Google is very much a black box when it comes to ranking. Google issues general guidelines—remove user-generated spam, keep links to a “reasonable number,” create a “useful, information-rich site,” don’t overdo guest blogging, and so on—but as long as you aren’t blatantly trying to game Google, it can be difficult to know exactly why one site beats another in the rankings. While I can’t answer the question of why MetaFilter’s pages dropped in Google’s search results, I can ask a different question, which is whether the drop is justified.
I took a look at roughly 100 MetaFilter pages, most of them on its Ask MetaFilter discussion section, and came up with search queries that I thought would lead to them. For reference, I compared the results against results from Bing (which also powers Yahoo and partly powers DuckDuckGo).
Looking at the results, I can say a few things with fair confidence. First, MetaFilter is not being penalized across the board. It had appropriately high-ranking pages for a number of my queries. Second, for a decent number of pages, Bing and Google treated MetaFilter roughly equally. Sometimes Google ranked MetaFilter a little higher, sometimes a little lower, but there was nothing to raise an eyebrow. But third—and this is the important point—there were a nontrivial number of queries for which Google ranked MetaFilter pages way lower than Bing did, and seemingly without good reason. It was much rarer to see Google rate MetaFilter way higher than Bing did.
For example, if you search “maybe there is a god,” Bing returns, in the No. 2 spot, a MetaFilter page with agnostic book recommendations, titled “Books about ‘maybe there is a God’ for an agnostic with some doubts.” Fair enough. On Google, though, the MetaFilter page doesn’t show up until the sixth page as the 60th result. (Search engines make tiny algorithmic alterations all the time, so placement may be approximate.)
Or consider the query “typology of joy” (not in quotation marks). Bing aptly returns MetaFilter’s page—“A typology of joyful pursuits?”—as the top result. Google’s top result is a page called “Topical Bible: Types of Christ: Paschal Lamb.” The MetaFilter page does not show up until halfway down the second page of results.
One last one: Searching for “most amazing woman ever” on Bing will give you MetaFilter’s helpful “Who is the most amazing woman who ever lived” as the third result. (Answers, by the way, included British spy and French Resistance leader Nancy Wake, world’s first programmer Ada Lovelace, slave rescuer and activist Harriet Tubman, and Chinese pirate Ching Shih.) Google puts it at the bottom of the second page of results, in 19th place. Google’s top result? A list of “the 100 most beautiful women ever.”
I turned up a number of other examples of this sort of puzzling deranking of MetaFilter pages, in which I judged the deranking to be unjustified, with many worse results above. Again, this deranking (i.e., lowering of the rank of a site) did not happen across the board, but it happened surprisingly often across the hundred or so queries I tested. My data is firmly anecdotal, due to a lack of time and resources on my part—I wouldn’t dare claim that Bing is better than Google, and some of the derankings may just be the luck of the draw. As a site that tries to attract users with solid content rather than flashy headlines and Share This buttons, MetaFilter may simply be getting caught between social linkbait sites like Upworthy, prestige properties like Yahoo Answers, and ultra-SEO’d sites. But the combination of the abrupt 2012 drop in MetaFilter’s fortunes and the preponderance of derankings leads me to suspect that something fishy is going on in the deep recesses of Google’s alchemical ranking algorithms, and that it deserves investigation. I doubt Google is intentionally penalizing MetaFilter—could it be a bug?
If so, it’s a shame: MetaFilter is one of the best-moderated sites on the Internet, and it’s tragic if it’s losing revenue to mistaken deranking, when linkbait and song-lyric sites frequently clog up the higher rankings. But all I have are suspicions. Will the Google oracle answer them?