20

In reaction to high failure rates of questions with titles containing the word "problem", the Stack Overflow community long ago forbade these. You couldn't ask a new question with the word "problem" in its title, nor could you save an edit to an existing question with the word "problem" still there. If you tried to either of those, a red popup appeared telling you why this wasn't allowed.

Thinking that ELU might benefit from something similar, four and a half years ago I collected aggregate data on how many questions with certain title words in them were getting closed or deleted. Those ancient data showed failure rates of around 50% for candidate titles.

Since that time, the associated failure rates have exploded dramatically: the current failure rates are almost half again as bad as they were then, closer to 75% now overall, not just the 50% of five years ago.

Enough is enough!

Stack Overflow implemented their title bad on the word "problems" when the associated failure rate was a lowly 20%, a mere pittance compared to what we are now suffering.

I therefore formally propose to the ELU community that we ban from our titles certain words and phrases with demonstrated failure rates above 50%.

So that the community at large can reach consensus here via voting, please give answers that are either (1) in favor of this proposal or (2) against this proposal. Be sure to include any comments or concerns in those answers.

If community consensus is to implement this ban, we'll then decide what text we want on the red popup messages that the ban will trigger. Those are details we'll hammer out as a community under a separate post here on meta later on down the road.

Here follow the raw data, taken today, ordered using four different criteria to help you better assess the impact these questions are currently having on the health of our site and of our community. The column headers are:

  1. IN TITLE: which word or words in the question title these data are for
  2. TOTAL: how many such questions in total, irrespective of state
  3. OPEN: how many such questions have been neither closed nor deleted
  4. !OPEN: how many such questions have been closed, deleted, or both
  5. FAIL %: the percentage of how many such question are no longer open and undeleted on the site

Remember that Stack Overflow banned "problem" when it was at only 20%. We're at much greater than 50%, often 80% or more. These should go.

1. Sorted by how many were closed or deleted (!OPEN)

IN TITLE       TOTAL  OPEN !OPEN  FAIL %
-----------------------------------------
correct         9282  2946  6336   68.26%
help            1827   352  1475   80.73%
grammar         1681   361  1320   78.52%
please          1375   214  1161   84.44%
grammatically   1468   403  1065   72.55%
right           1434   534   900   62.76%
which is        1401   578   823   58.74%
which one        867   225   642   74.05%
help me          617    68   549   88.98%
grammatical      916   399   517   56.44%
wrong            893   387   506   56.66%
which one is     638   153   485   76.02%
problem          649   219   430   66.26%
correctly        600   200   400   66.67%
please help      288    19   269   93.40%
incorrect        405   186   219   54.07%
please help me   204    10   194   95.10%
ielts            161     6   155   96.27%
need help        195    48   147   75.38%
is wrong         148    55    93   62.84%
problems         173    80    93   53.76%
mla               91    15    76   83.52%
grammer           69     1    68   98.55%
trouble          107    45    62   56.07%
the below         65    10    55   84.61%
gre               64    18    46   71.88%
your help         30     4    26   86.67%
urgent            27     4    23   85.19%
have a problem    16     5    11   68.75%
shit              12     2    10   86.67%
having trouble    15     5    10   66.67%
troubles          10     2     8   80.00%
grammarly         11     3     8   72.73%
corect             7     0     7  100.00%
grammaticaly       6     0     6  100.00%
grammatic          7     1     6   85.71%
plz help           5     0     5  100.00%
grama              3     0     3  100.00%
gramar             2     0     2  100.00%
fuck               5     5     0  100.00%
need your help    17    17     0  100.00%
plz               22    22     0  100.00%
urgently           3     3     0  100.00%

2. Sorted by failure rate (FAIL %)

IN TITLE       TOTAL  OPEN !OPEN  FAIL %
-----------------------------------------
corect             7     0     7  100.00%
grammaticaly       6     0     6  100.00%
plz help           5     0     5  100.00%
grama              3     0     3  100.00%
gramar             2     0     2  100.00%
plz               22    22     0  100.00%
need your help    17    17     0  100.00%
fuck               5     5     0  100.00%
urgently           3     3     0  100.00%
grammer           69     1    68   98.55%
ielts            161     6   155   96.27%
please help me   204    10   194   95.10%
please help      288    19   269   93.40%
help me          617    68   549   88.98%
your help         30     4    26   86.67%
shit              12     2    10   86.67%
grammatic          7     1     6   85.71%
urgent            27     4    23   85.19%
the below         65    10    55   84.61%
please          1375   214  1161   84.44%
mla               91    15    76   83.52%
help            1827   352  1475   80.73%
troubles          10     2     8   80.00%
grammar         1681   361  1320   78.52%
which one is     638   153   485   76.02%
need help        195    48   147   75.38%
which one        867   225   642   74.05%
grammarly         11     3     8   72.73%
grammatically   1468   403  1065   72.55%
gre               64    18    46   71.88%
have a problem    16     5    11   68.75%
correct         9282  2946  6336   68.26%
correctly        600   200   400   66.67%
having trouble    15     5    10   66.67%
problem          649   219   430   66.26%
is wrong         148    55    93   62.84%
right           1434   534   900   62.76%
which is        1401   578   823   58.74%
wrong            893   387   506   56.66%
grammatical      916   399   517   56.44%
trouble          107    45    62   56.07%
incorrect        405   186   219   54.07%
problems         173    80    93   53.76%

3. Sorted by total questions asked (TOTAL)

IN TITLE       TOTAL  OPEN !OPEN  FAIL %
-----------------------------------------
correct         9282  2946  6336   68.26%
help            1827   352  1475   80.73%
grammar         1681   361  1320   78.52%
grammatically   1468   403  1065   72.55%
right           1434   534   900   62.76%
which is        1401   578   823   58.74%
please          1375   214  1161   84.44%
grammatical      916   399   517   56.44%
wrong            893   387   506   56.66%
which one        867   225   642   74.05%
problem          649   219   430   66.26%
which one is     638   153   485   76.02%
help me          617    68   549   88.98%
correctly        600   200   400   66.67%
incorrect        405   186   219   54.07%
please help      288    19   269   93.40%
please help me   204    10   194   95.10%
need help        195    48   147   75.38%
problems         173    80    93   53.76%
ielts            161     6   155   96.27%
is wrong         148    55    93   62.84%
trouble          107    45    62   56.07%
mla               91    15    76   83.52%
grammer           69     1    68   98.55%
the below         65    10    55   84.61%
gre               64    18    46   71.88%
your help         30     4    26   86.67%
urgent            27     4    23   85.19%
plz               22    22     0  100.00%
need your help    17    17     0  100.00%
have a problem    16     5    11   68.75%
having trouble    15     5    10   66.67%
shit              12     2    10   86.67%
grammarly         11     3     8   72.73%
troubles          10     2     8   80.00%
corect             7     0     7  100.00%
grammatic          7     1     6   85.71%
grammaticaly       6     0     6  100.00%
fuck               5     5     0  100.00%
plz help           5     0     5  100.00%
grama              3     0     3  100.00%
urgently           3     3     0  100.00%
gramar             2     0     2  100.00%

4. Sorted alphabetically (IN TITLE)

IN TITLE       TOTAL  OPEN !OPEN  FAIL %
-----------------------------------------
corect             7     0     7  100.00%
correct         9282  2946  6336   68.26%
correctly        600   200   400   66.67%
fuck               5     5     0  100.00%
grama              3     0     3  100.00%
gramar             2     0     2  100.00%
grammar         1681   361  1320   78.52%
grammarly         11     3     8   72.73%
grammatic          7     1     6   85.71%
grammatical      916   399   517   56.44%
grammatically   1468   403  1065   72.55%
grammaticaly       6     0     6  100.00%
grammer           69     1    68   98.55%
gre               64    18    46   71.88%
have a problem    16     5    11   68.75%
having trouble    15     5    10   66.67%
help            1827   352  1475   80.73%
help me          617    68   549   88.98%
ielts            161     6   155   96.27%
incorrect        405   186   219   54.07%
is wrong         148    55    93   62.84%
mla               91    15    76   83.52%
need help        195    48   147   75.38%
need your help    17    17     0  100.00%
please          1375   214  1161   84.44%
please help      288    19   269   93.40%
please help me   204    10   194   95.10%
plz               22    22     0  100.00%
plz help           5     0     5  100.00%
problem          649   219   430   66.26%
problems         173    80    93   53.76%
right           1434   534   900   62.76%
shit              12     2    10   86.67%
the below         65    10    55   84.61%
trouble          107    45    62   56.07%
troubles          10     2     8   80.00%
urgent            27     4    23   85.19%
urgently           3     3     0  100.00%
which is        1401   578   823   58.74%
which one        867   225   642   74.05%
which one is     638   153   485   76.02%
wrong            893   387   506   56.66%
your help         30     4    26   86.67%

For comparison, overall since the site began we have a failure rate of 53.17%, with 60,707 deleted questions, 28,561 closed but not deleted questions, and 78,613 questions that are neither closed nor deleted.

  • 3
    @marcellothearcane I'm merely taking the approach taken by spam filters: something that's been often deemed spam in the past is unlikely to be deemed ham in the future. But we aren't supposed to have grossly offensive words in our post titles anyway. You can talk about them in the post body but there are all kinds of problems that come from having them in titles, potentially even getting us banned from high schools and the like. Cranking some machine-learning for this like they do with real spam filters would be optimal, but there's no mechanism for that in SE. – tchrist Sep 15 at 18:24
  • 1
    @marcellothearcane Oh good point. Let me fix that. – tchrist Sep 15 at 18:29
  • 2
    double double-plus question (haha because I keep thinking the same thing over and over).... (there's a but coming)... But -ban- the words? I feel like that will encourage simply 'word-smithing' around the banned word and the question that follows will still be horrible, the original bad word being a good proxy for bad quality later. If we ban 'ielts' the asker is just going to ask the question anyway just without 'ielts'. Is there another action that can be taken? Immediate closure and put on the reopen queue? Remote keyboard shocks? SWAT team deployment? – Mitch Sep 15 at 18:41
  • 1
    @marcellothearcane Oh drat I hand added those and flipped the columns. This is why you always need a totals line. Will fix once back from hike. – tchrist Sep 15 at 18:42
  • 1
    @Mitch that would fill up the reopen queue, or the SWAT team inbox, depending on what you went for. – marcellothearcane Sep 15 at 18:47
  • 2
    @marcellothearcane I will issue a separate proposal for removing and banning the grammar tag. It's as meaningless here as language would be. The problem this proposal aims to fix is the unending pile of offal that are useless proofreading of the "is this/which one is grammaticophobically correctissimo so my ESFLIETS teechur dont mark my test wronged" yes-no sort that will never help anyone else. Basically lazy GIMME DA CODEZ demands, not real questions. – tchrist Sep 15 at 18:50
  • 1
    @tchrist that could easily be solved by having a Stack Overflow style wizard with an option for proofreading and other undesired questions informing the user that it's off-topic and will be closed (aside from not allowing that wizard to proceed to asking when that option is selected). – JJ for Transparency and Monica Sep 15 at 19:33
  • 1
    @curiousdannii "Natural" has a 64% failure rate with 100 open and 175 closed or deleted. "Standard" is at 35% with 115 open and 61 not so. – tchrist Sep 16 at 12:01
  • 2
    @tchrist What is the overall fail rate for questions on EL&U - without this info it's difficult to analyse the figures. – Araucaria Sep 16 at 14:42
  • 1
    That really depends on the timeframe you look at but overall since the site began we have a failure rate of 53.17%, with 60,707 deleted questions, 28,561 closed but not deleted questions, and 78,613 questions that are neither closed nor deleted. – tchrist Sep 16 at 23:31
  • 2
    Great bit of analysis @tchrist. Sounds like it could save a lot of time. 😊👏🌟 – Jelila Sep 17 at 2:50
  • 5
    Please plz help me with my mla problem. I need urgent help grammatically. :D Very helpful analysis, though. I support this suggestion. – Brandon_J Sep 17 at 13:54
  • 2
    @tchrist So should we assume from that, that maybe the bottom 13 items (and maybe some others) do not have a very significantly higher than average chance of resulting in a close-vote/deletion, they just happen to occur in a lot of questions, and slightly more in questions that get closed. Some other items might be being closed for prejudicial reasons (e.g. for seeming to be related to homework). If these things were taken into account, then I would more happily back your proposal. – Araucaria Sep 19 at 14:43
  • 3
    Which of these words should I use to express to my grammar teacher that I have a problem going to the bathroom correctly and urgently need help so that I don't get in trouble? - if you can write a program that won't kick that back out, go nuts. – Mazura Sep 20 at 10:25
  • 3
    Code Review had a proposal to add a feature to warn users about low-quality questions without actually banning the question. It never went anywhere, though. – 200_success Sep 20 at 18:35
29

Yes, but not for trusted users the 'old hands'

This ban should only be implemented for new users.

I suggest that users with the Established User Close/Reopen privilege should not be limited, in the event that there is a legitimate question to be asked with any of the above words. (Thanks for the suggestion, @tchrist)

These users are already having a vote in whether such questions should exist, so should be au fait with what is on topic.

This would mean users with more than 1,000 reputation 3,000 reputation would be exempt from the ban.

There are some open questions on there, and as addressed in the comments, some words have other meanings (troubles for example).

  • 3
    Another privilege level that might be worth considering as a filter is the Close/Reopen-Vote privilege that kicks in at 3,000. I say this because such users are more familiar with our various criteria for closing questions, and thus one could hope they’d not to ask those. – tchrist Sep 15 at 21:37
  • I wouldn't know how to, but it would be interesting to see the data in the question distributed by the users' reputations. – marcellothearcane Sep 15 at 21:38
  • The data distribution by user reputation would require fancy SQL work in the employee-only version of the Stack Exchange Data Explorer. That's because the regular SEDE doesn't have access to specifics like that about deleted questions. I didn’t use SEDE to gather these data because I needed to include deleted questions (for example: deleted:all, deleted:1, deleted:0), which you can do at 10k for your own questions or as a diamond moderator for all questions. But the rest of that is only in the employee-only SEDE that only devs and CMs can use. – tchrist Sep 15 at 21:42
  • @tchrist edited, thanks :) – marcellothearcane Sep 15 at 21:43
  • 1
    @tchrist betcha it's mostly low rep though ;). And first (and only) questions. – marcellothearcane Sep 15 at 21:44
  • 8
    Truthfully, I think 1000 would be a more useful threshold than 3000. Anybody with more than a few hundred rep knows to avoid the problem, and preventing established users asking good questions because of a specific word is more damaging than helpful. – Tim Lymington supports Monica Sep 16 at 8:15
  • 2
    @TimLymington Fair enough. It would be good to see some stats on the upper bound of users asking poorly-received questions with the words in them. If this does get implemented, we could have another meta post to discuss where the cap should be. – marcellothearcane Sep 16 at 11:00
  • 1
    @TimLymington It might only take two or three posts hitting the HNQ to get 1000 rep. I don't think it's a reliable indicator of understanding community expectations. I regularly see people with 1xxx rep posting replies to poor and blatantly off-topic questions. – curiousdannii Sep 16 at 11:30
  • For what it's worth, I remember the trepidation of 1,000 whole reputation combined with not really having a clue. – marcellothearcane Sep 16 at 12:21
  • 3
    Technically, because a Community Manager will point this out, a trusted user is someone with 20K, so that term could not be used if you lower the threshold to 3K users. – Mari-Lou A Sep 18 at 9:23
  • @Mari-LouA Hmm. Not sure what we should call them instead then. Have edited though :) – marcellothearcane Sep 18 at 10:04
  • no level of rep makes "plz" okay – WendyG Oct 17 at 15:21
22

No, we should not adopt this proposal!

Vote this answer up if you think the proposal should not be adopted. If you disagree, upvote the corresponding aye-answer.

(Feel free to add your own answer if you have improvements that would make the proposal acceptable to you.)

  • 6
    Voting should only be in agreement with, you provided a "Yes" option, so why confuse users to downvote it, when upvoting this action actually means NO? – Mari-Lou A Sep 16 at 15:25
  • 2
    OK, maybe downvotes cannot be removed by mods can the downvoters retract their DVs and upvote the Yes proposal, if they haven't already done so. – Mari-Lou A Sep 18 at 9:13
  • 1
    I'd be interested to know why people believe this should not be adopted. Presumably voting Yes accepts the premise of the question as the reason for introduction -- and there are reasons for not wanting a restriction. What are the reasons? – Andrew Leach Oct 5 at 7:38
14

Yes, with proper scrutiny

There should be a new meta post in which (an updated version of) the list of banned words is posted as an answer, for everyone to see.

That post should have a separate community-wiki answer where users can propose new words to be added to and removed from the list.

The reason for this is that some words may well be part of a good title, even though it doesn't look like it now.

  • You raise a good point. I'd like to pursue that point a little. Keeping in mind that Stack Overflow implemented its titular ban :) on problem when 20% of those questions got closed or deleted, what precise metric do you exactly have in mind for us here? Just because it is possible to construct a viable question containing the words grammatically correct in its title, if we're closing most of those, why should we continue to put up with all the trouble this causes us? – tchrist Sep 15 at 20:01
  • I think this is a proposal for automatic blacklisting that stops questions frrom being asked, as in Tchrist's prior proposal in Why isn't "Please help" on our titular stoplist?, so there would be no chance to scrutinize. – Tonepoet Sep 15 at 20:03
  • 2
    @tchrist I was thinking more about other words which may have alternative meanings that are interesting subjects for questions. For example the word Troubles has another meaning as well, it's harder to find an alternative title if the question is really about that (whereas questions about grammar can be made more specific to avoid those blacklisted words). To some extent, there may be more of those words, especially when the process is automatic (which is fine, if it's transparent). – JJ for Transparency and Monica Sep 15 at 20:09
  • @Tonepoet especially when it's automated is there a benefit of being able to see what the automation does. By posting / updating the blacklist in a meta post every once in a while can we avoid needed words being taken away and undesired words being allowed. – JJ for Transparency and Monica Sep 15 at 20:10
  • It doesn't stop questions from being asked: it would force querents to refine their question titles. There is no question where "grammatically correct" should appear, because the question can always be phrased differently and more succinctly. It may be possible for the title to be edited after posting to include the banned words, if something like Troubles needs to appear. – Andrew Leach Sep 15 at 20:12
  • 2
    @AndrewLeach how would you phrase a title of a question on the etymology of Troubles as a word referring to conflicts as seen in Northern-Ireland, Mongolia, Spain and Corsica? Like I said, I agree with most of the words on the list (like those on grammar), but some may turn out to be on the list when it's not that needed. Maybe require a minimum number of bad questions (corresponding to the TOTAL column in question) before it's blacklisted. – JJ for Transparency and Monica Sep 15 at 20:32
  • @JJJ Perhaps that one could be case-sensitive. However, it's already been asked – Andrew Leach Sep 15 at 20:45
  • @AndrewLeach 'Is this sentence correct, grammatically?' That would probably pass through the filters. – marcellothearcane Sep 15 at 21:07
  • 4
    @marcello Two words there are proposed to be banned. – Andrew Leach Sep 15 at 21:18
  • 1
    @AndrewLeach oh, so any title including both 'correct' and 'grammatically' in any order and distance would be banned? I thought it was going to be 'grammatically correct' as a unit. – marcellothearcane Sep 15 at 21:23
  • 1
    @marcellothearcane I was just presenting different views of the data in case it helped get a better idea of what was going on with the different combos. – tchrist Sep 15 at 21:30
  • @tchrist The 'banned list' of words, and the 'rules' about what happens when someone attempts to use such a word in a title, can be more sophisticated than just one message coming up saying 'you can't use that'. Your list for this can logically look like: if word a, b, c - then 'action 1'. If word c, d, e - then action 2. If word f, g, h - action 3, and so on. I suggest you start to build such a 'rule table' - with a column for word groups (or single words) left, and the action on the right, then publish, for comment. D Bron started to identify such table rules in his comment on my question. – Jelila Sep 17 at 21:35
10

There are some problems with this proposal.

One is that this is a website about English Language & Usage, so every word we outright ban is going to be a word that can not be used in the title of a hypothetically legitimate question about the word. It'd be hard to talk about the origin, usage or orthography of plz if it's removed altogether. Sure, that might not seem like a large loss, but we've had serious inquiries regarding the discussion of abbreviated webslang before, such as What is the Meaning of Z.O.M.G., and if we do not already have a question regarding plz (I can't find one at the moment), then surely one has potential. Perhaps we can delete the multiple word strings without risking too much lost potential, since they seem fairly literal and hint more towards the nature of the question.

And look, I get it: I've played some online multiplayer games in the past, most relevantly including Realm of the Mad God. "Plz gimme X" is often considered a pretty frustrating sentiment for a variety of reasons I don't think need explanation.

However, this leads into the other problem which is even if we assume that questions using these sorts of questions only have marginal potential, we still need to ask ourselves if the benefits outweight the detriments, and to be honest the only benefit I am seeing is aesthetic. Otherwise, not only does banning single words as this proposed change suggests obstruct asking questions, but it also makes it harder to identify questions which should be closed for being proofreading, or too opinion based. Once the questions are closed, the better solution to make the site prettier is to simply delete them if they are utterly lacking in merit. It's more dirty work sure, but it is also work that needs to be done in accordance to broken windows theory.

Words like fuck, shit, bitch, tits, nigger, faggot, cunt and maybe damn should be added, because Stack Exchange has prohibited such words because the detriment offensive language cussing has on the network as a whole, in the form of getting the website on automated blacklists and making the network inhospitable to the broader S.E. membership, is more trouble than it is worth. It would save the community a whole bunch of hassle in editing these words out of questions.

However, if we don't even think damn should be censored in the titles because the website can't function properly or be taken seriously at that level of censorship, then what does removing innocuous words like incorrect or grammar, or even polite words such as please, say about us?

Now, I have less of a problem with blocking the multiple word phrases like "having trouble", "have a problem", "need your help" or "please help me" since those are fairly extrenuous phrases that do little to help and speak to the nature of the question, but even I doubt it would help much.

  • 4
    Just as with the CJK ban currently in place, it’s actually easy to get around any title ban PROVIDED that you know the simple trick that bypasses that ban. The goal here is very most certainly not to stop legitimate questions. The goal is to stop any newly arriving questions like the ten thousand odd currently closed or deleted questions from ever wasting our precious time with one or another variant of "Is this grammaticically correct". No more. The site is not maintainable as it now stands, and it is getting nothing but worse. – tchrist Sep 15 at 20:17
  • +1. It's a question of standard deviation - you have 95% 'bad' questions, but there's still the 5%. – marcellothearcane Sep 15 at 21:04
  • 1
    @marcellothearcane I'm pretty sure that 95% bad, 5% good is far beyond the reasonable mark. Remember Stack Overflow pulled the trigger at 20% bad and 80% good. I'm proposing a cutoff more like 50%, and I think even that is too timid. It's not like we couldn't exempt askers/editors with the Trusted User or perhaps even the Established User privilege. Plus like I said, we DO have various existing escape valves that allow the experienced or determined and resourceful asker to trivially bybass these bans. Those folks aren’t the ones getting their questions closed and deleted. – tchrist Sep 15 at 21:27
  • 2
    @tchrist oh I like the trusted user option! That's much more satisfactory. – marcellothearcane Sep 15 at 21:28
  • 5
    This isn’t censoring the words entirely, it’s banning them from the titles. This causes friction that will separate the folks that are willing to put some effort in to write a decent title from the folks that will just misspell words until they get past it or those who will just give up. It doesn’t solve the problem entirely, but it helps reduce really low quality questions. – ColleenV Sep 16 at 12:01
6

Yes, we should adopt this proposal!

Vote this answer up to say that you think the proposal has merit and should be adopted. You can vote it down to say that you don’t think so, or you can vote up the corresponding nay-answer.

You should also feel free to add your own answer if you would like to fine-tune the proposal.

  • I'm inclined towards supporting this, but I wonder whether those who are more determined than fluent-in-English would simply find (and find quite simply) loopholes to let their questions through anyway. There's something of a 'code smell' about this approach. Perhaps it would be more effective in conjunction with Matt's site-specific message proposal. – Lawrence Sep 17 at 14:18
  • I'm on board. People who think there is such a thing as a "correct sentence" should be sent to English Learners SE; they've obviously been taught badly. The supposed audience here includes linguists, lexicographers, and other language professionals who know that the notion of "correct grammar" is silly and doesn't merit attention, let alone pointing out yet again the facts. Learn to walk before trying to run. – John Lawler Oct 17 at 21:18
6

Yes, but we should commit to reviewing policy effects in a few months.

In general, I agree with the urgency of the issue. When a significant portion of questions are moderated, we should look for tools that increase the quality of questions. Yet, I see a few potential issues that we should commit to examining a few months after implementation:

  1. Convenience: Is this a nuisance to existing users, who could conceivably ask good questions with several of these terms?

  2. Effectiveness: Does the rate of closed and downvoted questions decrease? (Related questions that may be more difficult to assess quantitatively: Are users getting around the filter? Do they genuinely end up asking better questions? Do they give up?)

  3. Being Welcoming: Do we decrease the rate of upvoted questions from new users? (They may try to ask a legitimate question with the word and grow discouraged by the filter.)

Studying these three factors would ensure we're not discouraging high-effort contributions by new and existing users in aggressively filtering low-effort contributions. The method is proven in the main Stack Exchange, but they also have higher volume and a different demographic, so we should commit to being appropriately cautious. Especially with 2 and 3, I think comparisons of rates some months after implementation should be used to fine-tune the filters or take additional actions.

3

Why is it, that those words tend to be found in unsuitable questions? Does anybody know?

Is it because the person asking tupically ‘needs help or has a problem’ (that seems to be the main theme)?

...Does that then indicate questions that are ‘please help me with my homework?’ Or are from people whose literacy currently happens to be at a level which means that the question is poorly formed and can’t really be answered properly?

It would be interesting to hear, from those who often close questions, as I have rarely done that, so I don’t have a feel for it.

Rather than ‘banning words’, which I sense people feel is a bit contra to the ethos of the site, how about flagging up the title to the asker, before they post? (Similar to what happens when you try to post a 3 word comment). An error message could come up, guiding the asker to the help text and making suggestions for proper usage. It could also suggest the Learners site if that seems more appropriate.

  • 4
    It’s because the majority of these words strongly indicate the Q is off topic. “Grammar” usually indicates it’s a proof-reading Q which we don’t do; “IETLS” means they’re studying for Qs L2 English test and the question (a) is by definition covered in reference materials and (b) if asked at all, should by definition be asked on ELL; “urgently” indicates they don’t understand the nature of the site, and the Q will only ever help them and no one else (we’re not a helpdesk, we’re building a repository of knowledge useful to everyone), and so on. – Dan Bron Sep 17 at 11:41
  • I see, @Dan Bron. Well, that level of clarity about various words and likely/desired outcomes of each, is usually sufficient to derive a list of 'rules' that a computer system can be programmed to follow, when such words are encountered, provided that the resources exist for that. – Jelila Sep 17 at 14:19
  • 1
    That is exactly what Tom is proposing, given the specific set of resources we know we have at our disposal. – Dan Bron Sep 17 at 14:26
  • Mostly 'grammar' questions, or tests: "ielts" "gre" (homework), or swear words, and Qs generally written by people who don't know how to SE: thanks, urgent, please help, etc. - I find all this data very interesting, but "the website can't function properly or be taken seriously at that level of censorship". Maybe useful for kicking things into the review queue, but I assume 90% of those Qs are 1reps and those go straight onto it anyway. – Mazura Sep 20 at 9:27
  • The problem is smart phones. In 2010, about the time of this site's inception, 62.6m Americans had them. In 2019, 248.68m Americans have them. My assumption is that most of those 60m in 2010 were adults. Now I can only assume we are vastly out numbered, and when the majority of our questions are coming from ignorant children, this is what we get. – Mazura Sep 20 at 9:59
  • Add to this the tendency to use ‘what is out there on the internet’ as ‘evidence’ of ‘correct English’. Being widely used on the internet might mean that it’s ‘commonly used’ but that doesn’t mean that it is ‘correct’. – Jelila Sep 21 at 0:25
-3

No: I believe this proposal will do little to further the goals of the long term survival of the human species.

On the face of it this is horribly elitist. After reflection is it manifestly sick.

For the one proposal to allow trusted users to circumvent I think most of the people with 1-3-20k reps no longer need this site, they are here to help others so offering them tools to circumvent a novice ban is kind of pointless.

All I know is that every time another SE proposal comes along to bash bad questions on any of the stacks I loose a lot of respect and have to spend valuable time trying to justify why I provide my time to a system that is scared of 'bad' data. My proposal to all these reactions has always been the same because the basic problem does not seem to be understood.

The solution is to hide the scum from those that suffer from the righteous indignation about it.

  • Disk space is cheap.
  • People are not offended by weak, poor, sloppy questions if they are browsing SE.
  • Almost all visits to SE are as a result of a internet search.
  • Even bad data is data and even if a question only helps 2 people it was worth having.
  • Let the now rep members spend time editing, patching up and answering the bad questions and working for reputation.

So the answer is to have a flag, something like Beneath Me that makes a question invisible to fussy users. Have this flag triggered by a <99 rep user asking a question, a bad word in the title or the text, a >100 rep user setting the flag or >3 down votes. Questions with the flag set are not shown to users with rep >1000 by default and they can rest easy, all bad questions gone and site back to pristine operation.

My suggestions as above generally gain down votes but very few explanations of why it is not good policy to first be kind and avoid waste. It may be because people feel they have some personal vested interest in the site because they have volunteered for a long time and answered a lot of questions and have no desire to suffer fools gladly or otherwise. I do try to remind all the people that want to be elitist on the site that they are the product and the site is the marketing and they have no need to feel too much pride if the site also caters to others less wise than they are. The site will not crash or fall over or fail to work if it has junk in it. Google already knows how to direct people to the questions that have useful answers and there is no shame in having a good answer to a bad question. StackExchange is not paying you a salary, why are you victimising people to maintain standards that are not required.

So if anyone has a valid reason to down vote this then PLEASE share it with me as I also want to learn. Otherwise take up the crusade to stop victimising bad data for the sake of invisible appearances.

-5

I'm not sure about this.

  1. If you consider questions by failure rate then it's really just showing that questions riddled with typos stay closed.

  2. It doesn't measure at all the number one problem on this site: utterly unresearched questions. What if we prevented questions that didn't contain at least one link or block-quote from being submitted? That wouldn't solve the problems on this site, and it would still let through thousands of bad questions, but it would filter a lot of extremely poor ones.

    Recent examples that wouldn't meet that simple criteria:

  • Are you saying the 2nd question is off-topic? I don't understand, it has a decent title, it is quoting a piece of literature, it is asking about the usage of "it" in that quote. Q3 is off-topic for this site, but it's clear, well-written, and fairly interesting from a semantic point of view. – Mari-Lou A Sep 16 at 11:58
  • 1
    @Mari-LouA My point wasn't that they're off-topic, but that if you want a diagnostic tool for decent vs poor questions, including a link or quote is better IMO than excluding these words from question titles. Very very few good questions on this site do not have a quote. And if we can push people to be formatting their quotes by themselves rather than someone else from the community editing it later, all the better. – curiousdannii Sep 16 at 12:02
  • 1
    This technique not trying to detect off-topic questions and stop them, it’s putting a speed bump in that will help users with a legitimate question and desire to meet site guidelines not get lumped in with no-effort drive-by do-my-homework questions because they don’t know the impact titles can have on how their question gets received by the community. On-topicness is a different issue. – ColleenV Sep 16 at 15:56
  • 1
    @ColleenV interesting homework isn't on the list. – JJ for Transparency and Monica Sep 16 at 17:17
  • 1
    @JJJ Probably because the folks trying to get you to do their homework for them are rarely courteous enough to be transparent about it. ;) – ColleenV Sep 16 at 20:51
  • @ColleenV So what would be so wrong with adding a speedbump of needing to have a quote? How many questions here shouldn't have one? Whether it's quoting a text you're asking about, quoting a dictionary, or giving an example for a SWR, pretty much every question type on this site will use quotes. – curiousdannii Sep 16 at 23:40
  • 1
    I didn't downvote you - I think it's an interesting idea, but I think it's difficult to discern what is a relevant link and what's just there to bypass the automated system. There are plenty of on-topic worthwhile questions that may not benefit from a link. Not everything is on the internet or needs a quote block. – ColleenV Sep 16 at 23:41
  • @ColleenV Yeah that's why I suggest either a link or a quote. Any reasonable attempt of showing research or context for a question will pass, while "What does X mean?" won't. I'd like to see a good question with neither, do you have examples? – curiousdannii Sep 16 at 23:43
  • 1
    I think maybe it's something that an automated system wouldn't be good at, but human reviewers are ... Maybe extra text in the "first post" review queue to encourage reviewers to push for a citation or something. – ColleenV Sep 16 at 23:44
  • 4
    We really need the question wizard that Stack Overflow has. – curiousdannii Sep 16 at 23:46

You must log in to answer this question.

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