Why Search Sucks (for Opinion Content)

A little homework assignment for you: Pick a hot button topic (Global Warming, Intelligent Design, Stem Cell Research) and try to convince your favorite search engine that you want arguments pertaining to said topic. I think you’ll find the results disappointing.

It’s incredibly hard to get relevant opinion texts to rank high on a search engine. This is the case despite of the wealth of opinion texts on the web in blogs, newspapers, forums and elsewhere. Texts from a neutral point of view always seem to trump opinion texts in a search engine ranking. This is certainly a feature and not a bug. However, it makes a large part of human discorse invisible to what one could argue is our primary means of finding information on a daily basis.

It’s interesting to consider the history of Google’s citation-based ranking algorithm in this context. The model content that the creators considered for the algorithm during early development was scientific papers, where citations effectively increased the prestige of the cited paper. However, the purpose of a citation in the scientific world is to pull in data, not the conclusion, from antother researcher. That data became what search engines valued, not the arguments or conclusions.

Philosophy and the Humanities don’t have a culture of citation. One does not cite an argument someone else has made, one simply restates and maybe gives credit. Unlike citing someone else’s data it is a fallacy to cite the authority of someone esle in a debate. For these reasons opinion texts are unable to build the web of citations that search engines depend on to assign relevance.

Debate is something quite different from peer reviewed science process and Search becomes less usefull as we stray away from that model. The hard data relevant to a debate is typicically unambiguous. The conclusions that one can draw from those data is where debate begins.

It is unlikely and probably undesirable for the current generation search engines to change their algorithms to value relevant opinion texts. However, if someone sets out to create something that does then I think they’ll find that the current ways of doing things don’t work in the context of opinion.