distribute your rivals’ products, then you must be entitled to give them less favourable access to your distribution platform, wouldn’t you agree? I know that the judges who decided TeliaSonera would probably not, but I would have understood from previous posts that you do.

    Objective justifications in Google? It is not difficult to think of them. Aren’t search engines there to make it easy for end-users to find what they are looking for? When an end-user types the name of a city, or a (brick and mortar) address, it is a safe guess that the user may be looking for a map. Doesn’t it make business sense, then, to display a map (one, not four or five) on top of the search list to ease the search experience? Saying that ‘it does not follow’ that Google should choose its own services is very much like saying that tying may be efficiency-enhancing, but that it does not follow that the dominant firm should only tie its own products; or saying that vertical integration may be efficiency-enhancing but that it does not follow that the firm should only rely on its downstream division to distribute its products.

    You say: ‘indexing should thus be done under transparent, non discriminatory conditions’. Why? Because the firm has a ‘special responsibility’? Because discriminating is not a form of ‘competition on the merits’? Would you really go as far as to state that dominant firms are (or should be) under a general duty not to discriminate against rivals? The true issue, in my opinion, is whether indexing in opaque and discriminatory terms (assuming this is what Google really does) is abusive. And, as mentioned above, one will have to try very hard to convince me that this question should not be answered in light of Bronner. Assessing the abusive nature of this conduct against a different benchmark would be, in my opinion, as absurd as the ruling in TeliaSonera.


    18 December 2012 at 10:49 pm

  10. A last comment on this, putting my point diffently:
    In MSFT, the but for scenario (absent tying) was users with a naked OS. From a consumer welfare standpoint: tying scenario > but for scenario
    In Google, the but for scenario (absent algorithm manipulation) is popular links on top of Google’s searches regardless of their origin, rather than links to Google’s related services. From a consumer welfare standpoint, the manipulation scenario cannot be said to be better than the but for one.

    Nicolas Petit

    31 January 2013 at 3:13 pm

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: