Some discoveries of search engine results

Apologies for cross-posting.

With the interesting developments of a new political season upon us, I found this article to be exceptionally interesting: “Internet search engines may be influencing elections” ( This article summarizes some recent research published in PNAS by Robert Epstein, “The search engine manipulation effect (SEME) and its possible impact on the outcomes of elections” (

In essence, their research demonstrated that when people are searching for information to help them decide which candidate to vote for, candidates that place higher in the search results do much, much better than those lower. So, if you search for information on a topic that is important to you and candidate no. 1 comes up very high in the rankings, you will tend to vote for that person, or if no. 2 comes up higher, you will tend to vote for that person. They discovered this by making a fake search engine (Kanoodle–love the name!) where they could manipulate the placement where they could decide that one candidate or the other would come up first.

What I found the most interesting was the reaction of those people who actually figured out that the results were being manipulated, or in other words, those people who are doing the most thinking. Their reaction: “Very few subjects noticed they were being manipulated, but those who did were actually more likely to vote in line with the biased results. “We expect the search engine to be making wise choices,” Epstein says. “What they’re saying is, ‘Well yes, I see the bias and that’s telling me … the search engine is doing its job.’”

I find this conclusion surprising, but completely logical, when based on the commonly held assumptions of search engines.

There are many political considerations that arise from this finding (and that is what the Science article focuses on), but this is not the venue for those topics. I will say that the results themselves should not be all that surprising to those working in SEO (Search Engine Optimization, the attempt to influence search rankings) because it is such a flourishing industry. Manipulating search engine results is exactly what SEO is all about and, it works! This research only emphasizes that.

But from the point of view of how people respond to a library catalog search result, is the same thing happening? When they do a keyword search and see what pops up, do they look at the top few results and “believe” that the search engine is “doing its job” with putting the most “relevant” results to the top? (Although I have pointed out repeatedly that the commonly-accepted definition of “relevant” is absolutely *not* the same as the mathematical/information science definition of “relevant”)

Also, even if catalog searchers understand that this is an example of “bias”, do they believe it means that the catalog is “doing its job”?

If this is how the public responds to search results of our catalogs, of course it completely flies in the face of what cataloging is supposed to be about: to be as objective as humanly possible. The idea of cataloging has been for someone to search the catalog “correctly” for something, e.g. “Hiroshima-shi (Japan)–History–Bombardment, 1945” and the *arrangement* of the results had absolutely nothing to do with their relative usefulness. So, if a record happened to come up as no. 1, it did not mean that it was in any way more useful or relevant or important than no. middle or no. last.

Therefore, even though catalogers may proclaim that we remain as objective as possible (I am one of those who has proclaimed that) this research may demonstrate that our vaunted “objectivity” has already been taken out of our control. The public believes in “relevance” in search engines as well as in (most probably) library catalogs.

I don’t know what to do about it, but that is also the conclusion of the authors of the research.

James Weinheimer
First Thus
First Thus Facebook Page
Personal Facebook Page Google+
Cooperative Cataloging Rules
Cataloging Matters Podcasts The Library Herald