Vanilla 1.1.9 is a product of Lussumo. More Information: Documentation, Community Support.
Recently, the question of how to keep people from using MO for "cheating" has come up a couple of times. My opinion is that the best way to keep people from cheating is to simply have a completely public searchable record of what everybody contributed. It turns out that sometimes the asker is able to delete a question, in which case the answers are also deleted, so the answerer's contribution is hidden from the world. I don't like the idea of anybody trying to use this as a mechanism for cheating in any way, so I combed the database for situations where this might have happened.
I searched for questions which had answers and were deleted by the owner. I found 91 such cases, but I don't know what to do with them. I guess I'll post them here so that 10k+ rep users can have a look at them and vote to undelete them if appropriate (in the next comment because of the character limit). Note that 10k+ rep users can also see what questions have the most undelete votes by looking at the delete tab of the tools menu.
I welcome any ideas about how to better deal with this issue. Ultimately, humans have to look at the deleted questions to decide if they're worth undeleting, but perhaps there are better criteria I could use to narrow down the search space. Maybe I should even be widening the search space to include questions that don't have any answers, but do have "substantive-looking" comments. How would I programmatically look for such things?
Also, is there a good way to filter out questions that (somebody has verified) really should stay deleted so they don't add noise to the sample every time I refresh the list?
10824 11032 11539 12870 13424 13729 13865 14280 15044 15251 157 15715 1595 16021 16628 17088 17677 18167 1872 18919 19029 19177 19383 19621 20449 21732 2258 22870 2288 22920 23089 23687 24 24509 24662 25106 25416 26420 28198 28510 28841 29203 30451 31134 31566 317 33175 33742 33748 33799 33867 34209 34706 34905 35028 35378 35563 36346 36689 36883 37720 37847 3813 38196 38333 39139 39213 39220 39442 40749 41872 42070 42107 42479 42740 42871 43061 43120 43268 43488 43505 436 44011 44732 45754 46259 46271 6742 7927 8576 9814
One or two of those may be mine. Could you e-mail me if there is any problem (since I don't have ten thousand points)?
I'm hardly going to be offended if you keep it deleted, but it looks to me like Chandan and my answers to http://mathoverflow.net/questions/15251 are reasonable.
As far as filtering, once you (or someone) goes through that list, then, the next time around, you can delete anything with deletion date before November 2010 from the list.
Not sure why http://mathoverflow.net/questions/15715 was deleted, it seems like a reasonable question and answer, although I don't know all the terminology involved.
http://mathoverflow.net/questions/16021 is also a good, though basic, question about Lie groups with a good answer.
I begin to think that some people are embarrassed when they get a simple answer to their question.
A lot of these come from miwa http://mathoverflow.net/users/3818/miwa. He or she seems to have a pattern of asking an analysis question which is possibly at the right level, and certainly well formed, but not very interesting, then deleting it as soon as he gets an answer. I'm not sure what should be done about this, but it does seem suspicious.
I have now checked through 25416. Besides miwa, and the questions I have linked above, all the others are off topic, offensive, or not research level. I think I am one of the more liberal members of MO, so I doubt anyone will disagree with me on this, but, of course, you are free to look for yourself.
http://mathoverflow.net/questions/31134 is a very basic question about derived categories, with a correct answer. I'm not sure what criterion I'm supposed to be applying here.
It seems to me MO has narrowed its focus since we started. We used to welcome basic questions from someone moving into a new field, as long as that field was something usually learned at the upper graduate level. By that standard, I think this question is fine. But, nowadays, I would expect it to get closed. That's probably a discussion for another day though.
Since David is checking in order, I'm checking in reverse order.
http://mathoverflow.net/questions/8576 is a question about elliptic curves and I am not sure about it. It looks like Scott Carnahan edited the question to be more reasonable than it initially was, although I am not in a position to judge the level of the current question.
http://mathoverflow.net/questions/6742 looks like a reasonable, if basic, question in commutative algebra. I am not sure about its level.
http://mathoverflow.net/questions/37847 appears to be a fine question in algebraic topology.
http://mathoverflow.net/questions/36883 appears to be a fine question in knot theory, although I'll note that the answer is from the OP, so it's not as if anyone is being disenfranchised by this question remaining deleted. This is perhaps a general case we should ignore.
http://mathoverflow.net/questions/36346 appears to be a fine question in algebraic topology.
Aaand I've now checked through 34905. As for filtering out questions somebody has verified, can we tag deleted questions with a specific tag for this?
http://mathoverflow.net/questions/317 is a good question about dualizing sheaves with a good answer.
http://mathoverflow.net/questions/33175 is a "big list" question in combinatorics. It's no worse than many others, but no better.
Harry, http://mathoverflow.net/questions/33799 is one of yours. A basic confusion about homotopy theory, but on a very high level.
I have now checked through 34706, so Qiaochu and I together have checked everything.
Yeah, I agree with that appraisal. His questions definitely seem not to be thought through.
It's worth someone rechecking the questions I checked, since (like I said) I was unable to reasonably judge the level of many of them. Several were of the type "question with what seemed like an embarrassingly simple answer" and I wasn't sure what to do with them.
Woohoo! It worked. Thanks David and Qiaochu for pointing out interesting cases. It looks like three of those questions have been undeleted so far (thanks also to Pete Clark and José Figueroa-O'Farrill for voting to undelete). I like the idea of keeping questions with embarrassingly simple answers around, but I realize other people aren't as eager to be embarrassed as I am. I can think of two good general reasons for voting to undelete:
I'll regularly post updates on this meta thread, following David's suggestion of only looking for questions deleted after the last update. It looks like there will be fewer than a dozen every month, so it won't be much work to keep up.
I just went through and linkified all the URLs above.
It's a bit unclear what we should be doing here... I feel a bit bad just unilaterally undeleted some of these, even though there are many linked to above where personally I can't think of any reason the poster would have wanted to delete. I guess those without moderator powers should vote to undelete more freely.
Ideally I guess we would email the deleters and try to persuade them to undelete of their own accord, but I'm not sufficiently motivated to be doing this.
@Scott: Undeleting is not a permanent action. The owner can again delete the post if she wants to. If somebody asks me a question and says, "this is probably a stupid question", I don't feel bad about saying "I actually think it's pretty interesting." Under normal circumstances, there's no reason to feel any worse about voting to undelete, is there?
Note, by the way, that once a <10k rep user deletes a question and leaves the page, she can't get to it anymore. So it's not reasonable to assume that a user really wants to keep a question deleted just because they haven't undeleted it.
Regarding the update:
It is curious that the answer of 46985 was deleted less than one minute before the question was deleted.
The following questions were already closed, and their deletion is probably not a significant loss: 31140, 48820
The following may be rescued, but I'm not sure: 48134, 48904, 50573, 50584, 50600
24218 was answered by the questioner, and the question did not seem to make sense.
48100 is a rather elementary ultrametric manipulation that perhaps should have been closed. I don't think there is a reason to rescue it.
48944 is a "tell me about X" question, and the answer is a link to question 35882. No need to rescue.
49345 was already closed, but the comments on the question were reasonably interesting.
49530 would have been closed if it weren't deleted, and the answer is rubbish.
50167 is duplicated in 50600, but with a minor revision. I'm not sure what user FS is trying to accomplish by deleting a question and asking it again. Perhaps the two can be merged.
50167 and 50600 have been again duplicated.
48134 looks reasonable, but the answer is not very substantive. I'll wait to hear commentary from more stats literate users before any action.
48904 had a good answer, so I undeleted it (and voted up the answer).
50573 is a mess, and the "answer" was actually the OP writing an addendum and unable to get back into their account. I would leave it be.
50584 was "replaced" by 50741. The answer (by Dave Roberts) was reasonably good, but its content was incorporated into the new post with acknowledgement. I would be inclined to leave it alone, or discuss with Martin and Dave before undeleting.
50904 (the newest duplicate) has nothing of substance that isn't already in 50600, so there's no point in merging it (if we do, we should delete all the comments, but I'd rather just leave it). I see no downside to merging 50167 into 50600, so I'm doing it.
One thing that's weird is that the answers are all undeleted but now have a vote to delete (something only questions should have). The owners of the answers actually deleted them (look at the revision histories). I deleted the answer which just points the OP to math.SE. It now has a bizarre revision history.
Here are the questions with answers that have been deleted by the owner since 2011-01-01.
49630 50951 51127 51374 52133 52133 52240 53250 53769
49630 was answered by the owner, so can be ignored. 50951 and 51127 are of the form "what's your favorite X?" so should probably stay deleted.
All look reasonable to me, although the last one could probably be usefully undeleted without hurting anyone.
37572 should probably be undeleted. Maybe 57647. I didn't look past the first six.
I've looked at the rest.
I feel a little weird unilaterally undeleting 37572, since my answer is really a request for clarification (and unsurprisingly didn't get the 2 votes necessary to prevent deletion).
61920 doesn't seem like a bad question, although Harry deleted his answer before Dan deleted the question.
I've undeleted 57647 and 60737. Weird side effect: Will Merry's answer to 60737 has a vote to delete, since he deleted it before the question was deleted. Should I re-delete it?
I don't know enough to judge 60698, 59877 (which showed up twice in your list), or 58662.
Here are the questions with answers that have been deleted by the owner since 2011-05-01.
25247 31991 34050 58134 61699 63334 63559 63893 64200 64516 65066 65488 65599 65862 65912 67412 67927 68396 68614 68909 68988 70480 70585
Based on a quick look, the following should stay deleted: 25247, 34050, 64516
The deletion of 63334, 65862, 65912, 68396 looks a little strange to me, but I do not have expertise in those topics.
Some of the answers have been upvoted. I always thought questions with upvoted answers can't be deleted, is that right?
@Hailong: An answer has to have two upvotes for the owner to be prevented from deleting the question.
@Anton: Out of curiosity, how do you know such details?
@Mariano: In the SE 1.0 days, I was very attentive to the nooks of the software. If the old meta.SE were still up, I could point to concrete evidence of this attentiveness. The software hasn't changed, and I still remember most of these things. For things I don't remember or never knew, there is always the option of experimenting on faketestsite. In this case, I just remembered.
I've looked through all of the questions and undeleted 63334, 65912, and 68396.
Things I don't know enough to judge: It hurt my brain to read 65862. 65599 seems to need a set theorist. 63559 seems to need a number theorist or algebraist.
Question 63559 seems reasonable enough, and even somewhat interesting.
After all this time I'm still not quite sure what our criteria for undeletion are, but if the idea is "the question contributes non-negatively to the site" then I would say undelete this one.
Also 65599 seems like an okay question to me, and it got a serious, not completely trivial answer by two very serious set theorists. (I don't really understand the question either, but since it is cleanly stated and no one asked "What do you mean?" I take that also to be evidence of its nontriviality!)
@Andres: "If $\mathfrak d$ is the cardinality of the least dominating family in the set of functions from $\omega$ to $\aleph_{\omega}$, what are its possible values with respect to the cardinal $(\aleph_{\omega})^{\omega}$?"
@Anton: thanks a lot!
I agree with Pete L. CLark that 63559 is a reasonable question and perhaps should be undeleted. My only reservation (and the reason I did not put it in my initial list) is that the OP answered it himself (albeit with the help of Laurent Moret-Bailly), so I felt he should have more control on this matter.