I really like this question but I know there are complaints of it being more about art than programming. I'd like to hear both sides of this and see what is in the best interest of the community and the site's public beta.
I'm posting my own view as an answer so that the votes will show where it lies in relation to other answers.
I think popular questions are important to increase traffic, which in turn can bring extra views to the lesser seen questions. The whole stack exchange network sees links to our most popular questions. However, obviously not all popular questions are acceptable or even useful. For example "paste in your favourite picture of a cat" may well get lots of upvotes, but it would have no requirement for programming knowledge nor any learning or experimenting aspect. In particular, the increased traffic it would bring would not be people who necessarily know about programming or have any interest in the rest of the site, so it could just increase noise and potentially introduce votes on other questions from people who are not programmers, skewing the votes across the site.
So somewhere between random cat photos and pure code golf is a line we need to estimate in order to welcome questions that will bring in more programmers and golfers, without bringing in people in search of cats.
I see several questions on the site that are near that border, and in each case I judge it by whether the answers to it demonstrate skill and determination, or just aesthetic style. Those that require both, I see no problem with - I think the combination is a good thing. Those that are purely aesthetic and could be answered with little programming knowledge I see as noise, potentially taking us back a step.
Recently there seems to have been a move towards voting to close questions that have an asthetic or artistic aspect, regardless of how much programming skill they require. I think it is important to discuss how much programming skill makes a question acceptable, because we are in danger of closing questions based on how much art is in them, rather than based on how much programming is in them. As I see it, you can't have too much art in a question, only too little programming.
The particular popularity contest mentioned in the question is one I see as beneficial to the site. I'm not a confident code golfer and tend more towards questions where there is a challenge rather than minimising bytes. In answering the tweetable art question I started off with something very simple, but found I had to learn a bit about golfing along the way. This allowed me to attempt a more complex answer which led to me working out more golfing techniques which I probably wouldn't have learned from the straight golfing questions because I wouldn't have had the confidence to try them. Now I'll be much more likely to have a go at some golf questions because I've been able to start somewhere in between.
For me, a question with a strict byte limit seemed more approachable than pure golf questions. It was a stepping stone for me. For others, pure golf may actually appeal more since there isn't a limit to stress about. Since different people enter into a topic from different directions, I think the more different variations we have the more routes in there will be and the more golfers we will end up with.
I want to see questions that don't require programming skill flagged quickly, but I don't want highly demanding questions that happen to have an artistic side to be lost without considering their programming benefit.
I'm not going to write a contrary opinion -- someone else will have to do that. Instead, I'll zoom in on my personal line for this.
My bar for a question should be that programming is the challenge, rather than just being a challenge where you program.
Code golf is clearly in the first category, as brevity pushes you to exploit intricacies of the language. Underhanded contests, even if sometimes maligned, reward using language features in sneaky ways. Fastest code pushes you to optimize both algorithms and computer cycles.
King of the Hill contests though sometimes feel more of games where the challenge is to find a clever strategy and predict what others will do, with the code just an avenue to make the strategy run. There's a challenge, and you code it. In theory, computerized AI's could be used to find strategies on the fly, which is closer to a programming challenge, but that doesn't seem to be happen much.
Fastest-algorithm questions also can seem to be about optimizing theoretical algorithms rather than code, though maybe that's close enough to get a pass.
"Programming is the challenge" is the line I favor for questions on this SE. (And yes, that would mean clamping down on King of the Hill which is very popular, so I don't expect it to happen.) So where does that mean for art questions like Tweetable Mathematical Art?
If the goal is simply to produce pretty art, and it's easy to write code to make whatever art you want, that's just a challenge where you program. In Tweetable Mathematical Art though, there's a harsh character restriction for the code, which definitely makes coding a challenge.
Draw random black-and-white forest is more borderline, but the requirement that the forest be completely dynamically generated makes it a programming problem. Simply writing code to generate what looks like a tree is a programming challenge, and doing it well is suited to be judged as a popularity context.
For Create a popular Penrose tiling though, entries are likely to be distinguished by how pretty the Penrose tile designs are. Simply generating a Penrose tiling is certainly a programming challenge, but it's a bar that every program must pass. So, entries cannot be judged relative to each other based on it. The programming is not intrinsic the to contest -- contestants could instead paint tiles on paper and submit photos of their assembled tilings.
I know that popularity contests make it hard to evaluate what the competition is about because that's up to the voters' subjective preferences. Given that, it's up to the poster to explain why coding is intrinsic to their challenge, and not rely on "answers not in the spirit won't be voted up."
I think there are two separate issues which sometimes occur together but not always. They relate to the two key requirements we (in theory, at least) impose on questions.
Some art questions are so imprecise that it's subjective whether or not an answer meets the requirements. "Too broad" is probably the best close reason, but I can see them attracting custom closure reasons.
I anticipate the objection, "But popularity-contest is objective". I disagree, but that's a separate issue.
However, where you have a pop-con question which is set up for voting on the aesthetic qualities of the output rather than of the code, it's gone off-topic for this site. A contest is not a programming contest if it doesn't measure programming skill.
User xnor was recently kind enough to warn me that "the community" frowns on art-related programming questions, although ample evidence in this thread and elsewhere suggests that "the community" is in fact a handful of individuals not representative of the greater consensus.
Consider this meta thread, where user githubphagocyte's highly popular response is summarized thusly:
Recently there seems to have been a move towards voting to close questions that have an asthetic or artistic aspect, regardless of how much programming skill they require. I think it is important to discuss how much programming skill makes a question acceptable, because we are in danger of closing questions based on how much art is in them, rather than based on how much programming is in them. As I see it, you can't have too much art in a question, only too little programming.
About a third as popular is xnor's response, which makes some distinctions clearer:
Code golf is clearly in the first category [of acceptable art challenges], as brevity pushes you to exploit intricacies of the language. Underhanded contests, even if sometimes maligned, reward using language features in sneaky ways. Fastest code pushes you to optimize both algorithms and computer cycles.
and
If the goal is simply to produce pretty art, and it's easy to write code to make whatever art you want, that's just a challenge where you program. In Tweetable Mathematical Art though, there's a harsh character restriction for the code, which definitely makes coding a challenge.
referring to this exceptionally popular code competition.
No dissenting opinion has been penned in this discussion. In other words, the observable consensus is that an art programming challenge should be allowed so long as programming is a principle and integral component of the challenge. Furthermore, the observable consensus holds that code golf necessitates a degree of programming mastery constituting such a challenge, which is an assessment I heartily endorse.
Dozens, if not hundreds, of art-based challenges exist on PPCG. They are invariably popular (both in terms of voting and participation), creative, and fascinating to behold. They range from simple creative challenges such as ASCII Art Calendar, ASCII Art "Hello, World!", Draw a Heart Shape, Draw the Olympic Games Logo, Create an Analogue Clock, Deoxyribonucleic ASCII to abstract challenges such as What do you see in an inkspot?, Rearrange Pixels in an Image, Images with all colours to broad, open-ended art challenges such as Make a Valentine Wish, Draw a Sun Map!,Tweetable Mathematical Art, Draw Random Black-and-White Forest, Make a Circle Illusion Animation, and many more.
There are also several popular-but-closed challenges, with the notable observation that the de-artifying of PPCG began in roughly mid-2013, and that the same small pool of individuals is responsible for nearly all closures for questions of this type. I will not name names, but a quick survey of the "[closed]" questions when running a PPCG search for "art", "turtle", "picture", etc. proves enlightening.
The most unsettling aspect of PPCG de-artifying is that the stated reason for many of the closures, which is
This question does not appear to be about programming puzzles or code golf within the scope defined in the help center.
is demonstrably untrue.
A good faith search of the PPCG help center yields absolutely no rules regarding the acceptability of art in programming challenges. Any mention of art programming is conspicuously absent from the What topics can I ask about here? page that defines on-topicness. Likewise, art programming questions are nowhere mentioned in What types of questions should I avoid asking?, which is fortunate given that most, if not all, of the excellent art programming challenges listed above would need to be closed for off-topicness if such a prohibition did appear.
In the absence of any kind of prohibition, a user new to the site might survey existing questions, which yields a vast corpus of popular, admissible art programming challenges ranging from the highly constrained to the highly abstract/open-ended. This same user might consult this discussion where the obvious and uncontested consensus is that open-ended art challenges are permitted subject to certain clear restrictions, as described above in "This Thread".
As xnor warned me, this question was shut down almost immediately after being created, with the cited reason being off-topicness.
I can accept a question being closed. There was no serious investment of time involved in creating it. What upsets me is that absolutely nothing besides the opinions of a small (tiny, in the context of the full user pool) group of influential users appears to contraindicate this kind of challenge.
Numerous similar challenges exist with the enthusiastic consent and support of the broader community prior to the "anti-art" era. The challenge is exclusively accessible to expert programmers, with programming being the clear, primary sine qua non, meeting the standard laid out in the consensus in this thread. The challenge is in fact so programming-centric that I'm hard pressed to think of any kind of website aside from a code golf challenge board where such a challenge could reasonably be issued.
Finally, and most importantly, nothing in the codified rules for this website contraindicate this type of challenge.
I've barely been here three months. I've participated in many challenges and accrued a reasonable sum of "rep", granting me the ability to vote to close any question that doesn't suit my fancy. If my personal opinion of what shouldn't be here happens to coincide with that of just four other users (out of thousands upon thousands), we can close a hundred challenges in a week if we want to. We could deem "ASCII art" inappropriate and liquidate the site. And perhaps we'd be justified in doing so if there was anything anywhere besides our five completely undocumented opinions that we don't like ASCII art and so screw community content voting, nobody gets it.
Moderators: I would appreciate it (and I know it would help newer users) if something somewhere was codified as to what is the appropriate cutoff between art and programming on this site--preferably in the help center. Explain why the litany of programming challenges above have stood unchallenged, while the litany of [closed] programming challenges have been closed. This thread, the clear community consensus, and the help center certainly don't explain it.
Is the anti-art era here to stay? We need a definitive ruling one way or another. My thanks for your consideration.
