Most new users who spam old questions do not check to see if the answer has been accepted by the user, or how old it is. Preventing 1-rep users from answering old questions where the OP has clearly stated that the question has already been satisfactorily answered seems like a logical step.

The protection can be silent; i.e. the question doesn't necessarily have to have the protection banner on it. I would be OK if the protection didn't kick in until a period of time has elapsed, e.g. the question has had an accepted answer for more than 30 days.

  • 3
    What do you mean by "spam old questions"? Do you actually mean spam, as in, irrelevant commercial messages? Or do you mean "attempt to answer"? May 25, 2012 at 1:59
  • I mean anything that fulfills en.wikipedia.org/wiki/Spam_(electronic), or satisfies our definition for link-only answers that clearly just want to promote their blog or service.
    – user102937
    May 25, 2012 at 2:03
  • 20
    Wouldn't there be collateral damage in cases where a new user genuinely comes across a question where the accepted answer is out of date or inadequate? May 25, 2012 at 2:06
  • Then put the banner on it and let them flag for unprotection, or lower the threshold for auto-protection by auto-deleting the first 1-rep post that gets a downvote or spam flag and auto-protecting then. Three 1-rep deleted posts is a pretty high bar for autoprotection on an old question that already has an accepted answer.
    – user102937
    May 25, 2012 at 2:08
  • 8
    I'd rather address spam by detecting spam content than by using unrelated heuristics. If an answer is spammy there should be ways of noticing it (e.g. a bayesian algorithm, the lack of keywords in common with the question, IP addresses that are unusual for good content, etc. etc.) that do not create collateral damage... May 25, 2012 at 2:16
  • Sure, If you think you can find a Bayesian algorithm that is reliable enough. It took the email industry a decade to come up with reasonably good ones.
    – user102937
    May 25, 2012 at 2:19
  • 3
    We haven't even remotely begun to try. It's a high priority for the dev team over the next 6 months. May 25, 2012 at 2:28
  • As for flagging by low rep users: I don't see flag links when not logged in. (Which makes sense, I guess.) It's also not listed on privileges, so I guess any registered user can flag?
    – Arjan
    May 25, 2012 at 3:00
  • @Arjan: stackoverflow.com/privileges/flag-posts
    – Dennis
    May 25, 2012 at 3:01
  • 2
    Ah, so flagging requires 15 rep, hence flagging for unprotecting unfortunately won't help. (Wow, @Dennis, I feel stupid. It's 6th from the bottom of the list I linked to; I guess I expected it to be lower when looking!)
    – Arjan
    May 25, 2012 at 3:04
  • 11
    @JoelSpolsky Buy something off the shelf, integrate it, and be done with it. At first blush, it would seem the SE sites are not drowning in spam. Six months of dev team priority for something that isn't currently a perceived problem is disturbing.
    – casperOne
    May 25, 2012 at 11:49
  • 1
    Robert is there a way to analyze historical data to estimate potential gain? Like: 1) choose a time frame, say Oct through Dec 2011 2) for given time frame, 2a) find total amount of 1-rep answers to "eligible" questions, 2b) find amount of 1a kind answers that were spam, 2c) find total amount of all spam answers. Value of 2a:2b:2c like 10:9:9 would mean your idea is spot on, one like 10:9:999, well, would likely mean opposite etc. Side note - "protection" you describe sounds more like a privilege, like remove new user restrictions
    – gnat
    May 25, 2012 at 17:11
  • @gnat: Protection is opt-in for questions, because the only practical way for new users to gain rep to earn new privileges is by answering questions.
    – user102937
    May 25, 2012 at 17:13
  • 1
    @joel is right to argue that the collateral damage of "protecting" so many questions is way more harmful than a handful of spam. And anyway, our communities love nothing more than to attack any whiff of spam with the fury of a thousand suns. Spam is just not a serious problem on Stack Exchange compared to quality. Jun 16, 2012 at 8:17

5 Answers 5


Before I joined Stack Overflow, the site was something that I just thought was just another resource on the Internet where I could come to find solutions to problems I was facing, solutions which were asked and answered by others.

I didn't join the community, until one day I saw an answer posted about Comet that helped me solve a problem I was searching for days to solve. Someone downvoted the answer, left a snide comment, and there it sat with a -1 score.

I really wanted to leave a comment defending the answer, and I really wanted to tip the scales and upvote the answer, but the Stack Overflow system denied both of my attempts to right this wrong.

Determined to fix this problem, I quickly started looking for a question or two that I could answer in order to earn the reputation score needed so I could leave my comment and upvote the answer. I had my first 20 reputation points in just a few minutes, left my comment, upvoted the answer, and then became addicted to answering questions on Stack Overflow.

In summary, if someone really really wants to do something on Stack Overflow bad enough, such as providing a great answer to an old, outdated, protected question, and that person is really passionate about the subject matter, then he or she will likely go get those first few upvotes and gain the necessary privileges to provide an answer to the question.

Now, I'm not sure if protecting these accepted questions is indeed the right answer, but I am almost certain that the 10 or 20 rep user that answers those first few questions didn't spend 10 minutes of their time gaining reputation, just so they can post spam on a protected question.

I hope this helps Stack Exchange make a good decision on whether or not to implement this.

  • 3
    I am sure this happened with all of us, start using the site, then feel like you owe something to the community, then... addiction
    – ajax333221
    May 25, 2012 at 4:05
  • 1
    Mostly "...addiction" May 25, 2012 at 6:05
  • Addiction, then feeling like I owed something to the community, then more addiction, more payback... It's a vicious cycle... like gambling.
    – jmort253
    May 25, 2012 at 6:08
  • 3
    You would reserve Stack Overflow to addicts. I strongly disagree. I want to have drive-by experts as well, who contribute the odd pearl of wisdom but don't care to stick around. The proposed restriction would not drive you out, but it would prevent a lot of viable contributions. May 25, 2012 at 18:38
  • 1
    To give an example: my first post on SO was to a 2-year old question with a long-accepted answer, where none of the existing answers had worked for me. I quickly took on, but I'm not sure that I would have stuck if I'd been told “this question is too old, you may not answer it”. I've always been more attracted to the Wikipedia quadrant of the site — building a repository of knowledge (but without WP's notability and no-original-research requirements). This isn't so important if you're attracted to the Reddit quadrant, but keep in mind that Reddit isn't what drives the best content. May 25, 2012 at 19:00
  • @Gilles - I do think there are better ways to solve this problem than protecting all of the questions. This is but one of many possible solutions. Before the example I cited, there were other contributions I wanted to make but didn't simply because I had no privileges or account. In my opinion, 3k or 10k users need to be brought into this process. So far community moderation (not diamond moderation) is working out quite well to close off-topic and NARQ questions, so why not try the experiment with "not an answer" flags too? I see Robert Harvey's proposal as a last resort, not the first step.
    – jmort253
    May 25, 2012 at 19:27
  • @jmort253 NaA flags reach 20k users. They can do something about it (viz, delete the post) if the post doesn't start with a high score. May 25, 2012 at 19:54

You seem to believe that an answer being accepted has some sort of special status. I cannot disagree more!

An answer being accepted doesn't mean anything. Any answer can be accepted by the OP for any reason he sees fit, or no reason at all. It doesn't mean that it's correct or that it's good, and especially it doesn't mean that it's perfect.

Additionally you seem to imply that if the OP received an answer he likely believed satisfactory then the matter is "settled". Again, I cannot disagree more!

The question is not there only to help the OP. In fact, it's not even mainly there to help the OP: it's there to help the internet.

Hundreds, thousands maybe millions of people having the same problem will run into that question and at that point what the OP originally found helpful quickly becomes irrelevant.

  • Then suggest an alternate metric, if you don't like the one I proposed.
    – user102937
    May 25, 2012 at 16:35
  • 1
    @Robert: my post was kind a tangent but honestly I haven't quite understood what's the actual problem you are describing that needs fixing. I don't see a lot of spam.. in fact i don't see any spam. But even if there was a problem, using the "accepted answer" as a discriminator between questions that can be answered by new users and questions that cannot makes no sense May 25, 2012 at 16:37
  • You don't see any overt spam because the community and the mods work very hard to keep it off the site. The larger category of problem answers are link-only no-effort answers and non-answers, posted by 1-rep users.
    – user102937
    May 25, 2012 at 16:43
  • 1
    @Robert: so the real problem is not having spam on the website, but the fact that it takes too much time to clean it up? This starts making more sense. But the part about using accepted answers as discriminators still doesn't. May 25, 2012 at 16:53
  • 1
    @RobertHarvey I have 10k on SO, have spent time in /review. I do see some of the spam. And I conclude that it is not enough of a problem that an alternate metric is necessary: anything close to what you suggest would be worse than the statu quo. May 25, 2012 at 18:40

I strongly disagree with this policy as proposed. More specifically, I disagree with forbidding the scenario where the world expert on a narrow topic finds a question on Stack Exchange, which he had never participated in before, and decides to spend 5 minutes to share his wisdom.

My position applies to Stack Overflow as well as the other, smaller sites.

I've seen spammish answers on new questions as well as old questions. If you include link-only answers that aren't specifically promoting the poster's blog, they're far more of a plague on new questions. As an experiment, I just went through 30 of today's late answers in /review. The result:

  • 3 not-an-answer flags
  • 1 link-only answer to a question that called for it
  • 2 answers that weren't much more than links to code, where I left a comment asking to include the essential parts of the code in the answer
  • several more answers that looked like the answerer had tried all the prior answers (including a couple where the answerer explicitly wrote “I tried the existing answers and none worked”), but I did not know enough about the topic to lend my support by upvoting

That's a 1 in 10 definite bad answer rate, 1 in 4 bad-but-salvageable rate. This isn't markedly worse than new answers on new questions. It does not justify forbidding the drive-by experts from answering. I could only countenance this if there was a clear majority of bad apples, and the balance is clearly in the other direction.

If you really think there should be a restriction for new users, a longer minimum length is the only potentially viable idea I had before my visit to the late answer tab today. After my visit, I've decided that the length or the presence of a link isn't really telling: the me-too/follow-up-question posts that I flagged shared the 100–200 characters range with some valid-looking questions. So I do not propose any restriction.

And to address the unsaid subtext that Stack Overflow moderators have too much work: the solution is better tools for the community to perform clean-up tasks. Give us more delete votes. Give us tag sorting in the flag list and the suggested edit list. Give us a /review/delete tab. Give us migration paths with target community acceptance.


I think this idea comes with good intent, however I think it will prevent having better answers on old questions. I have answered a question in the past that had accepted answers that were kind of bad (example). I came to that question after doing a Google search on the subject, and it would have been bad if you couldn't add a much better answer.

What I think would be better is to have that kind of protection, but for questions that have a lot of answers (at least 5 or 10). On these questions, it's usually safe to say that the subject is fully covered by the answers.

  • 1
    How tough is it to get 10 rep, so that you can qualify to answer any protected question?
    – user102937
    May 25, 2012 at 2:17
  • @RobertHarvey People that comes from Google search don't necessarily have account here. They might want to add an answer to a question which has bad or average answer. If they can't answer it, they will just leave that question as is and never answer it. This create a lose-lose situation, it's quite frustrating not to be able to answer and we lost an opportunity to have better answer.
    – HoLyVieR
    May 25, 2012 at 2:20
  • 2
    You make a fair point, but I personally don't believe random people coming from Google should be slapping answers on old, satisfactorily resolved questions unless they actually have an account.
    – user102937
    May 25, 2012 at 2:23
  • 6
    It's not tough, but why would you bother? You're doing us a favor by helping us out with some of your knowledge, and we're making you jump through hoops. I'd like to see some actual stats on how many late answers by unanswered users are valuable vs. late answers that are spam (should be pretty easy to figure out from the data) May 25, 2012 at 2:23
  • 3
    I look forward to seeing that analysis. :)
    – user102937
    May 25, 2012 at 2:25
  • @RobertHarvey - Was wondering how many of these you deal with? Are there enough 3k+ and 10k+ users to effectively deal with these types of situations? I feel like bad content doesn't sit on SO for too long, even without diamond mod intervention. The community is pretty good about self-policing IMHO.
    – jmort253
    May 25, 2012 at 3:33
  • 1
    @jmort253: The vast majority of these errant posts are "Not an Answer," and they probably comprise one fourth or more of the 1000 or so flags we process each day. Of these, we could probably eliminate about 75% of them if we simply disallowed 1-rep people from posting to questions that are more than six months old.
    – user102937
    May 25, 2012 at 3:38
  • @RobertHarvey: That's quite a bit. Apart from the fact that non-moderators cannot convert answers to comments, is there any reason why not an answer can only be handled by moderators? If there isn't, maybe a 25k convert-answer-to-comment privilege could help...
    – Dennis
    May 25, 2012 at 3:56
  • Then maybe that would be a better solution. Can we give 10k users the ability to convert answers to comments? If not 10k, 20k or 30k perhaps? I see where Joel is coming from with not wanting to put up barriers, but I'm sure SO moderators are slammed with the constant flow of traffic and could use some extra community support.
    – jmort253
    May 25, 2012 at 4:00
  • @Dennis "trusted users" (i.e. 20k rep) can vote to delete any answer at -1 or lower, so they can downvote then vote to delete. Takes 3 such votes, I think. May 25, 2012 at 4:41
  • 1
    @jmort253: Most "Not an Answer" posts don't get converted to comments by moderators. Doing so defeats the purpose of requiring 50 rep for the privilege of posting comments.
    – user102937
    May 25, 2012 at 5:58
  • @RobertHarvey - Ok, makes sense, but instead of protecting questions, would it help if the threshold to vote to delete was lowered? I'm not quite 10k yet (will be this weekend though) but I feel like I have a good grasp on delete vs don't delete.
    – jmort253
    May 25, 2012 at 6:02
  • @jmort253: I don't know if that would help or not. There are already about 2400 users on Stack Overflow who have vote to delete privileges. stackoverflow.com/…
    – user102937
    May 25, 2012 at 6:08
  • 1
    @JoelSpolsky: I don't have numbers, but I have asked a lot of questions on SO that are over a year old, and a very high acceptance rate. Nearly all the new answers I see on old questions are from 1-rep users and don't add anything to the page. The compromise of only auto-protecting if there are 5 or more answers seems reasonable to me.
    – Kip
    May 25, 2012 at 16:25
  • 2
    @JoelSpolsky I just did an experiment: 30 posts from /review/late-answers. Not a statistically-significant sample, especially as the dregs of the dregs might have been deleted already, but my conclusion is that adding hoops would be detrimental to the site. May 25, 2012 at 18:43

I am in favor of this.

I asked a lot of questions on Stack Overflow at the beginning. So it is not uncommon for me to log in and see new answers on my old questions. They are usually from 1-rep users, and it is extremely rare that they actually add anything. Most of the time they are just a repeat of things in other answers. I usually protect these questions. (I wouldn't be surprised if I have one of the highest protect-vote:other-vote ratios for non-moderators on Stack Overflow.)

You must log in to answer this question.