Open Letter to the European Commission on Google

To the European Commission,

Google has recently asked you to approve its planned acquisition of DoubleClick.

In support of its request, Google’s management has engaged in a campaign which the media has relayed, I am afraid, without giving it a proper counter-weight. Your experts are not to be swayed by such practice but the public opinion may well be. I feel therefore necessary to respond to Google’s artful presentation of its case. And if my arguments makes your impending decision easier to explain to the very consumers and competitors your duty is to protect, all the better.

Before I proceed, I must admit a conflict of interest. My claim to expertise stems from research and developments I have carried for some years now within a for profit company called ePrio Inc. It is inevitable than my positions further the commercial interests of ePrio. This natural bias however invalidate neither facts nor arguments. Competitors are a normal source of information when it comes to anti-competitive behavior.

I mentioned facts. They must indeed be the foundation of any decision you take. But do I really need to recall here what the media and your own analysis have already made clear aplenty ? Do I need to present Google to the reader ? I would rather draw the attention to the proper framework in which to cast them.

This framework is very simple. It encompasses three concepts, all well known :

 utilities - companies which have achieved dominance in the delivery of some essential goods or service

 software layering - the fact that software architecture resembles multi-layered cakes

 abusive salesmanship - taking advantage beyond a legitimate transaction of customer weaknesses, be they rooted in ignorance, lazyness or fear,

The legitimate goal of any company is to win against competitors and gain the benefits of such dominance. For a software company, it means to turn its offer in the layer on which it does business, e.g. operating systems for personal computers or Internet searches, into a platform whose widespread adoption creates a bottleneck. Whether they operate in the layer immediately below or above, companies must become compatible with this bottleneck or else abandon any hope of being more than a niche player.

Would have Google succeeded without having been accessible from the Microsoft Explorer browser ? Can one imagine today launching a new Internet browser incompatible with the Google search engine ? Can one imagine developping a keyword search organizer to run on top of search engines which would not work with Google ?

Whatever the industry, dominance enables a utility to use monopoly pricing and grab back most of the economy of scale which becomes available to both its customers and suppliers. But the software industry is special in three ways :

 a dominant platform player does not have to support transaction costs with companies in the lower and the upper layer. They already behave as suppliers and customers, without the protection of a contract.

 whenever a player in an adjacent layer becomes dominant, the dominant platform player can try to replicate its functionality and absorb it into its platform, killing the target in the process.

 whenever a dominant platform player fails to absorb a would be dominant player from an adjacent layer, it opens itself to be displaced by this new player and let the bottleneck move to this new player’s layer.

Do not blame Microsoft from having killed Netscape and Real Networks. It had to. Its biggest issue nowadays is its failure to replicate Google’s search engine success.

In this framework, the case of Google is easy to present :

 Google has become the dominant player of the keyword-based search layer, packaged as contextual advertising in a brilliant business model

 the highest growth will be in adjacent personalized Internet advertising

 it is vital for Google to absorb personalized advertising into its platform

How convenient for me to cast personalized and contextual advertising as two adjacent layers, since it implies my conclusion. Why indeed, if not for user profiles.

Personalized advertising is based on user profiles, of which users’ searches is one of the largest potential contributors. But when Google tracks my search patterns over time and buys a company which implants cookies on my computer to spy on me, it burdens its architecture with a module, personal profile management, which have nothing to do with its core business of contextual advertising. DoubleClick on the other hand bought Abacus, a consumer database. Even if it disposed of it after a few years, we are talking here of a strikingly different layer, one which happens to sit on top of search engines.

Having set the stage, we can now proceed and hear Google’s arguments in its defense. They can be cast in three categories. A strike against Google is a strike against :

 innovation

 the consumer

 economic growth

The recent decision against Microsoft in the European Courts does justice to the claim for innovation. What discourages innovation is when entrepreneurs develop a widely acclaimed offer only to see it disappear into the dominant player’s platform. Besides forbidding the dominant player to cross-subsidize any outside business with the profits of their quasi monopoly, much as with traditional utilities, it is enough to ask software utilities to publish clear, open interfaces for innovation to flourish next to a dominant player.

Google has a lot to contribute to personalized advertising. It only needs to make it easy for emerging companies to plug into and aggregate its raw stream of individual searches.

In the case of personalized advertising, claiming better service for the consumer is a stretch. Personalized searches taken in isolation would be such a service. But is it credible, coming from Google ? It turned customer keyword search into contextual advertising and went on to sell advertising based on its users’ mail. Its duty is to its advertisers. If it magnanimously offers to keep user data for eighteen months only, advertisers must have concluded that such data go stale afterwards.

The only serious argument which could prevent the European Commission from restraining Google forays in personalized advertising is that it would hurt the economy. The European Commission can hardly block the future of advertising but neither can it restrain Google to the exclusion of other possible entrants. Such are aspiring dominant social sites Myspace and Facebook, which control another major source of user data and exhibit similarly exuberant appetites.

Hence Google’s campaign. “How can personalized advertising proceed without granting some personal data acquisition right to the industry ? If it can’t, why not agree that 18 months are better than 30 years ? And if so why single me out ? Be fair and reasonable.”

Yet personalized advertising is a very special case. Many companies do not even require a user click to claim user data ownership. Google for one does not ask its users before it archives their searches according to the personal computer they come from. Most of the rest require the user to unclick some consent box, knowing full well that few will make the effort. Aren’t such attempts to aggregate user data as users interact over the Internet cases of abusive salesmanship ? Shouldn’t user profiles be better protected ?

Hence my letter. Do not be fooled by Google’s representations. Ask yourself :

 does the growth of personalized advertising require companies to collect, hold, transfer, resell and otherwise aggregate personal data for any length of time in violation to the right of individuals to privacy ?

 is there any architectural benefit to integrate customer profiles with services which, like search engines, electronic stores, social networks, are natural sources of information on those profiles ?

The answer, I am pleased to report, is negative on both count.

My work has shown how companies can engage into personalized advertising without anyone acquiring access to any confidential data. The key is to set up a mechanism to hold the user profile at arm’s length from any entity involved, i.e. the profile information source, the advertiser, the space seller and last but not least the advertising system operator. The interfaces of this mechanism are quite clean and open. Profile information sources contribute user data, advertisers contribute advertisements, space sellers receive personalized ads and the advertising system operator receive accounting reports to bill advertisers and pay for the usage of both space and user data.

This is not the place to detail the technical solution. The key point is economic. Google, Myspace, Facebook, Amazon, Ebay and others generate user information as a by-product of their core activities. It would destroy value to refuse them to turn this information into a new revenue stream but it would put individual privacy at risk to let them aggregate it to this purpose. What I have shown is that one can implement user data aggregation without creating an aggregator in the process. This role truly defines an independent software layer, adjacent to and based upon, those core activities.

No doubt solutions based on my or others’ inventions will require further investments. No doubt my words should not be accepted on faith alone. But should you compromise and let any company become a user data aggregator, a step as unnecessary as it is dangerous, without giving yourself the time to ascertain the facts ? Don’t the rights of European customers justify both vigilance and due diligence ?

Respectfully yours,

Philippe Coueignoux PhD

President, ePrio Inc.

PS : see here for a short description of confidential, personalized interactions