Google’s ‘Misinformation Graph’ Strikes AgainBy: Chris Crum - October 29, 2013
Users have encountered another blunder from the Google Knowledge Graph with Google showing some quite questionable content, and presenting it as “knowledge” on a very high-traffic search term. This is only the latest in a series of misfires from the Knowledge Graph, but probably the highest profile example yet, given the search term.
Do you consider Google’s results to be reliable? Let us know in the comments.
The term is “st. louis cardinals”. As you may know, the team is currently in the World Series, so it stands to reason there are a lot of searches happening for that particular term. It’s currently number five for baseball teams on Google Trends:
Search for “st. louis cardinals” on Google right now, and you’ll probably see a Knowledge Graph result that looks something like this:
Okay, looks legit. Last night, however, things looked a little different, as Ben Cook pointed on Twitter (via RustyBrick).
— Ben Cook (@Skitzzo) October 28, 2013
@innov8ion yeah, ironically it stays longer in G than it does on Wikipedia lol
— Ben Cook (@Skitzzo) October 28, 2013
We’ve reached out to Google for comment on update timing, and will update if we hear back.
Update: We just got a response from Google’s Jason Freidenfelds, who tells us, “We crawl sources at different rates; for fast-changing info it can be within hours. But in this case it was a technical issue on our end that let outdated information through. We’ve fixed the issue.”
It’s unclear exactly how long this text appeared in Google’s search results, but it was at least for a few hours, according to Rusty. And the Cardinals did play a World Series game, so quite a few people probably saw it. Some even accused Google of being a Red Sox fan:
— Van Carver (@ThaCarverIV) October 29, 2013
We’ve talked about the reliability and credibility of Google’s Knowledge Graph results a few times in the past, mainly because things keep happening. In fact, it hasn’t even been a month since the last mistake we saw, when Google was showing an image of the singer/actress Brandy for brandy the drink.
After a little media coverage, they appear to have corrected it, but it took them a while, even after said coverage. They couldn’t blame Wikipedia on that one because the Wikipedia page for the drink showed a drink.
There have been other cases where Google has shown erroneous info in the Knowledge Graph. A while back, for example, it got a football player’s marital status wrong.
As I’ve said before, the errors may be few and far between, but how can users know for sure whether or not they can trust the information Google is providing as “knowledge”? Typically, users aren’t going to question the information they see here unless it’s obviously wrong.
In the case of the St. Louis Cardinals, it was obviously a prank, but people looking to spread misinformation can be a lot more clever than that. There’s no telling how much factually incorrect info Google is highlighting to users at any given time. Even if Google is able to quickly correct it, people can still be seeing it. As we see with the Cardinals example this can even happen on major search queries.
In some cases, we’ve even seen Google promoting brands on generic queries.
Meanwhile, Google continues to expand the Knowledge Graph to more types of queries, and to provide more types of information, potentially opening the door to more errors.
A side effect of Google’s Knowledge Graph is that people have less of a reason to click over to other websites. When Google is presenting the “answer” to their queries right in the search results, why bother to look further? You just assume it’s the correct answer. Not that there isn’t going to be misinformation on third-party sites, but at least going in, users can decide for themselves how much they want to trust a particular source. I think most probably trust Google enough to assume they’re displaying factual info on their search results page.
Google’s voice search also draws from the Knowledge Graph to provide users with answers, and this kind of searching is only gaining momentum as smartphone use grows. Users count on Google to give them factual information when they don’t point them elsewhere. Should they second guess the info they’re getting every time they get an “answer”?
Part of the issue is Wikipedia’s own credibility. Is sourcing the majority of the Knowledge Graph to Wikipedia a good idea in the first place?
This comes at an interesting time at Wikipedia itself. Last week, executive director Sue Gardner announced that Wikipedia had already shut down hundreds of accounts for paid edits. People have been manipulating Wikipedia for their own monetary gain, and apparently, some of the higher-ups had allowed it to happen, which is why it was even able to. Gardner expressed “shock and dismay” over the whole thing, and the investigation is ongoing.
Gardner, by the way, announced earlier this year that she was stepping down from her position, saying she was “uncomfortable” with where the Internet is heading. While Wikipedia, in general, has been an invaluable source of information for years, these things make you question its reliability, and by default, the reliability of Google’s Knowledge Graph, which leans on Wikipedia so heavily for its information. This is, by the way, where the Internet is headed – at least where Internet search is headed.
“This is a critical first step towards building the next generation of search,” Google’s Amit Singhal said in introducing the Knowledge Graph.
And in case you’re thinking about Bing, it has a practically identical feature (though I’ve not seen any reports about the Cardinals blunder related to Bing).
It’s very possible – perhaps likely – that the majority of the answers and information that Google’s Knowledge Graph feeds you is completely accurate, but if you’re ever searching for anything important (remember, Knowledge Graph includes nutrition and medical knowledge now), you may do well to remember that St. Louis Cardinals example, and continue your research. Verify the important facts. If Google can get it so wrong on such a hot search query, it can probably get it wrong on more obscure stuff.
Is this the direction search should be going in? Do you trust the Knowledge Graph? Share your thoughts in the comments.