29

This post on RPG.SE meta describes a heavily downvoted answer (current score -17) being shown as the featured snippet by Google when searching for the question title (or something resembling it):

Screenshot of a heavily downvoted Stack Exchange answer appearing in a featured snippet at the top of Google search results. The search query is "can you counterspell a counterspell" and the featured snippet begins with the text "No. You can't counter Counterspell on your own turn by using your Reaction to cast Counterspell." highlighted in boldface by Google.

The RPG.SE meta question asks whether the downvoted answer should be deleted to prevent it from appearing at the top of Google search results, but all the answers there so far say "no". The reasons cited for this include the fact that the answer does not technically meet the criteria for deletion* as well as the idea that the SE community should not subordinate its behavior to the whims of Google's algorithms.

However, it's (AFAIK) still part of Stack Exchange's mission to "make the Internet a better place", and promoting incorrect answers (as judged by community voting) to the top of Google's search results does not serve this purpose.

Fortunately, while we wait for Google to improve their snippet selection algorithms to take SE vote scores better into account, there's a simple technical measure that SE could use to prevent these kinds of egregious mishaps: the data-nosnippet HTML attribute can be used to instruct Google (and other conforming search engines) not to include text from certain parts of a page in snippets featured on result pages.**

In particular, Stack Exchange already shows answers with a score of -3 or lower as grayed out, clearly indicating (to human readers with normal eyesight) that those answers have been judged by the community to be of low quality and possibly incorrect. This is technically implemented by adding an extra class ("downvoted-answer") to the element surrounding the answer when generating the page HTML:

<div id="answer-86789" class="answer js-answer downvoted-answer" data-answerid="86789" data-parentid="62217" data-score="-18" data-position-on-page="5" data-highest-scored="0" data-question-has-accepted-highest-score="1" itemprop="suggestedAnswer" itemscope itemtype="https://schema.org/Answer">

As these HTML elements are also already given a number of custom data attributes, adding one more attribute based on the same logic should hopefully be fairly easy to implement. Thus, my request is:

Please add the data-nosnippet HTML attribute to any grayed-out answers with score -3 or less.

*) While clearly not enjoying community consensus, the answer is also clearly a genuine and well written attempt at answering the question, and the question itself is somewhat subjective, with a small minority of voters (+3 vs. -20) actually agreeing with the answer. Thus, while (arguably) wrong, the answer does not appear to be "extremely low quality [with] little to no scope for improvement" and thus, per the help center and the FAQ, it may deserve to be downvoted but not deleted.

**) AIUI, Google may ignore the data-nosnippet attribute when generating a Q&A carousel based on structured data. But those already take vote counts and acceptance status into account as long as these are marked up correctly, which SE already seems to be doing.

Improve this question
5
  • 1
    Communities in general, RPG no exception, are quite happy to ignore a range of technical criteria for a lot of things that are built into the network and it's systems. Arguing that technically it shouldn't be deleted (an easy quick existing solution that bears almost no defence, and which is very easily reverted by any moderator for the sake of a single flag) instead of seeking some change to the network, which may take a long time and may never happen, seems a bit silly, to put it politely.
    – Nij
    Jan 13 at 2:20
  • 1
    On some sites (especially meta sites), answers have lower score thresholds for graying out. For example, on here and MSO, it's -8.
    – Sonic the Anonymous Hedgehog
    Jan 13 at 4:22
  • 1
    I might extend this to answers with score < 0 (or even <= 0), because it really needs to have a positive/non-negative score to be considered something that's acceptable to be displayed out-of-context in search results. The mere fact that one person said something should not, by itself, allow it to be displayed as if it's authoritative.
    – NotThatGuy
    Jan 17 at 14:10
  • 3
    @Nij unless the plan is to delete every downvoted answer, the problem is practically guaranteed to occur again, unnoticed until someone happens to google something and see that the snippet is taken from a DV'd answer. It's probably already happening all over the network, with no way to tell. Ilmari's suggestion targets the cause, rather the bandaid fixing whichever things people happen to find. Seems useful to me, even if it takes a long time.
    – mbrig
    Jan 19 at 21:49
  • Ilmari's solution will take time, if it is ever implemented, if it is even considered. Using bandaids instead of a doctor might be a bad idea, but refusing to use bandaids because you might find a doctor next month (or next year, or maybe never) is stupid.
    – Nij
    Jan 19 at 23:44

0

Reset to default

You must log in to answer this question.