11

Unlike many here, I have some degree of confidence that users who are close to a ban can actually learn what they're doing wrong and improve. I also have confidence that problematic users are not just heuristic engines designed to post questions which marginally evade the filter. I think most of them would actually learn how to post good questions, if they were pushed to.

The question quality filter should become stricter as users move closer to a ban. This will force those users to post better questions, and hopefully help them learn. If they figure out what they should be doing, then they will, perhaps, avoid a ban entirely.

I'm not saying we should change the ban threshold at all. I think the banning mechanism is actually tuned fairly well now. However, we could be utilizing the question quality filter to push users in the right direction.

Improve this question
13
  • 4
    define should become stricter.
    – Mike
    Commented May 7, 2013 at 19:38
  • 1
    @Mike Raising the question quality filter's thresholds has always been somewhat of an abstract idea, since nobody but devs know how the filter internally works. We know of a couple existing checks (title, basic grammar, etc.) but that's only part of it.
    – user206222
    Commented May 7, 2013 at 19:38
  • 2
    I'm not sure how feasible this is. I would assume that the filter is already "performance tuned" to give the most reasonable results, trying to prevent the false positives.
    – Bart
    Commented May 7, 2013 at 19:42
  • 3
    @Bart I think that's the point here though; for users who have a history of producing low quality content we can error on the side of filtering; assuming that it's more likely for the content to really be poor than for it to be a false positive.
    – Servy
    Commented May 7, 2013 at 19:43
  • 1
    @Servy I of course know too little about the ban filter to confirm or deny any of that. But if the filter doesn't catch the crap questions that lead the user towards a ban, I'm not so sure that making it more strict will magically perform a better job. But of course I'll take a step back and wait for Shog9 to tell me I know nothing. ;)
    – Bart
    Commented May 7, 2013 at 19:49
  • People will upvote anything that makes it harder for a struggling user to post.
    – djechlin
    Commented May 7, 2013 at 20:05
  • @djechlin I think you're being a bit pessimistic overall. The goal here is to push users to have better posts, not make it harder to post.
    – user206222
    Commented May 7, 2013 at 20:06
  • 4
    @djechlin Yes, we are evil like that towards the poor struggling user. Oh how we laugh and laugh and laugh... Come on, we're trying to keep the quality up while at the same time not allowing a user to post bad content which might lead to a ban. I don't think this particular idea will work, but that particular sob-story of yours doesn't fit with this request.
    – Bart
    Commented May 7, 2013 at 20:08
  • @Bart we're experiencing a burst of ideas for hurdles to add here and our upvote-pattern isn't exactly being very discerning or critical for which are actually going to increase user retention or education. It's just, "that makes it harder, it might work, might not, so why not?" Why not just reject the next question attempt and say "try rewriting it" and accept the next attempt? Where's the line for hurdles we can throw before we're clearly wasting our time? The questions (incl mine recently) don't really identify the trade-offs, which makes me very suspect.
    – djechlin
    Commented May 7, 2013 at 20:11
  • @djechlin Then again, votes on feature requests are hardly any indication that they will actually be implemented. As far as they are concerned I regard them as a bit of a "let's fling stuff at the wall and see what sticks". And in the background I hear the laughter of the powers that be, who amuse themselves over the stuff we come up with while they happily ignore it. ;)
    – Bart
    Commented May 7, 2013 at 20:14
  • @Bart there's truth to that. I tend to view our job as more of maintaining the the items in the feature request queue but not prioritizing it or estimating effort. But I still think we should push ourselves to identify what sticks better, and I'm claiming we're not doing a great job on this topic. I'll open a meta thread on generally how we should be approaching the new user retention problem soon enough.
    – djechlin
    Commented May 7, 2013 at 20:16
  • 1
    @djechlin Make sure to identify that we have a problem that needs solving though.
    – Bart
    Commented May 7, 2013 at 20:17
  • This could be a good idea if the quality filter and the question ban mechanisms worked the same way, but they don't. They measure two very different metrics.
    – user102937
    Commented May 7, 2013 at 21:55

2 Answers 2

Reset to default
2

I think the current filter is not very forgiving to new users as it is. Of course I am biased for having been banned, but as a completely new user, who is earnestly trying his/her best to improve as a programmer and post quality questions as dictated by the guidelines, I still find myself struggling to adhere to them. Often times there are no direct warnings, and it seems as though your doing a great job, when out of the blue, you get banned.

Don't get me wrong, the filtration system is definitely amazing, but there could be some kind of less strict method to help rather then cast away new users.

So I would argue the opposite direction of the OP.

Improve this answer
3
  • 2
    I don't think it's very likely that objective users think they're doing a great job with their questions, when suddenly they are banned. A ban often has downvotes, closings, and deletions.
    – user206222
    Commented May 9, 2013 at 3:40
  • From my experience it only takes a few down votes. I understand that SO has a vast amount of very intelligent people. People who are just beginning though, as long as they listen to advice, and are obviously improving in the quality of their questions should be treated with a bit more leeway.
    – LearnIT
    Commented May 9, 2013 at 3:53
  • 2
    A few downvotes shouldn't do it. If by a few you mean 9 or 10 on multiple questions, I can see that - but I personally haven't seen a case yet where a user was banned after a few (being 3-5) downvotes. I can't speak for the team, though, since only they know the mechanism behind the banning system.
    – user206222
    Commented May 9, 2013 at 4:25
0

Shouldn't our quality filter be as strict as it can while producing accurate results? If we make it more guessy, how are new users going to learn the difference between a good question and a bad question, when some of their good questions we reject based on an unsure algorithm?

It's all the more important for a struggling (or apathetic) new user to get accurate feedback on this.

This proposal is analogous to auto-grading exams for struggling students so the teacher's time can be saved on students doing better. This is good for the teacher's workload but clearly bad for the struggling student. You could argue we should do this not to help onboarding a user but to reduce low quality question volume, on the grounds that so what if we screw it up for someone who has wasted our time once. But I suspect this will just lead to more "needsmoretextneedsmoretextneedsmoretext" in questions, making them lower quality.

Improve this answer
10
  • 1
    It doesn't do them much good either if we let their bad questions through the filter.
    – user206222
    Commented May 7, 2013 at 19:49
  • @KnightswhosayNi Sure it does - it will be reviewed by the community once it lands on the site and with possibly terse or harsh but basically constructive feedback. It's bad for us since we have to do that, not them.
    – djechlin
    Commented May 7, 2013 at 19:50
  • Case in point - Increase the strictness of the filter, so we don't have to. Plus, a posted question may still get an answer. A few downvotes won't discourage anyone if they think they absolutely need an answer.
    – user206222
    Commented May 7, 2013 at 19:51
  • @KnightswhosayNi a question ban will discourage. Basically I'm claiming that while this may be in the interest of the existing community, it's going to be harmful to onboarding, which from your first sentence in ? sounds like is the goal you are striving for. Analogously, do you automatically grade exams of struggling students to spare the teacher the time of failing them? Ultimately I think if we do this we'll just see a lot more of "needsmoretextneedsmoretextneedsmoretext..."
    – djechlin
    Commented May 7, 2013 at 19:52
  • 2
    I don't really understand what you're saying. A question ban doesn't discourage bad questions. It stops all questions. A question ban means "we've basically given up hope for you, sorry." A ban is different; this is trying to get users to learn something, as opposed to flat out banning them.
    – user206222
    Commented May 7, 2013 at 19:54
  • @KnightswhosayNi I guess your proposal makes sense in the sense that the user will have to try more tweaks to get it to pass and might happen to stumble on doing something better in the process. But in that case I strongly think that all that will happen is they'll just game the autofilter instead of increase question quality, so this fails in practice.
    – djechlin
    Commented May 7, 2013 at 19:57
  • 1
    I think people are way too pessimistic about these things. Because of that, I'm just going to have to disagree with you.
    – user206222
    Commented May 7, 2013 at 19:58
  • @KnightswhosayNi I still don't understand what value you get out of a stricter filter for users who have moved toward a ban. Shouldn't however that be made stricter apply to all users or at least all new users? If a new user was about to post a questionable question, isn't it as important they have to adapt to a stricter filter to post?
    – djechlin
    Commented May 7, 2013 at 20:03
  • No, because they don't have a history of bad questions. See Servy's comment above - if a user has consistently posted bad questions, we are free to err on the side of low quality. However, it's still possible to have good questions for new users that would hit an increased filter.
    – user206222
    Commented May 7, 2013 at 20:05
  • @KnightswhosayNi well, I think you're mostly just overestimating the strength of auto-filtering quality. NLP can't detect quality well yet, so we slapped in a few obvious things. But I suspect for each unit harder you make the autofilter, false positives increase exponentially. The calibration I think has to be rather precise. Effect will be new users just think it's a crappy site due to weird rules about posting. In any case we should be transparent and explain that this is an unusual measure taken for someone with a history of time-wasting questions.
    – djechlin
    Commented May 7, 2013 at 20:09

You must log in to answer this question.