13

It's not a good day when you see this:

skeptics.stackexchange.com result with "disinformation" stamped on it * Red “disinformation” stamp not included on actual Google seaches!

(Note the NSFW topic! All links are academic discussions of this.)

You can do the same search in Google to see the problem yourself.

What has happened to arrive at the above screenshot?

Google takes from a Skeptics answer part of a quote that was analysed in that answer and found to be untrue. Only Google didn't understand what quotes are and presented this part as a factual answer in its snippet view on some searches.

But the answer isn't telling you what Google's Featured Snippet is showing. Google is showing you a quote from the answer that's only there to explain why it's wrong! No rules were broken (and in fact the answer was otherwise a good answer), but in the end we seemed to accomplish the exact opposite of what the site is here for.

Demonstrated by the screenshot included above, the Google snippet presents something the Skeptics answer shows to be untrue, and Google users finding this snippet concluded that Google shows "disinformation". Since that Google result is based on the reference from Skeptics:SE, they might be lead to conclude that Skeptics is the source for "disinformation". In any case, an upset Google user then stamps a screenshot of his snippet view with "disinformation" and the Skeptics-URL is included.

This came to my attention via an anonymous suggested edit. I think I did what I thought was the right thing there, removing the entire offending part of the quote, though it will take a while for Google to see this. The edit summary linked to the article FSC's Mike Stabile Helps Debunk Google Disinformation About Adult Performer Life Expectancy (source of the image above), which explains that Google is looking into the issue. But I doubt disinformation will be an easy issue for them to fix (other than on a case by case basis).

That shows three problems:

  1. Google is using Skeptics inappropriately, like: takes quotes out of a quote context, and thereby may actually help to spread disinformation.

  2. They way Google presents these misleading snippets as shortened second-hand can taint the reputation of this site.

  3. Answers on Skeptics might be abused by selective quoting, whether in any automated fashion or by actual humans.

What can we do? How can search results like this be found and remediated before it causes harm? Hopefully getting the word out is a good first step.


Update: It seems like the time for Google to update is less than 24 hours. Now it’s pulling the snippet from the question and I’m struggling to think of a way to make this better with edits. It’s one thing to remove the quote from the answer, but almost every question here is just a quote of a source that can’t be trusted and many of them don’t even have answers.

9
  • We can hold out that Google will realize answering questions was never in its mission, and will return to the search superstar it once was. Maybe.
    – fredsbend Mod
    May 5 at 1:45
  • Although I hate to see it, I don't think there is much we can do. We are a community committed to fact checking. Our span of control is limited to this site. We can't be held responsible, nor should we feel responsible what others do with the info they find on this site. In this specific example, it seems that Google fixed the issue and I think it is an isolated case, seeing as this is the first example in ten years that I know of.
    – Jordy
    May 5 at 8:01
  • I can't see that snippet in Google search, and I'm not sure I understand what's going on: Is it Google taking a part of an S:SE answer, doesn't understand what quotes are, then uses just that quote and presents it to its own users as 'the answer', only to then slap that sticker on its own answer (since that parsing is now the misleading thing, done by Google alone)? Then all the 'harm' is entriely Google's fault? The miserable drive for shortness, the faulty parser, the abysmal 'fact-checking', the apparently brain-dead choices for presenting Google snippets as 'answers'? May 5 at 8:03
  • Re: Your update. I took the big red thingy as part of what google now shows itself, not realizing it to be an altered screenshot. You write "see for yourself". Expecting such a stamp overlay, I not even saw the stamp-less snippet (which may be an on/off thing. Ppl often post such snippet-shots, when my same search does not show any snippet at all. I followed this down to Twitter to realize my mistake) So: on first read I mixed up that thing, thinking that also the snippet (with badge) would be 'on Google to show you it's wrong'. Could you clear that up in this post's text? May 5 at 12:45
  • @LangLаngС Sorry about that 😅, is it clearer now?
    – Laurel
    May 5 at 13:13
  • No apology needed, clearly my bad as well. But the addition is so short that I feel the quotes below here would still be necessary. To avoid reliance on comments, I offer my ideas for an edit. That's not perfect, even a teensy bit redundant. Please revise further, cut as you like or roll-back. May 5 at 13:46
  • Just now I re-clicker your link to gg and the snippet is there, in misleading the form complained about. Scanning the results presented on that page, one gets a nice feeling of how bad the results generally are. Since we know squat about how the algo chooses the snippet, I suspect that it might be reinforced by all the other results that repeat the bad info, that say the same? Given your recent Qedit on main, we should try to monitor closely when that snippet changes (but I guess, we can't do anything sensible or reasonable? & if there is any harm from that, it's Google dealing it out) May 7 at 11:48
  • Edit this post to include a NSFW warning on the "FSC's Mike Stabile helps debunk..." link, please? Just ran up against my university's web blocker clicking on the link, which was an unpleasant surprise. I'd suggest an edit myself, but the SE software won't let me because this is a Meta site.
    – nick012000
    May 13 at 0:37
  • @nick012000 Point taken and edited the warning to apply to all the links. (Though none of the links here are too NSFW, it's just that not everyone uses the term "adult performer".)
    – Laurel
    May 13 at 1:09
15

We have two promising long-term options but neither can be done without developer help:

  1. Tell Google to not put anything on this site into a snippet.

  2. Tell Google to not put questions or bad answers into snippets and then edit other answers when we see issues come up. (Bad answers are anything downvoted or having a banner for lack of references.)

For more information see how to opt out of snippets.

This may be worth bringing up on the main meta as I think other sites would be able to benefit from this.

3
  • Opting out entirely sounds good to me.
    – fredsbend Mod
    May 7 at 5:42
  • 5
    Completely opting out of snippets might be a difficult sell (it would probably require an in-depth analysis of what it means for traffic). But the <span data-nosnippet> tag seems like it would be a good option to exclude specific text (quotes in questions in the case of skeptics). The use-case may be too niche to implement though.
    – tim Mod
    May 7 at 9:31
  • It's worth mentioning that this apparently happened once in the ten years that Skeptics existed. To put it differently, this happens once every ten thousands questions.
    – Jordy
    May 11 at 10:26
10

This is really unfortunate, but I think there is little we can do here at Skeptics.

Answers could be written in a way to de-emphasize inaccurate content (onyl short quotes, etc), but quotes in questions can't really be avoided.

Stack Exchange already properly marks quotes via blockquote tags, which Google could — and probably should — consider when creating featured snippets.

The feature in question is against Google’s snipped policies (inaccurate or misleading information), so I think the only thing that we can do is flag it via the "feedback" button.

3
  • I wonder if us all flagging it would make it change faster. Still, while this is something that you can and absolutely should do if you see a snippet like this, it’s unsatisfying as a solution because it’s reactive and can only be done when we see there’s a problem. :(
    – Laurel
    May 5 at 11:02
  • 1
    I still wonder if this is actually a big issue. AFAIK this is the first time it has happened with Skeptics, and it happened with an answer that is not really great. Notice that the answer in question only points out flaws in methodology (which is basically an argument from fallacy).
    – Jordy
    May 5 at 11:50
  • 1
    We should not change anything here — if this is Google's guilt alone. Also: its internal algos & workings are secret & subject to change. A moving target within a black box. Also: where does the stamp come from? The auto-systems are too bad for anything. So a human user does that? So why not tell that human to remove the snippet wholesale? If the slap is also machine-made, then we would need to know 'how'. Pass this problem on to MetaMeta is the utmost? Highlight everywhere how bad such snippets are/can be would be the better option. May 5 at 12:47

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .