I like to encourage people, to remove the number of characters from answers, given for questions, which are not code golf.


Here is a recent question which isn't a codegolf, which has 8 answers after 22 hours. 7 of them are decorated with the number of characters in the headline in bold, as if the people, giving the answer, didn't realize, that it isn't a CG. Here is another example - 8 of 12 answers contain the number of chars in a big, bold fashion in the headline.


  • Users will misinterpret these questions, and vote on these numbers
  • Users will mistakenly think it is a CG, and come up with golfed answers themselves or leave the question
  • Users will think this is normal behaviour and start their questions without winning criterion too.

suggested solution:

Everybody noticing such a situation should be officially encouraged by voting in this thread to remove these big, bold values from the headline.

alternative solutions:

  • Commenting the answers and hoping the best. I tried this before with questionable success. People did not remove the numbers, and I don't know how many realised, that they have to adjust their voting, in case they voted on these numbers. I guess later votes happened mostly after reading the comments, so with this aspect in mind.
  • Better preparation of questions. Questions which aren't codegolf should carry an emphasized hint. Said question is very vague about the winning criteria (efficient code). Since newcomers don't prepare their questions well, this will not work too. :)
  • Up to you.
  • 3
    \$\begingroup\$ Note that neither of those questions specified how answers would be scored, so I guess people just assumed it was code golf by default. \$\endgroup\$ – hammar Jan 6 '12 at 17:30
  • \$\begingroup\$ Is there such a default-rule? People should ask for clarification, if there is no winning criteria, and wait for an answer. \$\endgroup\$ – user unknown Jan 6 '12 at 18:41
  • 1
    \$\begingroup\$ It's probably not a rule, but I think there is an expectation among the users of this site that most questions are code golf, and they might not have noticed that the code-golf tag was missing. Making this mistake is less likely when the question is explicit about scoring. \$\endgroup\$ – hammar Jan 6 '12 at 18:50
  • 1
    \$\begingroup\$ If you want to suggest, that we assume code-golf if explicit winning criteria is missing, and therefore keep the numbers, please do so in an answer, so that we can vote about it. \$\endgroup\$ – user unknown Jan 6 '12 at 20:36
  • 5
    \$\begingroup\$ For both examples given, the questions were so trivial it made sense to me to assume they were code-golf. My rule of thumb has been that simple questions are code-golf unless the question is difficult, or unless the question specifically indicates otherwise. Looks like I'll need to revisit my rule of thumb. \$\endgroup\$ – Steven Rumbalski Jan 10 '12 at 15:35

I largely agree, though my main concern is making it more clear that code-golf isn't all that we do here.

I believe that while we remain in beta you already have enough rep to edit, and I would suggest that you do so. If the author rolls it back, well you tried, if not the site is improved for it. Acting very early is probably good, so that people coming to the question see the lack of decoration.


The problem occurs when people post a non-golf challenge, but either don't specify an objective and unambiguous winning criterion, or specify a criterion so loose that it results in lots of ties for first place. In such a situation, it's natural for answerers to start trying to find some way to one-up their competition:

"Hey, I just got a perfect score! No-one can beat my answer!"
"Oh yeah? Well I got a perfect score too, but I did it with fewer chars!"

In situations where there really is no objective winning criterion, I think I'd support just voting to close the question until one is provided. It might seem harsh, but if the goal is to keep such questions from spreading by example, I think it's the least harsh way that's going to have an effect.

However, I'm not sure how to deal with questions like this one, where the OP did provide a winning criterion (number of + and - chars) and even a backup tie-breaker criterion (number of / * ( ) = . , and 0-9 chars), but people very quickly came up with several answers that scored a perfect 0 on both counts. Certainly voting to close the question just because the OP neglected to provide a second tie-breaker rule seems a bit unreasonable.

(The fact that the answer currently chosen as the winner of that challenge is not actually one of the perfect ones is a bit annoying too, but at best only tangentially relevant to the issue at hand.)

Or perhaps we should adopt a policy that as soon as someone posts a theoretically unbeatable answer to a question, that question should be closed and the answer declared as the winner? I wouldn't really like that personally (I'd prefer to try and find some way to make a "better" answer, even if it means adopting an unofficial tie-breaker criterion such as character count), but it would certainly take care of that aspect of the problem.

  • \$\begingroup\$ Well, we can have 3 winners in the end: The accepted answer, the answer with most upvotes, and the answer which fulfills the winning criterion in a best way. Every puzzle could have an official period for about a week, to post your solution, and then a winner should be elected. Later answers could be out of competition for the 'accepted answer' but I would still like to see improvements and vote them up. For most puzzles, the last tie brake could be (implicitly) the solution, given more early. \$\endgroup\$ – user unknown Jan 17 '12 at 2:53
  • \$\begingroup\$ Wow, I just realized that the answer that I accepted (for the question that you link to) was not perfect. I probably went with the most up-voted one and just forgot about it. \$\endgroup\$ – PhiNotPi Feb 20 '12 at 20:03

I agree. Perhaps that should go into the FAQ too: namely that questions are to have character counts, and non- questions probably should not, unless the code size is somehow relevant.


I tend to agree with hammar in the comments: many non code-golf questions provide no objective judging criterion. My personal observation in such cases is that the Python questions tend to bubble, which I interpret as a language popularity vote. (I may be wrong, it's just my feeling)

I'd rather see a clear-cut winning creterion. Failing that, character count is better than language choice.

  • 1
    \$\begingroup\$ I guess a better solution is, to close the question until there is an objective winning criterion. But this is a way to threat the question, not the answers, so I start a new meta thread for that. \$\endgroup\$ – user unknown Jan 10 '12 at 20:05

I think that in the absence of a clearly-specified, objective winning criterion, character count is the most natural default. The two questions referenced in the OP are so trivial that I wouldn't have even bothered writing an answer unless I was going to code-golf it to actually make it challenging and interesting. I knew they weren't code golf, and I knew the code count was irrelevant, but I still did code-golf on purpose because there was nothing else to do... this site isn't supposed to be Java 101 homework questions.

So my vote is that all questions are automatically code-golf unless some other scoring criteria is specified, and we implement by editing the code-golf tag in if it's missing.

  • 1
    \$\begingroup\$ Why not ask for a winning criterion, and waiting until there is? Why not closing the question, until the criterion is specified? \$\endgroup\$ – user unknown Jan 10 '12 at 20:08
  • \$\begingroup\$ @user unknown: Why not ask -> no reason, and it's actually done most of the time. Waiting until there is -> that's kind of against human nature. Closing the question -> that doesn't go well with encouraging the author (most likely not a regular) to refine. \$\endgroup\$ – J B Jan 10 '12 at 23:24
  • 1
    \$\begingroup\$ Well, it does. On Skeptics, it is done over and over again. As we see, the asking persons often don't react. \$\endgroup\$ – user unknown Jan 11 '12 at 0:00
  • \$\begingroup\$ If they don't react, I feel this just proves my point. \$\endgroup\$ – J B Jan 12 '12 at 8:01
  • \$\begingroup\$ Is there a way to display some sort of a banner forcing the user to click "OK" that they have provided a scoring method? Or require either code-golf or code-challenge to be one of the tags to force the user to understand the meanings of those tags before they submit a new question? (maybe a question more for the SO devs). \$\endgroup\$ – mellamokb Jan 16 '12 at 22:10
  • \$\begingroup\$ @mellamokb: I think requiring at least one out of a given set of tags is technically possible, but I'm not sure it would help anything. I've seen several questions tagged as code-challenge that didn't specify an unambiguous winning criterion (e.g. this old one and this more recent one), leaving it to answerers to try and find some way to one-up their competitors (most seem to have chosen golfing). \$\endgroup\$ – Ilmari Karonen Jan 17 '12 at 0:13

You must log in to answer this question.

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