Too Big to Search

The "search giant" Google is undergoing significant scrutiny now in Europe where their 90% (according StatCounter) market share of search activity has made them the target of antitrust prosecution. This action is driven by economic concerns, that this level of market hegemony is bad for the consumer and for business development, and is aligned with the usual thinking that competition benefits the consumer and business environment. However, there is another activity in which search is also central and the effect of the dominance of Google is at least as far-reaching and potentially - this is in the marketplace of ideas. Increasingly, what we know, or what we actually use, is limited to what is online. And the way in which we access that information through search engines and in its monopolization is subject to the dangers at least as great as those worried about in the marketplace of commerce.

It is important to realize that search is not agnostic or for that matter entirely objective. Queries are guided at the first level by proprietary algorithms for autocompletion. The search engine working on this query is itself a secret algorithm influenced by a host of factors ranging from your location and previous search history to the various cookies on your machine. Using a proprietarily annotated or "indexed" snapshot of the web the engine then returns a ranked (from best to worst) list of suggested "answers". The algorithms can take advantage of all kinds of information. Business considerations can come into play, possibly influencing the promotion of certain links over others. This effectively turns the information superhighway into the information supermarket, where organization is influenced by competition for product placement.

Even the framing of a search return as an ordered list is a choice, reflecting a kind of path dependence and contingency like the QWERTY keyboard. When it comes to finding stuff out, is that really a process of search or one of exploration where a map might serve better than a list? Does every query have a single most relevant response, or are there different kinds of relevance that exist in relation to one another?

More broadly, are epistemology and advertising aligned when it comes to search optimization? The tension between the two was made clear in the problems and "solutions" to those problems that arose in the well-publicized censorship (mis-) adventures in China of the major search companies. More subtle is that a reliance on customer behavior as the metric for search utility seems to serve business concerns better than information discovery. The lack of an objective ground truth for search as well as the opaque nature of the proprietary algorithms - necessarily kept secret as the lynchpin of the overarching business - poses interesting problems for external evaluation and validation. The size of the unknown unknown depends on this technology "secret sauce."

It is possible to at least imagine that perhaps a government agency or NGO might do its own regular indexing of the web and then put over it, its own search engine, completely open source, with transparent parameters. Developers could work with that framework to create an open source search alternative wherein the implicit biases and assumptions that must pervade any such tool would at least be available for all to see, to better understand and influence the kind of information that they were guided to. Imagine if you could set a collection of parameters that would enable you to search like someone else. You might not be able to walk in another's shoes, but you could at least travel their digital trail. Such a service would be akin to the USGS and it's creation of the mapmaking groundtruth of the physical world. Interestingly, Google made available such their "topo map" in 2002, but then took it away in 2009. The current way of the web is akin to leaving that task to a global chamber of commerce and their creation of world wide map that looks like that found on the back of menus at a local tourist stop.

Access to the world's information is at present almost completely in the hands of a few large companies and largely in the hands of just one. As an aid to thinking rather than shopping, diversity in search might be crucial. Lack of such diversity might be bad for business, but might be even worse for humankind.