This question is brought to mind when I had a look at this (closed and deleted) question:

http://stackoverflow.com/questions/513778/what-do-you-think-of-junit-style-unit-testing-do-you-favor-behaviour-driven-tes/513807#513807

It also relates to my (closed) Meta question

Delete policy considered harmful

and it's a serious question.

Back in the Old Days, questions that were more open-ended were considered fair game. Some interesting questions were asked. Some people put effort and the sum of years of experience into answers.

Years later, five people come by and vote to close because it doesn't fit the Q&A format. The someone who happens to have mod powers deletes it.

Now, oddly, these questions still accumulate upvotes -- I noticed the issue on the question linked when I picked up an upvote for an old answer and wanted to see what I'd said. They were clearly considered informative at the time, and since they do still accumulate upvotes they apparently still are considered informative.

Given this, I've got to wonder, exactly what is the utility of the whole process? It appears to have essential flaws.

The Five-Vote Threshold is Flawed

Unlike the early days, SO now has a user community certainly in the many hundreds of thousands, and a question collection in the many hundreds of thousands or millions. The probability that a particular old question would be visited randomly would naturally seem to be low; purely on that basis, the chances of an old question accumulating five close votes would also seem to be low.

However, informally, I notice that many times I see the same names in the close votes. I don't hae the time or interest to perform a social-network analysis, but I would bet cash money that if one were performed, we would observe a small group of people who spend a significant effort closing questions. Consider, as a gedankenexperiment, a group of 5 users who decided they want to close all questions asked by Canadians before 1 April 2010. They compose a search and start voting to close; how many questions would be closed before it was questioned?

The point is that whether or not my social-network hypothesis is currently true, the five-votes-to-close policy is vulnerable to exploitation by cliques.

The Current Process Has Little Beneficial Effect With Old Questions

Consider: a question that is closed can no longer be answered, but the question and answer are still visible, and answers can still be upvoted and downvoted. Thus, the questions still require database space, and still generate transactions. Thus, closing an old question would not seem to significantly change the use of (increasingly inexpensive) resources.

Compare that to a new question -- it may be repetitive, it may be unconstructive, it may be illegible or incomprehensible. Those questions will tend to quickly receive close votes, and appropriately; the reason, however, isn't primarily for computation resources, but for the community's time and effort. They either have been answered, or can't be usefully answered, and so should be out of the way.

This clearly doesn't apply to an older question that was asked, voted useful, and received upvoted and accepted answers.

Given that, it would appear that the effort expended on closing old questions has essentially no beneficial effect for the community.

Closing Old and Upvoted Questions is Inherently Abusive

Again, consider an old question with upvotes and accepted, popular answers. Both questions and answers represent effort and knowledge on the part of the writers. In my case (because this is, after all, about me) I get paid a lot of money to write about computer issues, answer people's questions, and explain things. I have also contributed a lot of those same skills and talents to SO for free -- as evidenced by the fact that I'm currently in the top 300 all time contributors in terms of reputation. When a question I've answered, especially with a longer, informative answer, is deleted, that has the absolutely inescapable effect of destroying my work.

And in fact, it's not all about me: the same thing is happening to many other people. Each of them is having their work effectively denigrated at best and destroyed at worst.

The fact that it's contributed for free doesn't mean it has no value.

Suggested Improvements

It's certainly unfair to simply complain without suggestions for ways the process could be improved. Here are a few.

1. Increase the number of votes required to close old questions

The current voting process is, as shown above, inherently flawed and vulnerable to cliques. By increasing the number of votes required to close a question as the question grows older, this vulnerability is at least reduced. (Not eliminated: in fact, the process has the same vulnerability with brand new questions -- a conspiratorial clique could, for example, decide to vote to close all questions from Egyptians as soon as asked. However, with newer questions, that at least would probably generate controversy quickly.)

2. Establish a threshold based on age and vote count beyond which questions cannot be deleted.

At some point, a question which is still accumulating upvotes long after it was originally asked has proven its utility. Such questions should be preserved; they clearly represent useful information, and as discussed above, the benefit of close/delete on older questions is itself unclear.

3. Establish a place to put these older questions

As I've said above, deleting questions with useful upvoted answers is itself inherently an abuse of members of the community. A lot of this abuse now seems to be happening based on a new understanding of what is or is not "constructive" -- many of these questions were more discursive, at a time when discursive, pedagogical answers were acceptable. The work of community members in answering these questions should not be discarded.

Update

I notice that another person who's made similar complaints is Norman Ramsey. It's work noting that he's one of the more brilliant computer scientists in the world. Do we really want a policy that discourages people of that caliber from contributing?

share|improve this question
6  
If open-ended questions are good and accumulate lots of upvotes over the years, why are new open-ended questions bad? If they were left open for 4-5 years, perhaps they would then be considered good as well? –  Bo Persson Jan 19 '13 at 21:14
4  
Regarding deletion, I thought we hashed all of this out back in March with "Community-led deletionism: a protocol for sanity". –  Brad Larson Jan 19 '13 at 21:30
    
Does eyar == eyore? –  Rosinante Jan 19 '13 at 21:32
    
Gee, Brad, if policies once set can never change, then those questions shouldn't be getting closed, should they? –  Charlie Martin Jan 19 '13 at 22:50
    
Bo, frankly I don't think they are, and wonder if the policy should have been changed. However, the point is that now questions under the old policy are being judged under the new policy. –  Charlie Martin Jan 19 '13 at 22:51
1  
I would qualify this by saying that any question which has passed the age of six months/one year must be flagged for diamond-moderator deletion, and cannot be deleted by these people. This would stop close abuse. Additionally, if we spot these people, we need to report them for serially deleting old questions. –  Emracool Jan 19 '13 at 22:58
1  
I notice that when I look at Brad's link, the answers as upvoted seem to come down pretty heavily on the "too much deletion" side. I'll grant I'm one of those old hippie anarchist hackers, but I've noticed over the years that "some people are more equal than others" effect always shows up. –  Charlie Martin Jan 19 '13 at 23:01
    
Thinking about it, I suddenly realized that my reaction to having questions deleted on which someone clearly spent effort and thought asking, and responders spent effort and thought answering, is very much like my feeling about vandalism. Maybe I should say "other kinds of vandalism". But think about the reaction when a city comes along and paints over an old mural. Even if there's a new policy allowing it, you're still destroying someone's labor and the results of their skill. –  Charlie Martin Jan 19 '13 at 23:10
3  
I have to point out that, for all the talk of roving gangs of closers and deleters, the fate of your example involved only a pair - two to close, and one to delete. I've undeleted it to aid the discussion here. –  Shog9 Jan 20 '13 at 2:36
2  
Why are you valuing post age? I'm not trying to be a jerk but it does seem self-serving since more than 1500 of your 1600 posts are > 1 year old? –  Some Helpful Commenter Jan 20 '13 at 3:37
1  
The fact is many "questions" that are incoherent, incomplete, don't have any code and don't get any answers because nobody can understand what the OP wants, never get closed, while many that do have answers, many times accepted, get suddenly closed by a group of people that, in my opinion, don't have a clue of the subject. The general reasons described in the closing message justify everything. I agree with Charlie Martin, the closing process has flaws and some people take advantage of it to make closing a sport. That seems pretty obvious. –  faa Jan 20 '13 at 4:26
    
@Sho9 "I have to point out that, for all the talk of roving gangs of closers and deleters, the fate of your example involved only a pair - two to close, and one to delete. I've undeleted it to aid the discussion here." -- You're absolutely right -- if I'm reading that in fact, after one vote to delete it was summarily closed and deleted by a mod. Frankly, that strikes me as moderation abuse. -- Oh, and thanks for undeleting the question! –  Charlie Martin Jan 21 '13 at 1:26
    
@SHC, you're right, I don't have as much time for this as I used to. But you're mistaking my point: I'm not valuing old questions and answers per se, I'm valuing (I kind of hate using that as a verb, but what the hell) old questions and answers that have accumulated upvotes And yeah, it's self serving: it was one of my answers that brought the issue to my attention. Again. So what? –  Charlie Martin Jan 21 '13 at 1:29
add comment

3 Answers

up vote 5 down vote accepted

We need to consider methods of further protecting old questions. It abhors me to think that people would actually go around deleting old questions. Many and most of those questions are interesting and useful. This type of deletion I consider to be question abuse.

However, that is not to say all of them should be there. Many of them should be closed, and many of them should be deleted.

The function of closing and deletion with respect to old questions

Deletion is intended to be used only when a question is egregiously bad, and is no longer worth the community's time. This is probably one of the more overused features of StackOverflow, in that questions often get deleted which simply need improvement, or should not be deleted. Often, questions should simply be closed:

Close votes' intended superficial function is to lock a question so no more answers can be posted. The secondary purpose of closing a question, though, is to provide an example of questions that should not be asked. This is very similar to locking or protecting a question with the reason that This question exists because it has historical significance, but it is not considered a good, on-topic question for this site, so please do not use it as evidence that you can ask similar questions here.

This is effective and useful for old questions. Users can still reference the questions as useful, but the point is made. Those questions are no longer considered appropriate for StackOverflow. Many of the older questions I view have been closed for this reason - they are no longer on topic. It's a method of self-cleanup which is very effective and produces fantastic results.

The problem (and solution) with deleting old questions

This is, quite simply, information loss. There is a plethora of valuable and useful information stored on StackOverflow, and removing it from public view irks me. That information can be used by users to come - even if it isn't considered on topic anymore.

Users may delete these questions. In fact, it is very easy to delete these questions for old users. A handful of them go around deleting old and useful questions with the justification that they are off-topic. Thus, those old questions need to be protected somehow. We need to find a way to preserve the information.

Restricting deletions of old questions to diamond moderators

This is the first possible solution for handling off-topic questions. There are obvious benefits here: old questions which should not be deleted are removed. However, this causes more problems than it solves.

First and foremost, There is a high potential that moderators will be swamped with delete votes on old questions. As a simple matter of respect for their time, it is rude of the community to dump that problem on their hands. They are there to guide the community, not control its minutae. It simply isn't their job.

Secondly, it runs counter to the "power to the people" paradigm StackOverflow is constructed from. It is our community, and we moderate it. We guide it; on StackOverflow, the people balance the people.

Slowly increase the number of votes required to delete a question over time

I like sigmoids. They're fun. And practical.

The concept behind this is fairly simple: in order to better protect questions, we slowly raise over time the number of delete votes required to remove the question. I just took a couple minutes and worked out a sigmoid for a function which starts out around 5, and increases to 15 from 0 to 20 (rounds up from 14.5 at month 20).

Sigmoids!

Whenever a close vote is cast on a question, this function would be computed. That way, as a function ages, the number of votes required to close it slowly goes up. It's a form of artificial protection which would keep old questions alive unless they truly need to be deleted.

Graph of sigmoid

Of course, this is almost linear to the point of triviality, so in all honesty this could better be expressed as (where d is days):

Linear boo :{

But enough equations. Quite simply, the point of this would be to grant users slowly increasing protection over time. To build from this, any function should account for the number of received upvotes in the number of votes required for deletion. This can be as simple as +1 vote per 40 upvotes.

Why would this work?

The fact that a question has not been deleted for an extended period of time is implicit community approval of its existence. Of course there are going to be questions that are difficult to spot.

Cease thy serial deletions!

I have been active on Meta for only a few weeks now, and already I've seen several complaints about old deleted questions. Granted the small percentage of people who are likely to spot a deleted question, and the even smaller of percentage of people who would report deleted questions, this has the potential to be a rather significant problem.

The system should internally report suspicious deletion activity. If a user is serially deleting, or voting to delete, old questions, this should be reported to moderators. Though moderating these types of issues can be difficult, it's worth the time to stop old questions from being wrongfully deleted.

Additionally, there needs to be a clarification of the paradigm about old questions. What does the site do with them? Are old questions which no longer fit the guidelines of the FAQ okay? These should be answered in their own questions.

To address the OP's proposals

  1. My proposal is a detailed variant of this.
  2. See above: Often, old posts should actually be deleted. Though diamond moderators could handle it, it isn't efficient, and won't be effective.
  3. StackOverflow is the right place for StackOverflow questions.
share|improve this answer
2  
Until about a year ago, very popular questions required a huge number of votes to delete, effectively making it possible only for moderators. So instead of discussing the merits of a question publicly, it was far more effective to just badger the moderators until a question was deleted. That was not a good or effective solution. –  Shog9 Jan 20 '13 at 2:20
    
@Shog9 How many votes is a huge number? As far as deletions go, 15 votes really doesn't seem like it's that much. Either way, I would argue that popular questions should stay, or at least be very difficult to delete. –  Emracool Jan 20 '13 at 2:24
    
Unlike your suggestion, it wasn't capped - so some questions required hundreds of votes. –  Shog9 Jan 20 '13 at 2:34
    
@Shog9 Well, there's the problem. A limit needs to be set - but it needs to be unreasonable to attain, except in legitimate cases. Otherwise, the information will be lost. –  Emracool Jan 20 '13 at 2:37
    
Attainable in legit cases only - well, there's the trick... Right now, the number of votes required scales from 3 to 10 based on the votes a question attracts. 10 delete votes isn't easy - only about 200 questions have been deleted with >= 10 votes in the entire history of Stack Overflow. –  Shog9 Jan 20 '13 at 2:49
    
@Shog9 And I can, with near certainty, guarantee that every single one of those questions deserved deletion. Regardless, the formulas can be scaled as seen appropriate by admins. I was unaware that previously, questions with more votes were harder to delete - and I recommend reinstating some version of that. The primary object of focus here, though, is the deletion of old questions - older questions should be harder to delete, too. –  Emracool Jan 20 '13 at 2:53
1  
Don't be so quick to say that. I've personally cast delete votes on a few of those 200 questions and I can guarantee that they did not all deserve deletion. Back then, there was no "Historical lock" which froze the page (unvotable, uneditable, undeletable) and having such questions open gave newer users a reason to ask similar ones ("broken windows"), although the site's scope had changed in the meantime and such questions were no longer allowed.. Although I do stand by my decision back then to cast a delete vote, I wouldn't do so today, now that we have other ways to deal with the problem. –  Lorem Ipsum Jan 20 '13 at 4:10
5  
The real issue here is that the question Charlie referred to was deleted by a moderator immediately after closing, thus depriving the community of the chance to review the closure (not that it happens on SO anymore...). I don't follow SO very much these days, but my impression is that they no longer insta-delete stuff unless they're totally junk (spam/NA). –  Lorem Ipsum Jan 20 '13 at 4:13
add comment

Part of the real issue here is that historically moderators could delete, but the community could then undelete that question. A few years ago they implemented the policy that any post deleted by a moderator couldn't be undeleted by the community. This has allowed a lot of good posts to go the way of the buffalo.

Deleting should only be used for really egregiously bad questions, offensive ones, and completely off-topic ones (like programming questions on Meta). Closed as duplicate questions shouldn't be deleted because they are something that can be found in the search engine, and then followed to the canonical question. Any questions with good answers shouldn't be deleted.

share|improve this answer
add comment

The point is that whether or not my social-network hypothesis is currently true, the five-votes-to-close policy is vulnerable to exploitation by cliques.

... and? Unless you have reason to believe that there is exploitation happening to the detriment of the site, what does it matter that it could?

It sounds somewhat perverse, but in general, we only fix problems when those problems are demonstrated by evidence. Not merely that it could happen, but that it is happening and that it is bad. Otherwise, we let people do whatever they do. If you believe that there is such a clique, and that this clique is doing damage to the site, then you should be able to provide evidence in support of that position.

Personally, I think there are a number of people who vote to close questions in their areas of expertise. Are they a "clique", or simply frequent followers of one or more tags who have enough rep and the willingness to close crap when they see it? Just because a number of people are doing the same thing does not mean that collusion is happening.

Given that, it would appear that the effort expended on closing old questions has essentially no beneficial effect for the community.

Not true at all. The benefit is not in removing it from the database. The benefit is sticking a big, fat, "This needs to be fixed/is not wanted here!" sign on the question.

Therefore, when someone new finds this question and considers asking one just like it, they won't be surprised when their question is closed for the same reason.

When a question I've answered, especially with a longer, informative answer, is deleted, that has the absolutely inescapable effect of destroying my work.

Now you're getting off-topic. You said, "Closing Old and Upvoted Questions is Inherently Abusive". But now you're talking about deleting such questions. Deletion and closing are not the same thing; one has a much higher threshold.

So what is it that you're trying to claim? Is closing old questions "inherently abusive"? Or is deleting them "inherently abusive"?

Increase the number of votes required to close old questions

Establish a threshold based on age and vote count beyond which questions cannot be deleted.

Absolutely not. A bad question is a bad question, whether it is new or old. There is no reason why we should prioritize a question that just so happened to have slipped through the cracks over one that was caught right away.

Establish a place to put these older questions

We already have that; it's called a historical lock. Sadly, the moderators refuse to use them by request or acclimation. We instead have to prove that the deletion of the question is contentious by having the question deleted and undeleted a bunch of times. Only then will they get involved and provide a historical lock for the question.

I notice that another person who's made similar complaints is Norman Ramsey. It's work noting that he's one of the more brilliant computer scientists in the world. Do we really want a policy that discourages people of that caliber from contributing?

Non-argument. However "brilliant" Norman Ramsey is (and I have no idea who he is, so I'll take your word for it), he doesn't get veto power over SO just because he's a good programmer. If his beliefs about what kinds of questions should and should not be allowed on SO conflict with how SO needs to work to provide a great service, then the site is better off without him.

Your assumption is that "people of that caliber" all have the same beliefs about what questions should be allowed. I don't buy that, and I'd need to see a lot more evidence before I did. We have many skilled programmers contributing here. And if you're talking about this Norman Ramsey, it seems like he still contributes here.

share|improve this answer
    
"Absolutely not. A bad question is a bad question, whether it is new or old. There is no reason why we should prioritize a question that just so happened to have slipped through the cracks over one that was caught right away." Basically, this is the core of why I think this whole argument is incorrect: those weren't bad questions. The standard has changed. –  Charlie Martin Jan 21 '13 at 1:17
    
"Non-argument. However "brilliant" Norman Ramsey is (and I have no idea who he is, so I'll take your word for it), he doesn't get veto power over SO just because he's a good programmer. If his beliefs about what kinds of questions should and should not be allowed on SO conflict with how SO needs to work to provide a great service, then the site is better off without him." Yes, absolutely. merely being a world expert in the topic is no measure of why SO might want to encourage them. –  Charlie Martin Jan 21 '13 at 1:17
    
By the way, Nicol, Google is your friend. Here: cs.tufts.edu/~nr –  Charlie Martin Jan 21 '13 at 1:42
    
@CharlieMartin: "those weren't bad questions. The standard has changed." But they are bad questions now, and now is more important than "then". You can't just say that something is excusable now because people used to get away with it. Questions which do not fit within the rules of the site as they currently exist should be closed. Period. The good ones can get a historical lock. –  Nicol Bolas Jan 21 '13 at 1:53
    
Golly I wish I had such metaphysical certainty about things. Actually, no, I don't, because I remember when I did. You can certainly assert that these are bad questions now, because you're obviously so much wiser and more skilled than we were back then; however, it still means vandalizing work that people in those benighted past ages felt, however foolishly, they were doing to serve the common good. –  Charlie Martin Jan 22 '13 at 15:42
    
@CharlieMartin: Nothing is being "vandalized" by simply closing a question. I'm not a big fan of grandfather clauses; if something you wrote is no longer acceptable, then it should be closed as with anything else that's acceptable. We should not judge questions based on some nebulous criteria that nobody except those who were members at the time can even define. The criteria we have now are the most important criteria; they define the character of the site now. –  Nicol Bolas Jan 23 '13 at 9:13
    
Ah, but we're not talking about simply closing a question, are we. I refer you to the title of the question. –  Charlie Martin Jan 24 '13 at 13:51
    
@CharlieMartin: Your proposal focuses just as much on closing a question as deleting it, as does your "solution". My point is that closing the question is not the problem. The problem is the difficulty in getting a historical lock for it. –  Nicol Bolas Jan 24 '13 at 19:47
add comment

You must log in to answer this question.

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