2

The original stack overflow site has some "dirty" words and if you use them you get, I think, a warning before it lets you post. I think in some cases you cannot even post a question that has certain words in it.

Considering the history site keeps getting bad questions maybe we should add a list of "warning words" that try to get a poster to revise their question before posting. For reminder here are the undesired question types:

  • Genealogy
  • Asking for reference material
  • Questions answered by a simple Google search or to be found in a Wikipedia page
  • Predicting the future based on historical trends
  • Mythology
  • Conspiracy Theories or Pseudo-science

It should be pretty easy to think up some of the words and phrases that should trigger a warning given this list.

--- UPDATE

I am posting some statistics from the article linked by Yannis (The Efficacy Of Stack Overflow's Question Title Filter) below:

actively helpful - 37% of all titles gained some detail

benign - 62% of all titles became no worse

actively harmful - 2% (3 total) posts obviously hacked around the filter

--- UPDATE

I would add that it is very easy to test a prospective filter: you just run against the database of existing questions. If all of the posts it flags were put on hold, you know it is accurate.

  • 3
    "It should be pretty easy to think up some of the words and phrases..." I doubt it would be that easy. In fact, I'm pretty sure everything we'll think up will result in a significant amount of false positives. – yannis May 9 '14 at 9:15
  • 1
    @YannisRizos Well, StackOverflow does it successfully, but you are saying History which is part of the same system and uses the same software cannot? – Tyler Durden May 9 '14 at 13:59
  • "StackOverflow does it successfully" [citation-needed] Also, if it's that easy, where's your list of trigger words and phrases? – yannis May 9 '14 at 15:46
  • I am not privy to how the SO system works. All I know is that have a keyword-based warning system and if you try to make posts with certain words you will get warned or blocked from posting. Since history.stackexchange.com appears to use the same software system, I assume the support is there for the same kind of functionality. – Tyler Durden May 9 '14 at 15:50
  • 1
    @YannisRizos Are we reading the same article? The article seems to be saying the title filter is effective. Direct quote from the end of the article: "the question title check is pretty effective..." – Tyler Durden May 9 '14 at 17:25
  • 1
    I never said it wasn't effective. I know it is. It's also very annoying to well meaning users, and extremely hard to get right. Regardless, it's efficacy on SO is irrelevant. History isn't SO. We don't have even a fraction of the data SO has available (you don't "think up" problematic words & phrases, that's a recipe for disaster). Also, we allow a bit of subjectivity in posts. That completely screws up any attempt to find any kind of usable patterns. Anyway, if you do have suggestions for problematic words or phrases, please share. – yannis May 9 '14 at 17:57
  • Is there any way we could perform an experiment to tell which of these two vigorously held opinions is most accurate? I think it is an interesting idea. – Mark C. Wallace May 11 '14 at 18:44
  • @MarkC.Wallace SE Data Explorer. – yannis May 12 '14 at 12:49
3

Filter vs autoblock

I would support a warning filter. There are questions that I wish we could autoblock, but I recognize that my preferences aren't universal, and autoblock probably isn't inviting or ultimately prudent for the site. I think that there are questions that diminish the site, and if we can detect those questions, we should ask the user to take a step back, revisit [ask] and consider whether the question is a good fit for the site. Aside: As a general principle, we (I) probably need to migrate away from pointing out flaws and towards pointing out ways to improve the question.

Empericism

@YannisRizos points out that we could use SE Data Explorer to test the filter. I strongly support decisions based on experimental evidence.

How to bell the cat?

Could we use meta to develop the filter?

Post a question in meta asking for filter terms; Anyone can post an answer containing 1. A proposed filter term 1. A short justification for why the filter term would improve the site 1. A [Data Explorer] query showing the questions that would have been warned (not blocked).

Any answer that gets more than N upvotes should be added to the filter.

Questions

  1. How tough is it to update the filter?
  2. Who can do it?
  3. If we add a term and then later realize it was a bad idea, can it be reversed?
  4. Is the filter simple terms only, or can it use regular expressions?
  5. Can we link the warning message to advice? e.g. "Questions about Sujarkama's theory of Mesionic Inference have tended to get closed quickly because we generally feel that it is unclear what you're asking. Before you post this question about Sujarkama, we suggest you review [ask] and this advice. Those two sources will help to ensure that you get answers to the question rather than withering sarcasm." (Naturally this is just a mockup answer poking fun at myself - we'd want to provide more useful content.)
1

I think that the only way that this filter idea would have a chance at success would be for a large user effort to precisely construct the parameters of the filter. I really appreciate Tyler Durden's enthusiasm for the idea, but I share Yannis's caution. History is not programming. Not by any stretch of the imagination.

What sort of questions would we exclude? 9/11 Truther questions? But, what if the question was about the historical conditions in place that allowed 9/11 Truthers to flourish? The same things could be said for a great many other conspiracies.

If the idea is that there are common threads amongst closed questions then I have to disagree. I do not think that closed questions on History share many attributes. Usually questions closed on here are closed because it is not clear what they are asking, or questions are closed because they are objectively outside the realm of the site.

I think Tyler is on the right track in terms of enthusiasm, but I think that the fact we receive a lot of bad questions, in his estimation, is more due to the general public's lack of interest in formal study/understanding of history. History is often plucked from it's context as a piecemeal solution to arguments. The selective treatment that History receives has more to do with problems surrounding History than any phrasing or constructing of questions that we could capture with a filter.

You must log in to answer this question.

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