27
$\begingroup$

Almost $90$% of the questions I have seen from the time I have joined Maths.SE, I have noticed that the user NormalHuman suggests title edits, tag edits and question modifications almost instantly after the question pops up on the New Questions Tab and the language of the comments are almost the same.

So my question is: Does he do this separately for all the questions because it is surely very tiresome? Or does he use some kind of an alarm that posts this message to anyone asking a new question?

I want to know the trick for I want to use it too.

$\endgroup$
  • 20
    $\begingroup$ He's running a bot to do that. $\endgroup$ – Daniel Fischer Oct 27 '15 at 10:11
  • 15
    $\begingroup$ He is a very fast typer. $\endgroup$ – Najib Idrissi Oct 27 '15 at 10:40
  • 20
    $\begingroup$ In reality there are no normal humans. It's a ruse. It's a trap. It's a reverse Turing test. The only person to claim they are a "normal human" is in fact the internet incarnate. The ghost in the machine. $\endgroup$ – Asaf Karagila Oct 27 '15 at 10:59
  • 15
    $\begingroup$ Wow, people will complain about anything... $\endgroup$ – Najib Idrissi Oct 27 '15 at 17:52
  • 23
    $\begingroup$ @HenningMakholm The current wording is "Consider adding a tag for a broader subject area to which the question belongs. This will improve the visibility of your question." If you think this comes across as rude, can you suggest a more polite version? Btw I have made over 5 times as many edits as you had. $\endgroup$ – user147263 Oct 27 '15 at 18:02
  • 18
    $\begingroup$ @NormalHuman: If you know a better tag, you should edit it into the post instead of just teasing the poor newbie with, effectively, "there's a better tag for this post, but I'm not telling it what it is. Nyah nyah. Figure it out yourself". I don't think there's any way to make that message less rude by changing the wording -- the way to avoid being rude would be to cease the rude behavior, not to dress it up with irrelevant pleasantries. $\endgroup$ – Henning Makholm Oct 27 '15 at 18:04
  • 14
    $\begingroup$ Also, I fail to see how having made 5 times as many edits as I have justifies being rude to newbies. Why should having made a lot of edits entitle someone to that behavior? And even if it should, why would my edit count somehow be related to the threshold of being-allowed-to-bite-the-newbies? $\endgroup$ – Henning Makholm Oct 27 '15 at 18:08
  • 15
    $\begingroup$ @HenningMakholm If we had reason to believe the purpose of the bot is "telling new users off", we would disapprove of that, and make it known to the user running the bot. But if the purpose is to provide some guidance to new (and not-so-new) users, we approve of the purpose. How the implementation will be judged is a different matter, we shall see. $\endgroup$ – Daniel Fischer Oct 27 '15 at 19:05
  • 12
    $\begingroup$ @HenningMakholm : I remembered adding a new tags to a question, but the OP rollback the edit, and leave a comment like "It is rude to change my question without my approval". In some cases, editing the post might be considered more rude to some people. $\endgroup$ – user99914 Oct 27 '15 at 22:45
  • 17
    $\begingroup$ @HenningMakholm Empirically, someone tagging their question with "diophantine-equations" is capable of recognizing that "a broader subject area" is number theory, and adding a tag. I get a bunch of "thanks, fixed" responses every day from bot's operation, and this doesn't include the people who edit their question silently... Those who can't think of a tag tend to say so in their comment-reply -- at least they have someone to ping, because of that automatic comment. I get those replies and fix tags if still needed. $\endgroup$ – user147263 Oct 28 '15 at 12:01
  • 11
    $\begingroup$ Normal Human is normal in that he is imposing a norm $\endgroup$ – Omnomnomnom Oct 30 '15 at 16:18
  • 11
    $\begingroup$ @IanMiller I suppose your latest comment was in fact a reply to me. If the only problem is that somebody might not get it is auto-generated, I think I do not see the issue. As Normal Human remarked it is not the only comment that it not formulated by the user posting it. Some such functionality is even built into the site, and some more is supported since ages here on meta. And again there are in my mind plenty of hand-crafted(?) comments that are harder to get and more pushy. For instance the "What have you tried?" and variations on it. Frankly, suggesting extra tags is really a service. $\endgroup$ – quid Nov 2 '15 at 14:16
  • 8
    $\begingroup$ @AndréNicolas How is that, in any way, even remotely, "hacking"? This whole thing is starting to read like politics where party A would still manage to find a way to criticize party B if they somehow managed to cure cancer (because you know, oncologists would lose their jobs etc). Anything the "opposition" does is necessarily bad. $\endgroup$ – Najib Idrissi Nov 2 '15 at 16:37
  • 10
    $\begingroup$ And anyone who believes that running scripts using the (public!) SE API is somehow discouraged or forbidden is invited to take a look at stackapps. Please, everyone, take a step back: if SE displayed a warning (not an error) when a user used only one rare tag, or used one of several "bad" keywords in their titles, or displaymath in their titles, would you think this would be a positive or a negative change? $\endgroup$ – Najib Idrissi Nov 2 '15 at 16:44
  • 11
    $\begingroup$ @AndréNicolas one can have mixed opinion on the merits of automated comments (indeed generally I am not a big fan of "canned" comments). However, mounting claims about "hack[ing] the site" and not even being able or willing to follow up on it to clarify is completely out of line in my opinion. $\endgroup$ – quid Nov 2 '15 at 23:14
35
$\begingroup$

It's not 90% of questions. The bot currently makes about 100 70 comments a day, which is about 10% of new questions. (I made it more conservative regarding the tag selection). The source is on GitHub.

The comments are automatically deleted after 5 minutes, so the only way for you to see many of them is to open new questions soon after they are posted. I suppose the repetitive language gets old if this is how you use the site, so I added some variety to the text, choosing a different wording at random.

New: the comments are now logged in Normal chatroom, for those interested in the operation of the bot. This room is also a suitable place for feedback on bot's performance.


Aside on the effectiveness of the comments. Some are ignored. Many result in edits by OP. Some are also acknowledged by replies:

img

A few users also provided feedback on the algorithm, pointing out the instances where the bot's heuristic misfired. I modified it accordingly.

$\endgroup$
  • 15
    $\begingroup$ "The comments are automatically deleted after 5 minutes," I see why you might want to delete them eventually, but I find the time span too short and the situation can be confusing. I observed situations where an OP asked a follow up question and then the comment just went away. This could make an odd impression. Maybe either expand the span or do not remove it once there is a reply (if this is possible to detect reasonably easily). $\endgroup$ – quid Oct 28 '15 at 13:44
  • 7
    $\begingroup$ The current version cancels the deletion if the comment received a reply containing a question mark. I think this makes more sense that keeping it every time there is a follow-up, because the follow-up is often "ok, did that", making the comment obsolete. $\endgroup$ – user147263 Oct 28 '15 at 19:33
  • 12
    $\begingroup$ This seems like a good intermediate solution. Generally, I am impressed how effective this seems to be. $\endgroup$ – quid Oct 28 '15 at 19:42
  • 10
    $\begingroup$ I find it quite annoying, especially since almost all of the instances that I’ve seen have been unwarranted. $\endgroup$ – Brian M. Scott Oct 29 '15 at 22:32
  • 2
    $\begingroup$ @BrianM.Scott Thanks for your feedback. I did change the algorithm after you pointed out some instances in which tags were in fact adequate. But since this apparently wasn't enough, I'll scale down the frequency of tag-based comments in general. $\endgroup$ – user147263 Oct 29 '15 at 23:09
  • $\begingroup$ May I ask if the bot automatically inserts the soft-request immediately after a question is posted? Or does the algorithm know that no tags have been inserted? sometimes I see Questions where tags are visible but there has been no edit to the post, implying (I think) that the bot requests further tags. $\endgroup$ – Kevin Oct 30 '15 at 11:29
  • 4
    $\begingroup$ @Bacon A question cannot exist without tags. The problem is that some tags are not very good for classifying the post (and arguably, shouldn't even exist). For example, if there is chain-rule tag (a recently created tag with mere 25 questions), but no (calculus), (real-analysis), (multivariable-calculus), etc then the bot will point out that something is missing. $\endgroup$ – user147263 Oct 30 '15 at 11:57
  • 1
    $\begingroup$ @NormalHuman: I should have said that I did notice that the specific instances on which I’d commented seemed not to have recurred, so the modification evidently did its job. But I felt awkward making making such comments very often. $\endgroup$ – Brian M. Scott Oct 30 '15 at 19:14
  • 10
    $\begingroup$ Thanks for running the bot. I won't add my particular opinion on its functionality because my appreciation for people who create such content to better the site outweighs any small issues I would have with the algorithm. $\endgroup$ – The Chaz 2.0 Oct 31 '15 at 13:31
  • $\begingroup$ I greatly appreciate the effort done by Normal Human, and think it is quite effective. That said, I don't think its use should be popularized. One bot is more than enough, and widespread use would lead to redundance and overwhelming artificialness. $\endgroup$ – Aloizio Macedo Nov 4 '15 at 2:30
  • 1
    $\begingroup$ @AloizioMacedo Oh yes, I think one bot is quite enough. I've made the code available because this is what I always do, someone might want to use it on another site, or just learn from it how the API works. $\endgroup$ – user147263 Nov 4 '15 at 2:34
  • 3
    $\begingroup$ Coincidentally, today another bot was discussed on SO: Is TaggerBot a good idea? $\endgroup$ – user147263 Nov 4 '15 at 21:52
  • $\begingroup$ ah (autocomment)... I was just about to ask if someone from the StackTeam implemented self-destroying comments... $\endgroup$ – draks ... Nov 9 '15 at 13:37

You must log in to answer this question.

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