131

Update 2013/03/06: I've been playing around in the Data Explorer (first time I've used it) and put together a basic query to help identify typo related posts a little better, improvements welcome:

https://data.stackexchange.com/stackoverflow/query/edit/101573

select a.Id as [Post Link], a.Body
from Posts q
inner join Posts a on a.Id = q.AcceptedAnswerId -- only check accepted answers
where len(a.Body) < 200 -- posts less than 200 chars can be a giveaway
and q.ClosedDate IS NULL -- only search open questions
and (
      lower(a.Body) like '%typo%'
   or lower(a.Body) like '%spelled wrong%'
   or lower(a.Body) like '%spelled incorrectly%' -- add more...
)

This simple search yields an impressive amount of "too localized" questions related to typos by the OP:

https://stackoverflow.com/search?q=body%3Atypo+is%3Aanswer+isaccepted%3Ayes

There's been a lot of activity on this query, so I suspect much of the work left to do is in the "middle" of whichever way you sort the questions. There are several which do not qualify, and the more the bad ones get deleted, the harder it is to find more.

and these ones:

https://stackoverflow.com/search?q=body%3Aspelled+wrong+is%3Aanswer+isaccepted%3Ayes

https://stackoverflow.com/search?q=body%3Aspelt+wrong+is%3Aanswer+isaccepted%3Ayes

https://stackoverflow.com/search?q=body%3Aspelled+incorrectly+is%3Aanswer+isaccepted%3Ayes

https://stackoverflow.com/search?q=body%3Aspelt+incorrectly+is%3Aanswer+isaccepted%3Ayes

These are really easy, and almost all of them are a clear case of "too localized". I'm sure there are more variations of this query I didn't think of yet.

Most of the accepted answers are like this:

✓   You had a typo, "passwrod" should be "password".

OMG u r right!!! I feel so stupid!!! – CarelessUser May 10 '11 at 13:03

It might even be so bad where a moderator can just go in and just start deleting them outright, but I realize the mods are very busy (and as far as I understand, some mods are reluctant to get involved due to their deletes being permanent).

How can we help?

Would you all care to help me clean these questions up by closing and deleting them? Almost all of them are clear cut cases of too-localized useless questions. There are a lot, which is why I'm asking for help. I will do what I can in my spare time.

Here's one thing I just started doing that seems to help. Whenever I see a 3K+ user who has answered one of these questions, I'll post this comment:

There's a [campaign to clean up Stack Overflow](https://meta.stackexchange.com/q/167342) by removing these typo-related questions - we could really use your help! Would you mind pitching in a little by casting a close vote on this question?

So if you're reading this because of a similar comment, thank you!

If you haven't already, join us in chat!

Notes/Guidelines:

  • Obviously don't close/delete without reviewing first. The posts I'm targeting (strictly typo-related) generally take about 2 seconds to assert their uselessness which I think will make this an easy cleanup. Of course if the question/answer has other relevant content not related to a typo it might not be a candidate for removal.
  • If you have "delete" privileges - use them. This will make the process much easier once questions start disappearing.
  • Do not flag as this makes more work for mods, please simply cast close votes.
  • At the time of posting, the search returned 2,490 results.
  • Good question in the comments: Is there any way to filter closed questions from that list? The fact that the query searches for answers prevents using closed:no.
  • This may be debatable, but some questions are about clarifying possible typos in documentation and online resources (example). If you have doubt, err on the side of leaving it open in these cases. However if you find that the typo has been fixed in the original resource, go ahead and close the question as it is no longer applicable.
30
  • 30
    But what if I make the same typo? Then it's sooper useful!
    – yoozer8
    Commented Feb 12, 2013 at 23:25
  • 15
    NOTE: This question is probably not a dupe of Should questions where a problem arose from a typo be closed? That question was asking whether they should be closed. This question, building on that, is attempting to start a campaign to get them closed.
    – TRiG
    Commented Feb 12, 2013 at 23:28
  • 3
    Is there any way to filter closed questions from that list? The fact that theyre answers prevents me from using closed:no`
    – user200500
    Commented Feb 12, 2013 at 23:44
  • 1
    Oh, so your argument is that because there is a non zero number of users leaving, we have failed as a community? I thought your argument was that the number of new users we get is decreasing (because we have become more hostile). This isn't true.
    – user200500
    Commented Feb 12, 2013 at 23:51
  • This discussion has devolved. This question is a dupe. In summary: The community wants to clean up these posts, the too-localized "display" needs some love like the closed-for-dupe display received so as to better communicate what the issue is to newbies. Move on! meta.stackexchange.com/questions/123741/…
    – user133440
    Commented Feb 12, 2013 at 23:52
  • @Wesley exactly. We are assholes - some closures and comments on SO these days are just appalling, and I often don't feel comfortable sending people here any more. But you're not helping the situation by name-calling the community when it's actually looking to clean something up that no one will shed a tear for when it's gone. IMO.
    – Pekka
    Commented Feb 13, 2013 at 0:14
  • 32
    Personally, I'm happy the new users were helped out with their typoes. I answer a couple of those every week myself. But when their problem has been solved, the questions do not need to be kept around.
    – Pekka
    Commented Feb 13, 2013 at 0:23
  • 1
    Seems like we may have run out of steam on this; already some of my close votes on these questions have expired.
    – jscs
    Commented Feb 24, 2013 at 21:15
  • 1
    @JoshCaswell It would be nice if someone with insta-close/delete powers could take 10 minutes to review those. I'm not familiar with the expiration of close votes, how long do they take to expire? That seems really fast, I only posted this not even 2 weeks ago - are you sure they expire that quickly?
    – user159834
    Commented Feb 25, 2013 at 17:26
  • All close votes expire four days after the last one was placed, once the question has >= 100 views. meta.stackexchange.com/a/97581/159251
    – jscs
    Commented Feb 25, 2013 at 17:31
  • @JoshCaswell - thanks for the bounty, I forgot about this! Searching for "it was a typo" has been giving me good results. Commented Feb 25, 2013 at 17:33
  • 5
    @JoshCaswell Four days? That's way too fast IMO. No wonder the close vote review queue is eternally stuck at 50K questions... Wow, so they all expire, and does that also prevent you from close-voting again later?
    – user159834
    Commented Feb 25, 2013 at 17:33
  • 1
    @Jack While using temporary tags might be tempting, it's generally a bad idea, someone will definitely see it pop up in auto-complete and use it leading to a cascade, and then you don't know what's what all over again.
    – user50049
    Commented Feb 27, 2013 at 16:20
  • 2
    I created a shared Google Doc to help coordinate voting efforts: goo.gl/6vkPb if people feel this might be worthwhile, we could add it to the question
    – Pekka
    Commented Feb 28, 2013 at 19:24
  • 1
    Why did you remove the standard comment text for 3k+ users to pitch in?
    – Jack
    Commented Mar 1, 2013 at 9:34

8 Answers 8

32
+150

SE Chat room to help coordinate voting

Now with live backlog courtesy of Gordon, David X. Random and PeeHaa!

Syntax to point out questions that you think should be closed: [tag:cv-pls] (url)

It seems difficult to coordinate the voting on this; we seem to be needing a tool. (the Google doc is more of a crutch really)

The PHP chat room has an automatic close vote tool. It works like this (AFAIK):

  • somebody posts a link to a question in the chat and marks it [tag:cv-pls]
  • a cron job harvests the chat room frequently and adds any questions that haven't been closed yet to the backlog.

They even have browser add-ons. Holy crap!

I wonder whether this could be used for this purpose as well? In a separate chat room, of course.

I guess it would depend on how easily a copy of the entire cv-ring system could be set up that targets a different chat room ID, and uses its own backlog list. (Glancing innocuously in @Gordon's and @DaveRandom's direction)

8
  • Coordination of votes will absolutely help here. Perhaps we could all work on the same tag together, or one tag per day? Someone could announce the day's focus, and post it somewhere.
    – KatieK
    Commented Feb 28, 2013 at 19:43
  • 1
    The CV-Backlog is currently not aimed at a particular chat room. I saw no necessity for that since we are the only people using these cv-pls tags. So currently it will pick up marked links from all chats. If you guys want your own backlog, you can try installing on heroku. Limiting to a particular chat would require adding the room param to the Webpage class. However, why not just use our backlog in a joint effort?
    – Gordon
    Commented Feb 28, 2013 at 21:42
  • @Gordon So currently it will pick up marked links from all chats. hey, that's cool! A joint backlog would totally work except for one downside: people participating in this campaign can't tell apart PHP-specific closing candidates (which they don't necessarily have the expertise to closevote on). I might try getting the backlog working somewhere over the next couple of days if I get the chance
    – Pekka
    Commented Feb 28, 2013 at 21:50
  • Or they could simply not vote on them if they are unsure or disagree. That's how we do it, too.
    – Gordon
    Commented Feb 28, 2013 at 22:11
  • @Gordon yeah. Setting up a chat room chat.stackoverflow.com/rooms/25318/the-closing-room
    – Pekka
    Commented Feb 28, 2013 at 22:22
  • 2
    For the record, PeeHaa is giving us a hand with the backlog list for the chatroom. And, by the way, you got another close votes to spare on this campaign from me!
    – Alexander
    Commented Feb 28, 2013 at 22:53
  • 2
    @Wesley yeah! I felt the same way. Thanks to the guys from PHP chat who built the tools for this
    – Pekka
    Commented Mar 1, 2013 at 10:50
  • 2
    A few monkey patches to the cv-ring system are scheduled for this weekend/at the earliest possible opportunity. The browser extension auto-updates, so if you want to get started now new features will be made available at the earliest possible opportunity. Issue tracker here. Commented Mar 1, 2013 at 16:38
47

This is good.

In order to not add to the moderator woes of drowning, not waving, if you're at a level only to flag, don't.

There are plenty of users with close vote rep that can help wash out this debris.

Getting rid of these questions will allow for better search results and an increase in quality overall.

A typo is not the same as a logic problem and seeing results turn up for the former when yours is clearly of the latter is indeed a cycle of bother.

0
21

What a great idea! I'd love to see "It was a typo" questions cleaned up. There's nothing inherently wrong with them when they're asked, of course, but once solved they can't even really help even the OP again, let alone any future readers. They are red herrings for search, a kind of problem symptom wheel of misfortune, and they don't have a lasting place on SO.

Three points that I'd like to suggest watching out for (to elaborate on your first bullet about reviewing):

Some answers in this list use the word "typo" incidentally, or to mention something that isn't the core problem: Combine two shell commands into one output in shell and Regular expression to get url parameter are the two I've just run into. I'm editing these as appropriate when I come across them to remove them from the search. (One that can't be edited is Hide the toolbar which shows class functions, which is referring to the title of a menu item in an IDE.)

I'm running across a few questions which have plausible-appearing alternative answers: https://stackoverflow.com/q/4805619 and https://stackoverflow.com/q/9612878 are examples. I don't have the domain knowledge to decide whether these answers are useful in themselves, or whether they represent valid solutions to the problem presented in the question. I'd suggest caution here; do your best to determine whether the question should maybe be kept around just as a way to preserve that answer. The answers will appear in search results, after all.

There are also two I've seen where the typo was not generated by the OP, but was from a tutorial or documentation: https://stackoverflow.com/q/11730071 and https://stackoverflow.com/q/10535330. I voted to close the second because I couldn't find the listed code on the linked page. In general, though, I'd say if the typo comes from an external source, and that source still exists, someone else may very well end up with the same question -- "Is this wrong?" -- and that seems to be a perfectly reasonable thing to have on SO.

That said, round 'em up, deputies!

A further thought: might it be a good idea to start working from the end of the list (that is, the oldest posts)? I realized that my close votes have expired partly because, in the default "relevance" sorting, newer posts seem to show up first. Whenever someone goes through your search link, some questions with existing close votes will be buried on the second or third page. There's no way that I know of, unfortunately, for the link to go to the "last" page -- only numbers seem to function.

Working from the last page sorted by "newest" is awkward, of course, but I think it might focus the power of this fully armed and op... I'm sorry -- concentrate our effort and increase its effectiveness.

Suggestions for handling this kind of coordination are welcome have not only been suggested, but implemented by the inimitable Pekka.

2
  • I think you're right about the sorting, I just don't know how to get everyone to switch gears at this point. Suggestions? Feel free to edit the original post and add a notice about the sort (I totally missed your answer edit). I've been working pretty hard on this and am running out of gas.
    – user159834
    Commented Feb 28, 2013 at 15:48
  • Other than you chaning the link in your question to stackoverflow.com/… (going to the "newest" tab, i.e.), I don't have any good ideas, @Wesley. I just took a look at the last page, and a lot of them aren't the kind of questions we're after. We're just keeping on top of the new ones that pop up at this point. Without at lot more visibility, which I don't know how to achieve except via flagging, there's nothing to do but keep chipping away for a week or so and hope other people join in. Which they seem to be doing.
    – jscs
    Commented Feb 28, 2013 at 21:07
15

I've had good luck finding typo questions using:

If we get ten to fifteen people to grind through those phrases, that should knock a few hundred out.

Added 2/27/13

  • "I forgot" (4,782 answers, this should keep people busy for a while)
5
  • 17
    1,365 answer results for missed a semicolon
    – Zelda
    Commented Feb 25, 2013 at 17:52
  • Yep, "I had a typo" is most likely the OP self-answering, but sometimes it's some unnecessary crap like "Edit: I had a typo, fixed." underneath a valid answer to a real question.
    – user159834
    Commented Feb 25, 2013 at 18:12
  • @BenBrocka You forgot the quotes. After adding it, only 24 posts are left: stackoverflow.com/search?q=%22missed+a+semicolon%22+is%3Aa
    – Rob W
    Commented Mar 2, 2013 at 15:43
  • 1
    @RobW yeah, but that prevents stemming and sentence structure, ignoring a vast amount of very relevant results in the first query.
    – Zelda
    Commented Mar 2, 2013 at 17:35
  • Other variants: "missed [a] bracket", "forgot [a] bracket", "missed [a] paren", "forgot [a] paren", etc.
    – Aaron Bertrand Staff
    Commented Apr 14, 2013 at 16:23
13

CLOSE ALL THE QUESTIONS

Image © Allie Brosch. Used by tentative permission.

4
  • 9
    I was wondering when someone was going to post that Commented Mar 1, 2013 at 14:31
  • @TriG the OP didn't come with this himself, here is the source. Commented Jun 3, 2013 at 13:50
  • Yeah, @ShaWizDowArd. I know it's a very commonly used image, and is very very rarely properly attributed back to Allie. But she does, on her FAQ page, state very clearly that she still holds copyright and expects attribution. On SE at least, if nowhere else, this should be treated properly.
    – TRiG
    Commented Jun 3, 2013 at 14:56
  • @TRiG agreed, but it's only a drop in the ocean. Commented Jun 3, 2013 at 15:00
5

I don't have enough rep yet to vote to close: I flag ones that need flagging, but I won't be bothering the mods over this. I suppose people with closevote privileges may as well go ahead.

6
  • 4
    Problems caused include: Rubbish in search results, red herrings for people looking for answers to legit problems.
    – user159834
    Commented Feb 12, 2013 at 23:27
  • Fairy 'nuff. Good point. I'll still not be doing it: it's not worth the mods' time. But those of ye who can closevote may as well go ahead.
    – TRiG
    Commented Feb 12, 2013 at 23:29
  • I'm guessing there was a ninja edit after the comment to this answer; I like the answer as it stands. Commented Feb 13, 2013 at 0:26
  • 1
    @AndrewBarber: You are right, original post mentioned that the posts don't cause any problems. I'll have to ninja-edit myself so I can remove my own downvote.
    – user159834
    Commented Feb 13, 2013 at 0:27
  • Didn't realise my edit wouldn't show in the revision history. I must have been too fast. (I put a comment in and everything.)
    – TRiG
    Commented Feb 13, 2013 at 0:56
  • @TRiG If you edit within, like 2-5 minutes (or something close to that) of posting the question/answer , your edit won't show in the revision history.
    – ɥʇǝS
    Commented Feb 28, 2013 at 1:29
3

As I was doing the review queue by filtering to "too localized" for a while (now I switched to "Off topic") I realized how many times I had to open the actual question just to see that the current best answer was actually "hey you made a typo". Having the close vote queue showing the best answer at least for questions that have at least one "too localized" flag would be very helpful in sorting these questions out.

On the other side the close vote queue is getting bigger every week...

1

There's a fine line to walk here. Do we want to drive away first-time users by deleting their question before they see an answer? Do we want to punish the people who answer such questions by taking their points away?

Has anybody quantified how much noise people trip over because this stuff is left behind? Not how much is out there, but how much it affects search results etc. I think the downvotes that such questions get is enough to push them to the bottom of the barrel.

Once you put a pitchfork in someone's hands, it's hard for them not to join the mob.