Why I am leaving Code Golf

Question

The contents of this post are solicited opinion that I have been asked by a Code Golf moderator to share with you. I understand "If you don't like it, then GEETTT OOUUTT" applies here, so I have already submitted my account for deletion.

---

I joined Code Golf because the challenge of fulfilling certain criteria in the most terse possible syntax seemed awesome. Hacking languages to make them do stuff they weren't meant to in new and creative ways? Sign me up!!

Whoa then, what's this? Somebody posted a solution in a language I've never heard about, and solved the problem in 10 characters. Cool! Fifteen minutes of Wikipedia brings me up to speed on esoteric programming languages and I have a good laugh that someone went through all sorts of trouble to make a language that is basically useless except for getting a few tasks done in a very cryptic, dense syntax.

The next day, I log back into Code Golf and, oh look, that esolang has the top answer again. 4 bytes. Let's scroll down and... another esolang does it in 3 bytes. 5 bytes... Ok... getting old. When the solution language is designed for the exact problem, we are no longer hacking anything.

Analogy problem: Bake a moused-shaped cookie in as few steps as possible. Solutions:

1. Foil (15 steps), 2 upvotes: I shaped a piece of foil in the shape of a mouse and put the cookie dough inside.
2. Knife (9 steps), 5 upvotes: I baked a normal cookie, then cut a mouse out of it using the following technique...
3. MseShpdCttr (1 step), 25 upvotes: I used this device called MseShpdCtter that takes a lump of dough and shapes and bakes and spits out a mouse shaped cookie.

Using a specialized device to solve a specialized problem isn't fun after the first few times. Seeing people solve problems in Java, Javascript, PHP, C++, ..., and having others comment on it and find hacks to shave bytes is amazing. I really like that aspect of this community.

But the fact is every Code Golf thread is dominated by these esolangs and makes it really not fun to see a "competition" with the languages listed above. Yes, I know there is supposedly a winner for each language, but the fact is this is Stack Exchange and the top responses get the most focus. The current system is not suited for "there is one winner per programming language", and the people actually hacking up answers are at the bottom of the (fifth) page.

The line was finally crossed when I saw a Hello World challenge and the most upvoted response was a zero-byte solution in some language that prints Hello, world! with no input or code. That's not interesting. That's not hacking. For me, that's not fun.

I wish I had novel suggestions to improve this site. "Specify the language" doesn't seem like it would get enough independent responses. no-esolangs and esolangs-only tags are arbitrary. Maybe something like restricting answers by default to the top 10 most used languages on GitHub at the time or something would work, but there is a very real chance that is not in line with the community's zeitgeist.

Like everything, Code Golf isn't perfect. I can think of other ways this site can be improved and could be hashed out with the community and morphed into implementations we all benefit from. I think other people are sharing ideas on Meta with that purpose in mind. However, all of this is drowned out by the problem detailed above, and I don't see the point of working out details of improvements when Code Golf's foundation is fundamentally flawed.

So, because I cannot think of any feasible ways to help this community improve, I am just going to leave. I hope one day somebody here figures out a way to isolate the MseShpdCttr solutions from the ones utilizing universal tools, but until then farewell.

Answer 1

This answer assumes you have golfing languages in mind when you refer to esolangs. If you are actually referring to languages specialised for specific problems, then these are heavily discouraged in most cases*.

---

I'm sorry to hear you're leaving. This issue has been raised before, and I can sympathise with you in that I don't find languages with abbreviated syntax particularly interesting (in and of themselves). In over 3 years, I've never posted in CJam/Pyth/GolfScript - I stick with Python most of the time, and occasionally post in other 'normal' languages.

However, I have found that voters tend to side with cleverness and creativity over the number of bytes. In my experience, a clever solution in a standard language (like Python, JavaScript, C or even Java) will always outscore straightforward answers in golfier languages.

But I think you are missing the converse. An answer may be written in a standard programming language without being new or creative. In fact, many answers in common languages are quite poorly golfed. On the other hand, answers written in golfing languages can be amazingly creative and hack-y, especially since expert golfers tend to use these languages.

tl;dr - if you post a clever answer and explanation, you'll be upvoted regardless of language.

---

* As for the hello world question, it is unique - even boring answers in normal languages (e.g. print"Hello, World!") have 10-20 upvotes. And you mentioned answers sitting at the bottom of the fifth page - questions with five pages of answers aren't a fair representation of PPCG either.

Answer 2

I joined this site 85 days ago, with no "code golf" experience at all, so I'm relatively new to all of this.

I kind of understand the reaction you have against (I assume) languages like CJam and Pyth notably, which win almost every code golf challenge on this site. Users like @Dennis will always be on top of the rep chart because they answer everything in those languages.

Similarly I can understand how you got frustrated seeing my answer for the Hello World! challenge get more than 100 upvotes, when it doesn't even contain any code at all.

On the other hand, I think you are underestimating the efforts it takes to produce those CJam/Pyth answers. Its' true that they have very powerful 1/2 bytes operators, but as far as I know (never learned those languages, so correct me if I'm wrong), they are still relatively general purpose languages and they don't really have super-specific operators that are useful for a particular challenge. Having a one byte operator to generate all possible subsets of a list may seem like cheating, but it's really not that convoluted of an operation.

Moreover, answers by the best users of those languages are usually pretty well explained, and sometimes display remarkable tricks to save some bytes, which in themselves are sufficient arguments to justify their presence on this site.

But the most important point I think is that you care too much about who wins the challenge, or what trick is interesting.

Seeing people solve problems in Java, Javascript, PHP, C++, ..., and having others comment on it and find hacks to shave bytes is amazing.

I don't want to generalize too much since I don't have that much experience here, but from what I've seen the most interesting "hacks" are usually actually from those "esolangs" rather than general purpose languages, because you either have a plethora of different operators that can help you solve the problem in a very different but shorter way, or because the esolang is so esoteric that very ugly solutions are possible. Languages like Java don't really have both options and they usually end up being not the most interesting answers, even if we don't pay attention to the bytes count.

Finally, while it's true that CJam/Pyth are very often at the top of answers, and are almost always the winners of challenges, it's also true that a lot of languages that are bad at golfing usually get very respectable upvotes/scores.

APL or J for instance, while they still may look like noise, are languages that were not designed for code golfing, yet can usually yield very short answers.

Python and Ruby are probably the two "standard" languages that I see the most on this site, and they usually do pretty well and display very interesting tricks often. I'm surprised you have not talked about answers in those languages in your post, because to me they seem to be both popular and numerous.

Finally, my own answers are almost all in Prolog, which is an absolutely godawful language for code golf (can't really do any hacky things, very long built-in names, declarative paradigms forces you to make a lot of things explicit, etc.). But I still keep using it because a) I learnt a lot about Prolog using it and b) I love trying to golf the logic of the problem with respect to the way Prolog tries to unify things. This answer is ahead of Pyth, while considerable longer, so even godawful languages can "win" challenges in a sense.

I even started writing my own esolang called Brachylog which is to Prolog what Pyth is to Python (except way worse, but you get the idea). So that's one more reason for you to hate me I guess, but I find it rewarding to actually use a language I designed (on the spot, but still).

In conclusion, yes some languages like CJam or Pyth will win most challenges, but I don't think they are aggregations of ultra specific operators, they can still be beat in specific cases by other languages that were not designed for golfing. Answers in "normal" languages like Python or Ruby seem very popular and numerous to me which is something you should like, and even awful languages can get a good number of upvotes regularly.

EDIT One last thing that I forgot to mention is that you can appreciate the quality of some questions on this site rather than the answers themselves. A lot of my reputation comes from the questions I actually asked rather than my answers, which is also more satisfying.

Answer 3

I joined this site only 51 days ago, with no "code golf" experience at all, similarly to Fatalize. I understand that you feel "robbed" by the Pyth/CJam answers. It's very hard, and sometimes impossible, to create something shorter in other, more "standard" languages. But that doesn't have to take away the fun of solving challenges with these languages.

Since I joined, I've used Javascript in practically every answer I have posted, and I've thoroughly enjoyed golfing each and every one of them. Some of them are near the top, and one is the most up-voted answer, even though they are by no means the shortest.

But the fact is every Code Golf thread is dominated by these esolangs and makes it really not fun to see a "competition" with the languages listed above.

It's true that these languages are dominant in the realm of "who can write the shortest code?", and this could take away the fun of answering in "standard" languages. But it doesn't have to. The reason I continue to participate in code-golf, though I don't often win, is because I like to see how short I can get my answers in my favorite language (currently JS). For the same reason, Martin Büttner's amazing leaderboard template not only has an overall leaderboard, but a per-language leaderboard.

I guess my main point is that the primary goal of code-golf, as with any other recreational activity, is to have fun and improve your skills, not just to win as many challenges as possible. It's not a prize-paying championship; there's nothing you're missing out on by not using these golf-optimized esolangs. So if you'd like to try answering a challenge with C++ or PHP or JS or Java, go right ahead; there's no one holding you back, and no one who's going to criticize you just for doing so.

One final thought: I've found that instead of just looking at the water and deciding it will feel cold, it's best to go down a few steps and see how it really feels. You just might like it. :-)

Good luck!

Answer 4

To be honest, I don't really see the problem. Code golf is about writing shortest entry in any language that is accepted. People created some strange languages to be able to write short programs. Some people learned it and use it.

While I understand that seeing all these "exotic" answers might be bugging you, I also think that your approach is wrong(but if you don't like it, you are doing the right thing leaving - it's your decision, and it's perfectly fine).

First of all - answers in CJam or Pyth tend to be creative. They use some different approach to the problem to minimize characters. I like that.

Secondly, code golf is all about fun. If you get the shortest entry, you don't win a cookie. You just get the shortest entry. It's literally meaningless, so why care? If you wrote some brilliant entry, people might appreciate that and you can feel good, but that's it. We're here to have fun, and being too focused on winning is bad.

And if you're so inclined for winning - remember that programming is not only the ability to use X programming language - it's about being able to use the right tool for the job. If you want to make simple operations on text file, you use sed or awk. If you want to find something in file, you use grep. You don't go for C++, because it's not the right tool for the job. When the job is Code Golf, the right tool is often CJam/Pyth/another esolang.

With this in mind, I often see many answers in "mainstream" programming languages that are really short and impressive. And I think this is what Code Golf is about.

Anyway, good luck! I hope that you either stay here and enjoy it, or go and find some other place where you can enjoy programming whatever you want.

Answer 5

Just compete with yourself and others using your language. Forget about the rep you earn. I enjoyed answernig a leap-year challenge using z/OS assembler and then seeing exactly how short I could get it. I play CoD online too and my kill/death ratio is always less than 1 - I'm not going to compete with the guys who get to play all day every day and just enjoy the game and getting a few good kills.

Answer 6

Even now, this is Code Golf and Coding Challenges. The former name (Programming Puzzles and Code Golf) had more emphasis on the non-golf challenges, but the name change didn't make us golf-only. Whenever I get frustrated at being outgolfed in esolangs, I simply have a stroll through the code-challenge or old popularity-contest questions and pick a fun one.

In a sense, there's more than one community that shares this site, and it's a choice whether to mix with the most competitive golfers or to find another island to settle on. It's a bit like choosing to do a fun run somewhere scenic rather instead of aiming to break world records.

FWIW, my first golf answer outscored the winner by more than 2:1, despite being 88% longer. Choosing a language seemingly unsuited for the task and explaining how it operates will have influenced voters, I think.