I have noticed this in hundreds of questions.

Most of the users use invalid lower case letters. "i" instead of "I" and "how to do..." instead of "How to.." Other than that I can see a lot of abbreviated words like "plz", "thx", etc.

However, they are not nice to see on a Q&A site. And these are the things that can be corrected by a bot (or even when submitting the question).

I just wanted to add this as a new idea.


4 Answers 4


That's what the segment of the community over 2K rep is for. It's all a collaborative effort and everyone pitches in to clean up, whether it be with fixing up formatting or even going through and cleaning up the grammar and english.

The incentive for this is the Strunk and White badge and feeling a sense of relief and a calming of the night terrors as lowercase usage of the pronoun I haunts you leaving the soles of your feet pitch black and an empty bird cage next to your pillow.

  • 1
    It gets a bit boring after a while, doesn't it?
    – alex
    Commented Sep 23, 2009 at 9:22
  • 3
    Depends on if you have Paint open.
    – random
    Commented Sep 23, 2009 at 9:22
  • 3
    You, sir, seem to have mastered the fine art on creating masterpieces in Paint :)
    – alex
    Commented Sep 23, 2009 at 9:44

I think some automated cleanup would be beneficial, although it would have to be carefully considered. Wikipedia uses bots quite successfully in this way. Despite what users on meta claim, manual community cleanup is not sufficient to keep up with the incorrect grammar and typos. Take a look through any random selection of questions on Stack Overflow. You'll find them full of errors, even ones with lots of votes and answers.

I think the truth is most people just don't really care, as long as the intent of the question can be discerned well enough to provide some kind of answer. I don't have any hard evidence, but I suspect the vast majority of 2K+ users don't edit much at all. It's easier to just vote to close. Personally, I edit every now and again if I like a question, but it's pretty boring and completely thankless work.

In a related way, I also think a lot of tag cleanup tasks could be automated (I think some already are), but that wasn't a popular opinion.

One thing that does surprise me is how ambivalent/negative the reaction seems to be to this sort of proposal. On a website dominated by people whose job it is to automate things, I find the answer 'we have people to do that manually' difficult to understand.

I'd definitely appreciate some insight from those who prefer the other side of this particular fence.

  • 2
    You can't really automate intent or sometimes, context. What if they need to include such words as "kthxbai" but the autocorrectabot transforms and rolls it out into something that doesn't make sense any more? Or invalidates/nulls out the question?
    – random
    Commented Sep 23, 2009 at 10:39
  • Just wait until the sentient brother of Sneakers O'Toole is rolled out to the public site.
    – perbert
    Commented Sep 23, 2009 at 19:59

There are patterns that can be used in search of editable content, such as:

lexic misconceptions

"founded" (incorrect unless you are talking about foundations).

Misconstructed syntax

"Ho do I can...?"
"Do you can...?"

These ones came to my mind days ago and some search results show hundreds of pages with unexperienced writings in titles, questions and answers in StackSites that mix different writing contexts (code and explanations). The list can be endless and goes far away from common slang. Described and similar patterns are useful to find editable content and some of such patterns could make a topic hard to find. Anyone can use this to give a hand in cleaning and earn some reputation in exchange. However results and posibilities lead to another concern: set aside editable content and focus only on Editworthy content. Most of results are poor quality entries that go beyond grammar: they can be Old, obsolete and lazy writings for example. Edition, queuing, revision and approval of such stuff lead to its revival on main page, requesting unnecessary attention in present time and then requiring moderation, involved reanimators advise and voted deletion; rather than actions regarding site main drives: New Questions and answers.

These patterns are also the easy ones. Remembering my first School MS Word homeworks I recall long sentences that were hard to conceive their suggested corrections. Long patterns use to fall in semantic issues (and even far away, streaking the unclear what you are saying area), but their appearance is still signal of unexperienced writing, and its search and automated analysis is still a subject hard to conceive and load into cpu memory in a worthy way.

If there's already an autocorrect bot, it will then need to update its directives and do its duty silently, at least with short and easy (typos, misconceptions, slang) patterns. All these may become a subject of study or community filled data to make some more automated usefull cleansing.

Last but not least to say: For recommended and non automated (member side) help, editworthy entries migth be searched and found using unexperienced traces along with good quality potential or actual presence in despite of its grammar (upvotes or fluent/rich explanations/feedback/activity).


I am against automated post-processing of messages, but e.g. Firefox already comes with a built-in spell-checker. So we should try to find a way in motivating people for using it instead of risking automated "corrections" which could slightly change the meaning of a post.

You must log in to answer this question.

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