More Procrastination

I've engaged in search engine comparisons in the past, both using sites providing the support framework for the tests and running redundant queries on my own; in fact, when James Fallows wrote a Bing vs. Google note a couple of years ago, I wrote him to recommend Yahoo!, which I'd found very good at filtering out irrelevant results. That was before Yahoo! dropped support for their superior engine (for me, in English) to use Bing for their searches.

Yesterday, ReadWriteWeb described "Bing It On" as "A 'Pepsi Challenge' for Bing and Google," an official effort on Microsoft's part to test the same queries side-by-side with the two search engines. It is a test, presenting only the top (first page) results returned, and one 'plays' five times with the search terms of one's choice. I presume Bing has some way to mine the results to try to figure out why they lose when they do, but I frankly don't know how I would structure that analysis. (But now I'm thinking about it).


For Dave Copeland, the ReadWriteWeb writer, Google won 3-1-1. He includes information about some research findings which may explain the results. The research "found that Google outperforms Bing on simple one-word queries. Bing generally delivered more precise results for simple, multi-word queries and complex multi-word queries." His queries were simple multi-word, not simple one-word, so it does not seem to me that the explanation works.

Nor does this seem to explain the 4-1-0 Google scored with my query set!  None of my queries were one-word, even excluding articles, prepositions and conjunctions. Two were in French, one of which included a spelling ambiguity. My list was not "designed" or planned, it just happened that way, but probably is representative of the queries I submit.

My list:

• stem family joint family
• le module sp\xe9cifi\e9 est introuvable
• Is Zoroastrianism pagan?
• limage populaire
• cooperatives, private equity

The draw was on 'limage populaire', a curious name I spotted on a tombstone recently in a small village churchyard in Belgium. It also contains the spelling ambiguity, as "l'image populaire" is much more common, although "limage populaire" is a possible--not very comprehensible--construction.



The "Bing It On" site also provides a link to learn about the study in which they found "people chose Bing web search results over Google nearly 2:1 in blind comparison tests." The link is on the tally page you'll see after you try five queries, too.

---

Tags: :

Labels: Bing, criticism, Google, search engines