Guest User

Untitled

a guest
Nov 7th, 2017
159
0
Never
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
text 7.96 KB | None | 0 0
  1. oerjan: I really /really/ hate Stack Exchange
  2. ais523: I kind-of reluctantly moved there because that's where the codegolfing was. and although PPCG's community is great, Stack Exchange is sufficiently bad that I was starting to resent code golf by association. I wasn't enjoying myself at all, it was something of negative addiction cycle towards the end.
  3. quintopia: the incentives are all wrong
  4. ais523: in particular, the fact that answers with more votes have more visibility, thus more votes, means that getting a vote lead early is the main thing that controls how much reputation you'll get. and the site puts huge emphasis on reputation, then when you actually get it, it actually makes the site experience worse rather than better
  5. it applies even on PPCG, because if you go to the people who can actually change the site software to, e.g., report a bug that disproportionately affects PPCG, they just tell you you're using the site wrong. this is a particularly bad example of an SE bug that should be trivial to fix (given that it just involves removing a misfeature), and frequently affects newbies at PPCG: https://meta.stackexchange.com/q/276269 but in general, none of the features of the Stack Exchange software actually work for their intended purpose (this is particularly noticeable at PPCG but true elsewhere, too) when a feature does turn out to be useful, it's because it's being used in a way contrary to the official recommendations.
  6.  
  7. quintopia: i have thought on many occasions that moving the PPCG community to more code golf friendly software (that still keeps what /does/ work well at SE) would improve the community
  8. ais523: yes the top two features that PPCG has and most golf sites don't have are: a) the ability to add an explanation along with the code;
  9. b) an ability to comment on other people's code. that way, it's actually possible to have some sort of workable communication between the people golfing and the people appreciating the results
  10. quintopia: and c) the ability to redirect comment threads to group chats
  11. ais523: note that the use of SE's comment system for b) is actually banned according to the comment policy but everyone ignores that.
  12. quintopia: no, that does more harm than good in several ways the way it's currently implemented. and also requires subsequent moderator action in 90% of the cases where it's triggered, either intentionally or accidentally
  13. quintopia: i've only seen it done intentionally
  14. ais523: if I'm having a productive conversation about improving a post, directing it to chat is probably a mistake, because you want it attached to the post (and can clean it up later). OTOH, if it's a case where one person is being unreasonable and other people are arguing with them, the conversation is probably better off not happening at all
  15. quintopia: yeah, having (hidden) attached comment threads would be a better solution
  16. but i like the idea of being able to coerce people into live chats somehow
  17. at the least, the ability to link to a live chat that opens in the browser
  18. ais523: yes, ability to link to chat is very helpful, but just doing it with an actual regular link is much better than the autolink. in fact, the autolink works more than once and IIRC creates a new chat room each time
  19. quintopia: if we take those few features, add anagolf like automatic code testing and scoring, add en /entire forum/ for challenge sandboxing instead of one stupid overgrown thread... i didn't know there was an autolink like i said, i've only ever seen it done intentionally.
  20. ais523: quintopia: it comes up every time you try to add a comment
  21. also the created chat room has no owners and no way to add any, which is a problem in its own right (and there's a trivial solution, making it the OP of the relevant post, that hasn't been implemented). that said, I have had thoughts about what an optimal golfing site would look like
  22.  
  23. first, and most importantly, you have some sort of system to keep out spambots
  24. ideally, this should be requiring actual golf entries to recently-posted problems to be submitted that are at least decently competitive
  25. as that proves that people understand what the site is about, and will also help repel human-aided spambot farms
  26. once you're sure that there's no sort of automated attack going on, that then allows you to let people express opinions without worrying about the system being flooded
  27. then, you let people write up golfing problems, with some sort of "automatic marking"
  28. I think you express the specification in the form of a set of possible inputs (which could be infinite), and a program (that isn't necessarily golfed, and in fact probably shouldn't be, i.e. it should be as clear as possible)
  29. then an answer is correct if it gives the same answers as the program (tested on some manually given test cases + some randomly selected ones); you have a recheck feature which allows people to give counterexamples to other people's programs
  30. all submissions to a problem, and all problems, should be categorisable with tags, and people should be able to tag each other's answers
  31. and people should be able to attach commentary to problems and submissions (whether it's explanations, feedback, or something else; keep it decently freeform)
  32. I think there should be some way to improve problems after they're posted, perhaps by marking the problem as provisional for a set period, and allowing wiki-style edits to it (which invalidate any answers that violate the new spec)
  33. incidentally, when I left PPCG, I was sent a number of messages by Stack Exchange staff that were worded to look like they were concerned about my reasons for leaving, but which (from the way they were worded) were clearly written without actually having read my stated reasons
  34. so I'm assuming they were just boilerplate misleadingly written to imply that they cared about what people thought about the site
  35. meanwhile, the actual PPCG community listened to my concerns (and largely agreed with them), but I think they're powerless to fix the fundamental issues with the site software
  36. oh, one other thing is related to I/O; I think each challenge would specify the inputs and outputs using a well-known format (probably JSON), but each language would have a "default boilerplate" that went and decoded/encoded the JSON into whatever form was most convenient for the language, and that doesn't count against the byte count
  37. so in most cases, what you'd actually submit would be a function body, or the language's closest equivalent to that
  38. this would end all the arguments that you get in individual questions about which I/O formats are acceptable, and change them into a more centralised question about which I/O formats are most suitable for each given language
  39. which only has to be argued once per language, not once per challenge
  40. oh, for non-golf challenges, you should also have the option of specifying a scoring function
  41. another useful feature (which no golfing site has yet that I'm aware of) would be an auto-encoding feature, where you enter your program in whatever encoding is most convenient for working on the program
  42. but the site automatically works out the best possible encoding of that that works in the language
  43. for example, you could type Jelly using Unicode (which is normally easier), but have that automatically transcoded to Jelly's character set (which is more compact in terms of the number of octets needed)
  44. something I'm torn on is efficiency: should we require the answers to run within, say, a factor of 10 of the model answer? one approach I see is separate leaderboards for answers that do and answers that don't
  45. meaning that there are two golf competitions at once (the shortest answer, and the shortest answer that uses an algorithm that isn't considerably slower than that of the model answer)
  46. doesthiswork: is it possible to dispute a tag on someone's answer?
  47. ais523: it should be
  48. I have a system in mind where tags are voted up and down, but with hysteresis
  49. something like "a score of +3 for a tag adds it, a score of -3 removes it, votes change the score by 1 point, the OP sets the original status"
Advertisement
Add Comment
Please, Sign In to add comment