Supercharge ROOMBA — again!

Question

Following Shog's trough here and the last post asking not to focus efforts on closing old questions here, I would like to apply the most logical solution: delete that crap already (this can be done if people downvotes the questions, but then there are users that just don't downvote :/).

Since the only way to delete a question is closing it, and we are discouraged to do so, maybe we need the ROOMBA to do it for us, more aggressively. I propose the following formula instead, since our needs seems to be different:

If a question is more than 90 days old, and...

• has a score of 0 or less
• has no answers with a score > 0
• is not locked

...it will be automatically deleted.

A test of what it would select in this query. Checked some of them, and were close-worthy.

Anyone with another formula? Or some other approach?

This query match 35987 questions that follow the above criteria, plus doesn't have accepted answer. This match those that have accepted answers.

Answer 1

I at first thought this was a good idea (FWIW I still think it is a good idea, just with a little more tweaking), however, after taking a brief look at your improved query (removing accepted answers) I am a little worried about one thing.

I checked the first 6 out, and of those 6 I found 3 that had good answers that just needed upvoting (and they were good questions too). That is a %50 percent chance of pulling a good question out.

If this trend continues, even anywhere near this high, we would be deleting an awful lot of good content and I don't think clearing out more old cruft is worth what we would be deleting.

Therefor I suggest we also remove questions that have an answer. I know that will still still some junk lying around, but I think the amount of good questions it will save us is worth a little more cleanup on our part. I had Braiam modify the query once more to filter out questions that have an answer. It still finds some 18k questions and after going through the top 7 I wouldn't mind losing any of them.

Answer 2

The existing question sweeps do enough and your alterations won't fix the review problem.

The post you mention, "asking not to focus efforts on closing old questions", was (and remains) a frustration that old answered things end up on the review queue, that people should be more vigilant about how they review and also that the review system itself should be prioritising today's junk over yesterday's so that everything new gets eyes-on and it never turns into yesterday's junk.

While shenanigans like this keeps happening, I'm going to be grumpy about this. I know your post isn't directly about this so I'll be quick:

• One user finds it (I can't imagine why, there are no recent events I can see) and votes to close
• That generated a review task that sucks in 7 more people resulting in a 4/3 split to close (meaning the original stood).
• This left it on the close queue and 4 more people closed it
• Then one user and a mod re-opened.

14 separate people involved just to preside over a 3-year old, extremely answered question.

Getting back on topic, your solution is not remedy to the problem described above, you're just lowering the amount of fuel. We could keep adjusting the gearing on this until we just have a bot that deletes everything 2 hours after it's posted unless it has a +10 answer. There'll be a lot less crap in the review queue, that's for sure.

So ignore the review issues. They're a problem but they aren't this problem.
Our princess is in another castle.

Unlike Shog's suggestion, you're suggesting we delete a number of questions that have had no votes whatsoever. I think you summarised the ethos quite well in this —trimmed by me— comment:

Any answer not upvoted is potentially bad at the same time is potentially good, but that doesn't mean we should preserve all of them, since most are based on bad questions

I'm just not sure automatically drawing lines around what is potentially good or not is the right solution to any of our various competing problems. You might as well start every question at -1 instead of zero.

And as has already been mentioned late answering isn't common but we have over a thousand questions that got their sole ≥0-scoring answer after 90 days. You'd be condemning the next thousand to just being deleted or needing to re-ask their question. This aspect is actually bigger (questions with more than one answer but the latest >90day one is the best) but my SQL-fu isn't strong enough today.

I'd much rather we fix some of the ultra-wasteful time sinks with a view to reinvesting that back into making sure zero-vote posts get more eyes.

Answer 3

I must say that, to me, the most logical solution seems to be ignore that crap already! One of the good things about the SE system is that old, ignored questions are forgotten as long as they're left alone. Yes, some posts should be deleted because they may (and this has never been proven, but let's just say it's true for the sake of argument) serve as doorways for more crap. I might point to crappy question X and take it as evidence that my crappy question Y belongs here. Perhaps.

However, old crap is not really harmful. In most cases, it will never be seen. So, sure, deleting old questions with no answers or no upvotes or whatever might clean some bad stuff out. But is that worth losing even a single good question? As long as the automated systems can delete the good along with the bad, I see no point in it. Why not simply ignore them and move on? Why should we bother so much about deleting old stuff when we can be doing more useful things like answering new Qs?

I realize you are not advocating focusing on the old and ignoring the new. What I'm saying is that I don't see any great benefit from deleting old ignored Qs. Closing them, perhaps. But deleting? Why?

Anyway, if the community really feels strongly enough about this, at the very least the time limit should be way more than 90 days. You never know when a question might get a great new answer, what makes you think that it will only be relevant for 90 days. What if I have the same problem 5 years later, solve it and then post an answer. The next person to come along will now have a solution and that would have been impossible if the question had been deleted.

So, personally, I would not even try to delete more than the system already does. If, however, you all feel that we must, then delete questions that have <=0 score and 0 upvoted answers and are at least a year (or maybe two or three) old.