I've been experimenting with the flagging system for the last couple of days. I have learned a few strategies, but I feel that I'm making little difference to the mountain of accumulated rubbish.

Having easily run out of my 48 flags within a few minutes tonight, I am now helpless for another 19 hours and thought I would share a couple of tips for finding likely posts to flag with the rest of you, to help keep our poor moderators busy and entertained.

  1. The simplest: Look at the first posts of new users. Apart from general scanning, if they have a negative score, they're very likely to be comments or question updates.
  2. Search: body:"same problem". This finds bucketloads of "Me too!! Any progress?" answers.
  3. Search: body:bump. Gets lots of thread-bumping non-answers.

A couple of thoughts:

  • If there was a way to search only within answers, not questions, then other searches, like 'update', to find updates posted as answers, would probably be effective.
  • If it was possible to add regex matches like start of line (^), then searches like 'bump' would be much more effective.

Please post any other ideas you have found that make finding detritus easier. There's more than enough junk out there for everyone... Anything we can do to make it easier to spot improves the editing efficiency of the collective.

  • 4
    If there's a query for non-accepted answers posted by OPs, I'd imagine it would be helpful. There are lots of answers posted by OPs who are unaware of the edit and/or comment features of the site. Mar 14, 2011 at 6:03
  • 3
    per blog.stackoverflow.com/2011/01/how-to-say-thanks-in-an-answer we actually detect a lot of these 'answers' and force the answerer to click through a EULA when we do: stackoverflow.com/questions/how-to-answer -- latest stats Answer Eulas Invoked: 20834 | Answer Eulas Ignored: 9793 Mar 14, 2011 at 6:07
  • 3
    +1 more than doubled my flag weight today thanks to these tips :)
    – Jon
    Mar 14, 2011 at 9:23
  • 29
    Well, this post explains a lot... Mar 14, 2011 at 14:20
  • 1
    +1 my flag weight went up by almost a factor of 3.14 today thanks to these tips (which is appropriate, given today is π day) Mar 14, 2011 at 20:24
  • Hopefully my ten submissions helped out, amazing how quickly you can burn through those. I'd upvote this, but need more rep on meta first. :) Mar 14, 2011 at 21:10
  • 12
    @Bill: It also explains what's happening to the alots. Mar 15, 2011 at 6:54
  • @Jason: I tried to ask for such a query, but got no answers: meta.stackexchange.com/questions/84538/…
    – Gabe
    Apr 3, 2011 at 15:07

12 Answers 12


This is more for laughs than anything else, and doesn't really answer your question, since they are more candidates for editing than flagging, but here goes:

Pro tip: if you have less than 2k rep, go edit these posts, or better yet, build a bot to edit these posts, for massive rep gain!

And for hardcore editors, if you want to somewhere to start, go append views:1000 to those search terms for the posts receiving Google visitors. These posts should receive the most attention.

  • 6
    Ah, if only we could detect "loose", though. I imagine the number of misfires greatly outweighs both its number of correct uses and most of the ones listed here.
    – Grace Note StaffMod
    Mar 14, 2011 at 14:43
  • 9
    @GraceNote I'd wish for a case-sensitive search. Then I would be able to find my personal favourites - JQuery, jquery, html, javascript, etc...
    – Yi Jiang
    Mar 14, 2011 at 14:50
  • @YiJiang +1 for the tanks!
    – Alexia
    Mar 14, 2011 at 14:57
  • 1
    Also, "definately" and "wierd" are good "possibly needs edited" indicators, along with lots of others, I'm sure. Mar 14, 2011 at 15:21
  • 9
    Pro tip: if you have less than 2k rep, go edit these posts... My only caveat would be that single typo changes on their own are not helpful - see Should tiny edits be accepted or rejected in review? Mar 14, 2011 at 16:10
  • 4
    @ire_and_curses I personally would love to see less alots here. Somehow the population just went out of control, and now we have some serious problems culling them. Seriously though, most of the time there's more than one typo/spelling error to be made here - esp. those posts with 'friendly remarks'
    – Yi Jiang
    Mar 14, 2011 at 16:14
  • 2
    @GraceNote "loosing" is a runner-up. Mar 14, 2011 at 19:10
  • 2
    Somebody should flag this answer. Should be a comment in my eyes.
    – Franz
    Mar 14, 2011 at 20:33
  • 1
    I just noticed you culling the alots. How could you be killing such beautiful creatures! Mar 15, 2011 at 6:52
  • 1
    @AndrewGrimm The conservation efforts went too well! Now our ecosystem is struggling under the load of all the alots pooping on our beautiful questions, so something must be done, quickly, before a lot of alot poop clog up the systems
    – Yi Jiang
    Mar 15, 2011 at 6:54
  • you could also add "FIXED!!!" in addition to "SOLVED!!" @YiJang Mar 15, 2011 at 9:34
  • While we're at it, can someone cull thankyou in instances where the sentence isn't referring to a thankyou card.
    – amelvin
    Mar 15, 2011 at 9:59
  • Is there any way to find "too localized" questions?
    – Daniel
    Oct 11, 2011 at 21:44
  • After having gone on editing sprees on some sites where I wanted to build up reputation for a bot account, I would recommend two things: prefer editing relatively new questions over old ones, and try to keep the volume of edits down on low-traffic sites. In chat, @rene suggests examining the site's front page to get a sense of the recent average number of new questions per day, and staying significantly below that number, so as to avoid littering the site's front page with old content.
    – tripleee
    Jun 14, 2021 at 6:42
  • An alternative, in some sense less disruptive approach is to propose new tag descriptions, but that's obviously more work, and requires knowledge of the related domains.
    – tripleee
    Jun 14, 2021 at 6:43

For the search results to contain only answers, add isaccepted:0, which restricts the results to non-accepted answer posts, i.e:

isaccepted:0 +body:"same problem"

Here's two more I've found useful:

isaccepted:0 +body:"help me"
isaccepted:0 +body:"plz"

  • 12
    The 'help me' search was a gold mine for non-answers. I flagged 51 of them.
    – jjnguy
    Mar 14, 2011 at 15:40
  • We might as well auto-delete everything @yhw42 flags ;-)
    – Ivo Flipse
    Mar 14, 2011 at 20:56
  • Don't assume I'm infallible @Ivo! I had one of my flags dismissed as invalid this weekend! :)
    – yhw42
    Mar 14, 2011 at 21:07
  • Wow @yhw42, a whole flag out of probably a 100! :P You had nearly 750, before they tweaked the algorithm and are already up to 621, I think I'll take the risk of you making a mistake
    – Ivo Flipse
    Mar 14, 2011 at 21:12
  • yhw is close to 650 now :P @Ivo. Mar 15, 2011 at 7:44
  • 4
    Heh, I wonder what the highest flag weights are.
    – moinudin
    Mar 16, 2011 at 7:36
  • @jjnguy: When I posted this, the "help me" query returned 3103 results. Currently (about 4 days later) it returns 2169. :)
    – yhw42
    Mar 18, 2011 at 17:24
  • I have to say, these are the nasty parts of the site. "Plz send me teh codes", asking for email addresses, "please read my blog entry over here then email me the answer" ... shudder. I don't normally come across that in "my" tags. Hope I helped a bit the last few days. Mar 19, 2011 at 13:42
  • 2
    Also "desperate" and "bump". Sep 24, 2012 at 0:09

Another useful query for non-answers is body:anyone isaccepted:0:



In addition to "any idea":


which had a good number of non-answers (before everyone cleaned those out), another trigger phrase is "thanks in advance":


Aside from when people are being quoted, the latter phrase really has no place in an answer.


Random has provided a query (written by sysadmin1138) that finds answers that have a comment that includes "not an answer" at https://data.stackexchange.com/stackoverflow/s/1040/finding-bad-answers-through-comments. I would like one for "should be a comment" also. Many of the answers I am flagging today have such comments already, which were never acted on.

Edit: Having used up most of the "not an answer" and "should be a comment" I am now working through https://stackoverflow.com/search?tab=relevance&q=isaccepted%3a0%20%2burgent%22

Edit2: in addition to urgent, it seems that guys, cheers, and worked are popular in non-answers. Also dear (often addressing the OP asking for teh codez), any luck and any one (just plain anyone is listed in another answer already.)

Came across one that puzzled me and when I searched, it has lots of siblings: https://stackoverflow.com/search?q=%22does+not+meet+our+quality+standards%22 Unbelievable. They get around the quality filter by pasting in the actual error message.

  • 3
    +1, except this is not an answer and should be a comment... :) Mar 17, 2011 at 4:06
  • :-) @Zach you're right but I have a plan about that. Mar 17, 2011 at 12:06
  • It was written by sysadmin1138 but if you tweak the query you can get all sorts of other comments to come back
    – random
    Mar 17, 2011 at 12:44
  • Whew... 40 flags gone Mar 17, 2011 at 13:00
  • 3
    @Kate: The comments and edits that have been accompanying your flags on SO lately have been enormously helpful. Thank you! Apr 10, 2011 at 16:22
  • @Bill thanks for letting me know, much appreciated. Apr 10, 2011 at 18:02

Late Answers usually have a lot of "This doesn't work for me" and "Thanks this works" answers.

Can find some spam in there also sometimes.


Reputation-Related Commentary

Follow-Up "Questions" and Duplicates


  • The System seems to automatically detect many of these and consequently put the post into the Very Low Quality review queue.
    – Raedwald
    Jan 2, 2019 at 17:06

I was able to find quite a few using is:answer body:"one more question".

This is often an answer that should be a new question, especially if it also contains "thanks" or similar.

I also found quite a lot of should-be comments/close votes using is:answer body:"this is a duplicate", most of them only containing a link to the duplicate question. is:answer body:"duplicate" url:"*stackoverflow.com/*" also works quite well.

  • 1
    Wow. I can't believe that the "duplicate" searches turn up nearly three thousand results! Aug 23, 2011 at 19:15
  • 1
    @Chris: What strikes me as even more bizarre is that a few of them had upvotes. One even had 14 upvotes for just the single word "duplicate".
    – hammar
    Aug 23, 2011 at 19:29

isaccepted:0 +body:"how can i" helped me find plenty.

Likewise did isaccepted:0 +body:"thanks in advance"


One search that seems to get a lot of posts that just link to other questions without any additional info is

is:answer url:"*stackoverflow.com*"


These searches mostly find old bad answers, so they are great for cleaning up the site, but within a few days enough people will have used them and flag the worse answers they show.

The danger is that lot of people will start from the first result page, and therefore the first few pages will get lots of flags, then everyone will run out of flags – so it would be better if the search results come back in random order.

(I hope the mods that cope with the flag storm this may create….)

  • 10
    The results on the review pages are randomized, so tip 1 is already ok with this.
    – ho1
    Mar 14, 2011 at 11:19
  • 3
    this should be a comment
    – Lesmana
    Apr 13, 2011 at 14:28
  • When I use some of these searches and sort by newest first, I get very recent posts.
    – Esailija
    Jul 14, 2012 at 10:24

I am getting a lot from


As I will not have time over the next week or so to check the other 300 pages of results, someone else can get the flag weight for these. I have been finding about 3 or 4 none answers per page of results.

(I got up to page 19, don't all start looking at page 20! choose a ramdom page to start from)


One more: I would like to. Now 4795 answers. 3738 are not accepted.

(99 are "I would like to recommend", which seems legitimate. There might some other good ones in them, too.)

  • Ooooh! That is such a clever idea! Very good find, I am implementing with great success. Sep 1, 2011 at 15:46

You must log in to answer this question.

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