1

We have a user on Ask Ubuntu who just received the "do better or we won't let you play" nastygram, while writing a new question. They aren't banned yet, but they're clearly being told they will be if they carry on.

I understand the point of this feature is to stop people flooding a site with bad questions...
... But this user has a good —albeit brief— record:

  • 1 accepted question with a score of +4
  • 1 close-dupe question with a score of +2
  • 2 automatically-closed (by the Community user) with scores of ±0. I've looked at these questions and they weren't bad, they just didn't attract enough of the right people to get a vote or an answer. Ask Ubuntu is broad enough that some people just don't see some subjects like that.
  • 1 flag against them (the duplicate post)

So they hasn't attracted a single negative vote... But they're in "danger of being blocked from asking"?

What is the system actually counting to denote a "bad user" here? I suspect their community-deletions and the dupe are counting against them but I don't think they should.

Dupes happen, especially on technical sites where you can ask exactly the same thing in a dozen different ways... And unanswered, unvoted questions are just as a symptom of a less popular topic as a bad question (as I've said, the quality here was good).

Moreover, the ultimatum we show these users tells them to "improve" their existing posts. How on earth are they supposed to do that here? The Community-deleted ones are only salvageable by a moderator and the one that is duplicate will always be duplicate.

It's impossible for the user to fix themselves; they're stuck.

Certain flag and closure types attract automatic downvotes. These are what you should be counting to indicate bad quality. Until that point you shouldn't be counting users like this in people who are on the brink of being cut off.

And unless they're doing something demonstrably wrong, we shouldn't be banning users for things they can't actually fix.

It'd also be nice for moderators to be able to countermand any automatic blocks. I feel helpless when users come into chat begging to be allowed to try asking another question but the system won't let them. This user in particular has done nothing wrong but this is far from an isolated case.

Please fix.


The message seen:

Wait! Some of your past questions have not been well-received, and you're in danger of being blocked from asking any more.

For help formulating a clear, useful question, see: How do I ask a good question?

Also, edit your previous questions to improve formatting and clarity.

  • what exactly did the user receive? i myself someday received a warning when i was asking a question, it was look like some message box that i could click "ok, i'm going to do better this time" and it disappeared and i could still post a question. did the forementioned user receive a warning like mine or was she completed banned from posting anything? – 把友情留在无盐 Jun 5 '15 at 9:12
  • I'd say, self-deleted questions are a fair game, it's just deleted-because-boring questions that shouldn't count. – John Dvorak Jun 5 '15 at 9:33
  • As for helping the user, you can undelete the questions that Community deleted. If that doesn't lift the ban, upvoting them will (and if none of the user's posts deserve an upvote...). – John Dvorak Jun 5 '15 at 9:35
  • @JanDvorak That's it, I think there's a gaping void between "good enough for an upvote" and "so bad we shouldn't let them post any more". Just being mediocre shouldn't get you blocked and in this case the user has shown they generally ask good questions. – Oli Jun 5 '15 at 9:42
  • @soubunmei Added. – Oli Jun 5 '15 at 9:44
  • In that case just undeleting those zero-scored ones should lift the ban – John Dvorak Jun 5 '15 at 9:46
  • 1
    @JanDvorak It seems there is no ban present for the forementioned user. It is currently just a harmless warning. – 把友情留在无盐 Jun 5 '15 at 9:50
  • In that case... undeleting these zero-scored posts should lift the warning :-D – John Dvorak Jun 5 '15 at 9:58
  • @soubunmei No, it's an ultimatum. It scared the user into opening a meta post to ask what they'd done to upset the system. In this case the user really has done nothing wrong; they are a good user. They should not be seeing the message. That's what this is all about. – Oli Jun 5 '15 at 11:56
  • 2
    @Oli your use of the word "auto-block" in the title is very misleading. Based on the title and the original version of this question, I think people were assuming you were talking about the post-ban. And not the low quality warning. – psubsee2003 Jun 5 '15 at 12:00
  • @psubsee2003 Maybe, but I think one is just an extension of the other, both on the same scale (eg warning at -3, block at -5). My problem (as I've explained a few times and have just edited in) is what contributes towards that score. Posts the user is not at fault (like all these) shouldn't tar them as a bad user. – Oli Jun 5 '15 at 12:06
  • 1
    I suspect this is a quite rare "false positive" - that message is shown based on a different criteria than actual post ban, i.e. it will take lots more to actually have that user banned. I suggest adding "How close is this user to being banned?" stat for moderators only, which will use the actual, super secret, criteria - then moderators can know for sure who is really nearing a ban, and who just get warning while being still far away from ban. – Shadow The Curly Braced Wizard Jun 6 '15 at 10:08
4

The user simply received a warning that was added to the system to help guide the user. There is nothing about receiving that warning that will benefit from a moderator being able to manually remove it after the user has already seen it. I don't know this, but I would even go as far to guess that there isn't even a mention of this warning in the user's history. It isn't a black mark, just a way of saying "heads up".

Since the user does not really have a lot of history (4 questions), this warning is very likely due to the fact that the algorithm has very little data from which to make a solid assessment of the user's actual performance.The problem here is the same with any kind of trending and projection. It takes more than a few data points to come up with any kind of reliable projection. Since the user only asked a few questions, there is little information to work with, deciding between a good user with a couple of unlucky breaks and a bad user with a small history of bad content is very difficult.

The message might be wrong in this case, but a user that is actually pushing close to the post ban could actually have the exact same inputs to the algorithm. You need to warn these users so they don't push themselves further into the hole. Stack Exchange made the decision to do that, at the expense of warning a few users who probably aren't in real danger.

I strongly suspect that the thresholds for this warning are intentionally set fairly low to help put users on the right path earlier and not wait until they are beyond help. That was always one of the criticisms of the post-ban. Once the user was told they were in trouble, it was often too late to do anything as they had so many bad posts. It was not recoverable without disassociating said posts. By warning them early, you hopefully and give them more guidance to keep them out of a complete post-ban.

Given all of this, maybe the solution isn't changing the algorithm but having a 2nd version of the message for users that have such a limited history and have no history of downvotes. Something like:

Some of your past questions have not been well-received. While you are not in danger yet, if the trend continues in the future you could be prevented from asking any more.

If you think you need help formulating your question, see: How do I ask a good question?

You may also find help and good resources on meta.

It could probably further softened but it gets the point across without being as accusatory.

  • 1
    Re your first line, I never said the user was blocked. My problem is the message and imminent threat of being blocked for a behaviour they haven't committed. This answer does not account for the fact the user seeing this message has done nothing wrong. They shouldn't be anywhere near a "threshold" (however low) or we'd be banning people before they created their first post. – Oli Jun 5 '15 at 11:47
  • 1
    @Oli I think it does cover. Per your assessment of the user's history. They have exactly 4 questions, 2 deleted, 1 closed. That is not much of a history to build a trend from. True there are no downvotes, but 75% of the user's posts are closed or deleted. He might not be a bad user, but a bad user could have an identical history. So the issue is do we warn the good user or not warn the bad user. SE chose the former so that they wouldn't miss not warning the bad user. – psubsee2003 Jun 5 '15 at 11:52
  • Re the second part, what can the user actually do for themselves in this situation? Their "bad" posts are a dupe and two auto-deletes that they can't undelete, nor should they differentiate their question to de-dupe it. That's not what the system was designed for. – Oli Jun 5 '15 at 11:52
  • 1
    ... Which is why I'm calling for an adjustment. A raw close/deleted-vs-open/answered score is not indicator of user quality and in this case it's very near harmful. – Oli Jun 5 '15 at 11:53
  • 2
    @Oli no one knows what the algorithm is. It is probably not a raw close/deleted-vs-open/answered score, I was just using it for an example. I spend my days trying to project performance into the future and I know how difficult it is to get any kind of meaning trend based on only a small number of data points. With 4 posts, there is probably not enough data to decide if it is a good user or a bad user. So they err on the side of warning a good user. Maybe the solution is softening the message when there are only a few data points instead? – psubsee2003 Jun 5 '15 at 11:57
  • You still seem to think it's acceptable to even count these closed/deleted questions as a negative. They should be zeros because they weren't the fault of the user. I'm sure the threshold is fine for users with genuinely poor posts, it's what's being counted that troubles me. – Oli Jun 5 '15 at 12:04
  • 1
    @Oli I'm not saying it is or isn't acceptable. What I am trying to say that in this specific case, the warning is probably wrong. But it doesn't mean it will always be wrong given the exact same inputs. There is simply not enough data to make a good decision. – psubsee2003 Jun 5 '15 at 12:07
4

In the past, nothing was said until the user had definitely done something wrong - many negatively scored, deleted, and closed posts, not offset by anything positive. That was widely agreed to be too late. Now the system says something before the user has done anything wrong, simply they haven't been quite right enough. And now you're saying we should wait till they do something wrong. That has been tried and it did not work.

Your best response is to use chat or comments to reassure this user there is no black mark on their record, and that if they do something right (such as posting content that is upvoted) the warning will stop, and the whole thing will be like it never happened.

As an aside, on your theory that the closed questions aren't the user's fault: if they were actively good, people would have answered or upvoted them. Their deletion is something that was caused by the user's behaviour. It's not a big deal or a terrible thing, but it's not some outside action that the user is not responsible for the way you're presenting it.

  • I disagree with the second half of your answer. This is a black mark and this user will be blocked if they carry on doing nothing wrong. The criteria is the same and they can't fix their old posts (as highlighted in the question). And on "fault", there is an external force: topic popularity. You've been on broad-topic sites like SO, so you'll know some subjects just don't have the same number of people on them as others. Ask Ubuntu is a big site with a lot of topics, some clearly more popular than others. That isn't indicative of quality. – Oli Jun 6 '15 at 9:58
  • No record is kept of the warning being issued. That by definition means it is not a black mark. It is not possible to look back and see that a user was once warned. Please stop stating the opposite of simple facts. Also, you cannot know that the user will be blocked, none of us do, because it is an automatic process with secret details. The warning is set to be more sensitive than the block, in all likelihood. And finally don't lecture me about site audience please. I have accounts on 40 sites. I do know about small sites. – Kate Gregory Jun 6 '15 at 11:45
  • Also, at the time of the warning, it's not necessary to fix the old posts. As you note, that isn't always possible. Hence the warning. If the user's next post or two are well received (they get some votes or answers, aren't closed, and so on) everything will be fine. That's the whole point of the warning. – Kate Gregory Jun 6 '15 at 11:47
  • Sorry but a permanent record is kept and is available to moderators in the user history screen. And I didn't mean to lecture, but you were blaming the user for posting objectively fine posts in an unpopular topic. I honestly don't want to come across as combative here but the system is scaring users who aren't doing anything wrong. That this whole scoring process is secret (even to moderators) makes this all the more infuriating. – Oli Jun 6 '15 at 12:59

You must log in to answer this question.

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