After yesterday’s mass shooting in Las Vegas, Google briefly gave its “Top Stories” stamp of approval to two 4chan threads identifying (and triumphantly smearing) the wrong man as the shooter. Google apologized for including “inaccurate” web pages in its top results, saying that its algorithm had spotted a burst of activity around a little-used search term (the name of 4chan’s so-called suspect), created a Top Stories carousel, and favored “fresh” content there above more authoritative sources.
This is far from the first time Google’s search results have purveyed misinformation. In March, it finally instructed human quality raters — who manually evaluate web pages to train the Search algorithm — to flag offensive and factually incorrect material, which Search could then downgrade for users seeking general information about a topic. As the 4chan incident shows, though, it still has blind spots. And that’s not really because of a problem with Google’s algorithm. It’s happening because Google’s core business has never been about defining truth — yet that’s what Top Stories is implicitly promising.
Google publishes detailed guidelines for website quality ratings, where it outlines many ranking factors that include originality of content and “expertise, authoritativeness, and trustworthiness.” But it won’t go into detail about how various factors intersect, a crucial question that it says is too complicated to answer. Among other things, Google won’t explain just what makes a “Top Stories” carousel appear for a particular search term, except that it uses a special set of signals to detect whether users might be interested in seeing fresh or “breaking” links. Once the carousel appears, we don’t know how its stories are chosen compared to Google’s normal search results — except, again, that there’s an added focus on freshness.
I’m sure there are complicated answers to these questions, but there are also basic principles that Google could publicly commit to following, if it wanted. Are Top Stories supposed to be held to higher than usual trustworthiness standards than average search results? Does the carousel only appear if there’s a baseline general-interest newsworthiness, or is any internet micro-controversy supposed to trigger one? What is a “Top Story” even supposed to be?
“Top Stories” has only been part of desktop Google Search since the end of 2016, when it replaced similar “In the News” boxes. While the new name justifies a much broader and more flexible range of content, it leaves the overarching purpose unclear. If good Top Stories are defined by the same standards as good generic search results, they should just be the top-ranked links for a query. If the point is to showcase fresh content, they should be called something like “Recent Stories.” If they’re the most high-quality and definitive results, Google needs to explain its standards — and why they’re different from the larger ranking system.
A Google spokesperson told us that Top Stories could be valuable for immediately presenting a range of different types of useful information on a search query, especially when it’s newsworthy. But Google already has a News box, which sets search algorithms loose on a smaller list of approved sites. It seems easy to offer an expanded version of this with a larger list of general-purpose websites, exclude sites with low “authoritative” rankings, or otherwise provide special guidance for these sections. Conversely, if Google can’t define why “Top Stories” are special, then it might as well abolish them — there’s no reason to give a few arcanely selected web pages special treatment.
Google’s original PageRank algorithm was built to deliver the most popular and influential results for any search query, whether or not the content was true or good in a more philosophical sense. It’s gotten far more complex since then, but being the web’s best directory is still very different from being its ultimate arbiter of quality. As it puts more and more focus on its AI assistant, Google wants to be both — but it refuses to acknowledge that they’re different things, and that the tools that work for one might be bad for the other. As long as that’s true, Top Stories will stay like it is now: a vaguely named, poorly defined system that conflates popularity with value.