Meta Stack Exchange is where users like you discuss bugs, features, and support issues that affect the software powering all 158 Stack Exchange communities.

What is meta?
Here's how it works:
  1. Any Stack Exchange user can ask a question
  2. The community provides support, votes on ideas, and reports bugs
  3. Your voice helps shape the way Stack Exchange operates

The Low Quality Posts queue recently gave me this question to review. It clearly looked like one of those posts which are there just to test us (not a very short post, and it has a score of +5).

I opened the full thread page, and it was already closed. I decided to vote to close too, but the review system didn't like that, and I got a warning telling me I should have clicked "looks good".

Now, how can the system know if the question is constructive/on topic/a real question? It can't, right? If those "examplary" questions are selected algorithmically, shouldn't their status (open or closed) be taken into account?

share|improve this question
I have no idea how the review system determines this, but I can't see how the "not constructive" close reason even remotely make sense on a question about designing an algorithm. – Adam Lear Sep 9 '12 at 16:18
Because it's too broad and could be answered 100 ways. – Rosinante Sep 9 '12 at 16:28
I don't understand the rationale behind "not constructive" there either, and so I've re-opened it. But yeah, showing you a closed question as an example of something that shouldn't be closed is kinda bogus (even if you did cheat by opening it up and using the closed state to override your own instincts on it). – Shog9 Sep 9 '12 at 17:43
@Shog9 - I might have voted to close this one too, if I couldn't think of a way to edit and improve it. Asking for "tips" seems no different than asking "What do YOU think". It gives people permission to just give a tip or tiny suggestion without actually answering the full question. With that said, I think this can be edited, so reopening is justified. – jmort253 Sep 9 '12 at 17:48
You see that a lot though, @jmort253, where someone asks a question they kinda think is a pretty big request, and so feel the need to put something like that at the end lest they get no answers for making it too much work to answer completely. It doesn't really ruin the question, but it is better without it. – Shog9 Sep 9 '12 at 18:04
@Rosinante Sure, but it's quite on-topic (perhaps slightly more so on Programmers, but still) and if folks actually put effort into answering it, it's a good candidate for being the good kind of subjective - multiple possible solutions are fine, so long as they're properly justified. The best one would bubble to the top. It's highly unlikely that this question would be an endless list of opinions, which is the bad subjective we want to discourage. – Adam Lear Sep 9 '12 at 19:07
@AnnaLear aren't algorithm shopping questions just shopping questions at the end of the day? – Rosinante Sep 9 '12 at 19:31
@Rosinante While I'm sure it's possible to write one that's just a shopping question, in general I'd say no. Algorithm design is a topic that attracts real questions with real answers not unlike the typical code-focused questions on SO. Heck, even most shopping questions can be rephrased to be answerable. It's all about focusing the question on the problem rather than possible solutions. – Adam Lear Sep 9 '12 at 19:41
"Here's my problem. Is there an open source library that solves it?" "Here's my problem. Is there an algorithm that solves it?" If both are OK by you, I'll need to readjust my close button reflexes. – Rosinante Sep 9 '12 at 19:52
My first "instincts" told me "looks good", based mainly on score an length (probably the same reason used by the review system to select that question as a gotcha). After reading the question I was not so sure anymore, that's why I "cheated". And I agree with Rosinante on this one. The question at the end is: "Is this a problem that has been solved by an algorithm before? If so I'd appreciate any direction in where to look. (...) Just a little stuck on the opportunity finding algo." A bit vague. Also, the post describes the problem, but not what the OP tried. – bfavaretto Sep 9 '12 at 20:06
The above being said, question is not about whether that particular question should be closed or not, but rather about the review system choosing a closed question as exemplary of something that should not be closed. I don't think it should assume a question "looks good" based on just score and length. – bfavaretto Sep 9 '12 at 20:09
@Rosinante "Here's my problem. Are there code changes that fix it?". You have to look at a bit more than just the form of a question. (But we are majorly derailing the thread here.) – Adam Lear Sep 9 '12 at 20:20
@AnnaLear, now that this is status-completed we can go back to derailing, if we want to! :) – bfavaretto Sep 10 '12 at 17:18
up vote 7 down vote accepted

Humorously, the system was being a bit smarter than intended there.

The actual query for the "Looks Good" audit tasks isn't just "open with a positive score", it's looking for universal approval from the community but also trying to exclude crazy things like "Favorite Programming Cartoon" and whatever's trending on HN or Reddit.

By those criteria, that question did in fact "Look Good" (it had nothing but upvotes and a "reasonable" score). The error is that it didn't consider being closed a sign of not "Looking Good". A universally upvoted (but not Gorilla vs Shark) closed question is pretty rare, so this bug slipped through.

It won't happen again, starting with the next build. I'm also going through history and making sure no-one else has been bit by this bug (and will remove any black marks if anyone has).

share|improve this answer

You must log in to answer this question.

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