Earlier this month there was a kerfuffle when journalists noticed that the top Google hit for "did the Holocaust happen"—a popular question, as it turns out—was, due to the vagaries of Google's own algorithms, a link to a grunting anti-Semitic neo-Nazi website offering readers about as much insight into that question as browsing the contents of a garbage can behind your worst local bar.
Prominently linking to a Holocaust-denying site was a problem for Google; they have long proudly declared that they don't edit or filter search results except in extreme cases (illegal content, virus links, etc.) and people being goddamn stupid on the internet is not, sadly, an extreme case.
On the other hand, Google is a search engine. Its goal is to provide users with useful, informative content. If the top hit to a basic factual question sends the reader off to uninformative, malevolent, or simply fraudulent responses then the Google algorithm is failing to do the key part of its job—the thing that keeps customers coming back to it. If you did a search for how does electricity work and the top hits were sites proclaiming electricity to be the unchained souls of wandering ghostly hippos then both the company's long-term future and Mr. Miller's eighth grade science class would both be in a world of hurt.
So after an initial round of insisting that the results were the results and they just couldn't help it if a neo-Nazi hate group peddling misinformation gains top search spots, Google is now patching its algorithm:
A spokesperson for the company told Digital Trends, “Judging which pages on the web best answer a query is a challenging problem and we don’t always get it right.” [...]
Google made sure to emphasize it is still sorting its results automatically, and that the company had simply “made improvements” to its algorithm when it comes to “non-authoritative information.” The spokesperson said that this “will help surface more high quality, credible content on the web.”
So the site in question isn't coming up as one of the top results anymore, at least not at the moment, though Gizmodo notes that other racist queries haven't been given the same adjustment. Reading between the lines, this suggests that the Google measure of "non-authoritative information" may involve a graylist of some sort, perhaps not penalizing particular sites by name but penalizing certain phrases or the "authoritative"-ness of the site's other content. It's difficult to say.
But what we do know is after the issue of fake news has roiled other giants of the internet, Google is paying new attention to its own role as potential megaphone for such misinformation. We'll have to keep watch to see if the fix spreads to other racist queries or if it's a one-time patch for this particular hateful bit of revisionism.