31
\$\begingroup\$

Edit 2: See Grace Note's answer.

Edit: Answers may be allowed, so that question-specific sandboxes would be actually on topic, such as attempted for Stack Exchange Quine. It could also enable people to post more detailed comments to proposed questions and potential answers, helping mitigate more loopholes (limited agreement to user80551's answer).


TL;DR

I feel that the sandbox idea would be better suited to its own subdomain (e.g. sandbox.codegolf.stackexchange.com), retaining the standard format (specifically Meta) familiar to all of us, but with one key difference: posting requires 0 rep.

I'm just posting a personal opinion/proposal of mine, looking for ideas, agreement, disagreement, and criticism.


Basically, I'm looking at it thinking that the sandbox as a subdomain would be very beneficial.

Things you could do with a separate subdomain:

  1. more easily migrate questions not quite ready for prime time from the main site or meta to the sandbox
  2. implement an "easy post" button for using the sandbox question as a template for a new question on the main site
  3. apply challenge type labels right off the bat
  4. search sandbox & filter searches just like on main site

That could help it become more maintainable, especially as our site base grows. It would also help people in assisting more often with challenges that they prefer working with. This could mean higher quality code bowling and popularity contest challenges down the road, which at the moment are a little too fuzzy on whether they are too broad and/or unclear. It could also mean, instead of closing most of these bad questions that needed to go through the sandbox first, we could just migrate the repairable ones.

What are your thoughts on this proposal?

\$\endgroup\$
  • 6
    \$\begingroup\$ My thoughts are that you're not the first one to suggest this. ;) \$\endgroup\$ – Martin Ender Jul 15 '14 at 10:01
  • \$\begingroup\$ @m.buettner In fairness, OP has responded to my thread, so they have seen it. This post is probably trying to lend further support for the idea, which I commend. \$\endgroup\$ – Chris Jester-Young Jul 15 '14 at 10:05
  • 1
    \$\begingroup\$ @m.buettner Matter of fact, I posted a more detailed answer, in which this is but a really small overview of. \$\endgroup\$ – Isiah Meadows Jul 15 '14 at 10:06
  • \$\begingroup\$ Ah, I didn't see that. Well I already upvoted anyway, because I think it's a good idea and some new exposure specifically for this proposal seems like a good thing. Although I'm not sure SE will actually make an exception for us and give us another instance. I don't know how that works though. \$\endgroup\$ – Martin Ender Jul 15 '14 at 10:09
  • \$\begingroup\$ The main issue with this suggestion is that would require some cooperation from the SE dev team. If you can get that, I'm all ears. \$\endgroup\$ – John Dvorak Jul 15 '14 at 12:11
  • \$\begingroup\$ I'm getting ready for a 4-hour and then 11-hour flight, so I'll write up a full answer during those, but the essence of it is that this is basically already provided by the tools that Stack Exchange provides, and it's essentially just reinventing the wheel. See also my "kill the sandbox" meta question. \$\endgroup\$ – Doorknob Jul 15 '14 at 12:30
  • \$\begingroup\$ There would also be one big advantage people tend to forget. Posting to Meta requires 5 rep, so new people can't post their first question there. \$\endgroup\$ – seequ Jul 16 '14 at 8:46
  • \$\begingroup\$ I'll edit my suggestion accordingly \$\endgroup\$ – Isiah Meadows Jul 16 '14 at 8:51
  • \$\begingroup\$ @TheRare Updated \$\endgroup\$ – Isiah Meadows Jul 16 '14 at 8:53
  • \$\begingroup\$ Can supporters please explain why they'd prefer a separate subdomain for meta rather to posting questions with an in-construction tag on the main site? \$\endgroup\$ – xnor Jul 16 '14 at 13:17
  • 1
    \$\begingroup\$ @xnor First, a brand new challenge that has been polished in the sandbox should arrive at the top of the list. It deserves the same visibility that every other post gets. Second, tagging with under-construction does not immediately prevent answers. Third, the main site needs to be clean, but under construction questions are likely to be abandoned. I sandboxed two incredibly popular questions for more than three weeks each, and I would be upset if they were deleted while I was on vacation. \$\endgroup\$ – Rainbolt Jul 16 '14 at 13:43
  • \$\begingroup\$ @xnor I've only had one question that actually turned out decent that I hadn't sandboxed first, Implement an integer parser. Every other one flopped to some extent. I use the sandbox on a regular basis, but it's far too clunky on Meta, and in really don't want the barrage of answers to an incomplete question. \$\endgroup\$ – Isiah Meadows Jul 17 '14 at 16:39
12
\$\begingroup\$

I'm Grace Note, a Community Manager at Stack Exchange.

Originally, I was writing this answer explaining the mechanical issues with this idea. I won't bore with details, though, but in the long run, migration (and especially chain migration) is pretty messy and this would be a mess to implement.

Today we had a meeting to discuss something we'd been brewing on Area 51, Stack Overflow Academy. The specifics of implementation may have differed, but the goal of these was the same - you wanted a place that you could help provide guidance and sculpt posts with users, an environment that allowed you to have "questions" asked not for the purpose of getting answers, but for the purpose of building better questions.

We decided to close the proposal. Right now, over on Meta Stack Overflow, they're looking into ideas on how to make it work on the Meta itself. The fact is that the Meta really should be a place to get help in constructing good questions when the basis is present, a place that we already suggest to people now and then across the network when it comes in as a support email. There are always challenges, though - for example this site itself has seen 14 threads for sandboxes which get tons of answers, which paints a very clean image of the problem of drafting/sandboxing overwhelming the rest of meta activity.

The proposal here specifically for creating a subdomain, I will for now say that this is . Not because the idea of something beyond Meta is bad, but that this particular implementation of it is, we feel, sub-par. Meta is itself an illustration of why this is - we as a network created Meta out of the tools we had at our disposal, to serve a purpose that we needed without the full instrumentation to do what that purpose itself would need. So while Meta is what one might call "sufficient" at the job, it's not the best. Going full guns to create an entire Q&A section for the purpose of sandboxing and mentoring in asking, I feel, is repeating this same mistake - it uses what we have on hand in a makeshift "It works, that's what matters" that ignores what it means to actually work. That and personally it feels about as excessive as using a grenade to clean your sink - there's way so much overhead associated with a full Q&A that goes beyond what you need for this task.

This is something that is a real problem across the network, and I feel that if we want to work on providing an in-house solution to this, we should want to create something solid and robust, that has the tools needed for the job and nothing more. We don't have this solution yet, we don't have a full plan. But it's something we'll be working on. Not "in the future", not "in the near future", this is something to be working on in the now. And while we work on it, Meta will have to remain the existing solution. We'll see what we can do, what things we can help make it work better on Meta. Some ideas include a new meta tag besides to identify these, as well as better forms of filtering. We're going to be happy to try out a lot of different things, but there'll be a solution. Be it something we integrate into the main site, be it something that adapts Meta to a better way, be it ultimately a separate section of the site, there'll be a solution.


I will give one idea, though, that might help this site while we work on something more mechanically integrated - on Mathematics, they use a single thread for sandboxing, one that has a fixed number of answers. Rather than letting people just keep posting, you can only draft while there's an open slot for you to take. If the draft is "abandoned", it can be cycled out. This kind of strategy may help with some of the flaws that have been observed in the current sandbox strategy. Not all of them, but I hope it might be a useful direction to look at.

\$\endgroup\$
  • 11
    \$\begingroup\$ I've used the sandbox on math.SE and it serves a very different purpose: it's to let you work on a complicated answer with lots of MathJAX. I think the current PPCG sandbox suits its purpose better than that. \$\endgroup\$ – Peter Taylor Aug 14 '14 at 21:56
  • 2
    \$\begingroup\$ Thanks for the detailed response! I agree with Peter though that a limited sandbox will be problematic here, especially with quickly increasing sandbox volume and some questions lingering there for several weeks or months to mature (especially the more involved king-of-the-hill challenges). A 5th meta tag that can be easily filtered by (or filtered out) sounds like a very good solution though! (Although it still comes with the problem of new users having to pass the 5-rep barrier.) \$\endgroup\$ – Martin Ender Aug 15 '14 at 7:22
  • 1
    \$\begingroup\$ @MartinBüttner If the idea is to ahve a tag that easily gets filtered out, I don't see why it couldn't go on the main site. If you have a sugegstion for a question, you post it on the main site including that tag and by default, people won't see it. The moment it's ready, the tag is removed and it's history is wiped so it looks like a brand new question. When a question by a new user needs work, it's retagged and no one except for people willing to help can see it. \$\endgroup\$ – overactor Aug 15 '14 at 7:41
16
\$\begingroup\$

In response to the question which Doorknob asks in his answer,

Aren't you just reinventing Stack Exchange?

I think that he does Chris Jester-Young's suggestion an injustice. It's not reinventing SE so much as making a small extension which turns an existing feature (cross-site migration) into something much more useful. At present we can only use it to move questions between main and meta, which happens extremely rarely. (To pre-empt: yes, mods can migrate to other sites. That also happens extremely rarely to and from PPCG).

The major advantage that I see in having a sandbox.codegolf.SE stack is precisely that it's a separate stack. (Although hopefully linked similarly to the way meta is, so that there are people with the rep to migrate questions to main!) This fixes the big problems with the option which some people advocate of having "Under construction" questions on main:

  • Having "Under construction" questions and "normal" questions bumping each other off the front page. (I took a quick sample from the recently retired sandbox on meta: most of the answers I sampled have been edited more than 5 times, and some 10 times. Similarly the question with which professorfish trialled this approach. In my opinion, having lots of unanswerable questions bumping that many times would be a Bad Thing™.
  • People answering them anyway. I know that the idea is that they will be closed, but closing bad questions doesn't happen instantaneously, so it would be optimistic to assume that closing questions marked "Under construction" would. The idea has been mooted that all questions should start out closed and require 5 reopen votes, but that might be a bigger change to the underlying software than having a separate substack.
  • It looks ugly.
\$\endgroup\$
3
\$\begingroup\$

but with two key differences: answers always disabled and posting requires 0 rep.

Why would you want to disable answers. Criticisms/drawbacks to the question could be posted as one point per answer instead of in comments. Then the SE mechanism of voting would ensure that the most important drawbacks are upvoted and need to be fixed first. This would also give more space for suggestions to those specific drawbacks (as comments to those answers).

IMO this would be better than having single questions with many comments. (The current sandbox mechanism being single answer with many comments.)

\$\endgroup\$
  • \$\begingroup\$ Because the point is to migrate questions which are ready to the main site. Deleted answers would be preserved, and still visible to high-rep users. \$\endgroup\$ – Peter Taylor Jul 16 '14 at 13:31
  • \$\begingroup\$ @PeterTaylor Precisely. \$\endgroup\$ – Isiah Meadows Jul 17 '14 at 16:20
  • \$\begingroup\$ @user80551 You do have a few good points, but answers could also serve far better purposes, especially for larger challenges. Look at my updated question. \$\endgroup\$ – Isiah Meadows Jul 17 '14 at 16:34
  • 1
    \$\begingroup\$ @PeterTaylor Oh, ok, but still, the OP can edit-> Ctrl+A -> Ctrl+C -> New question -> Ctrl+ V -> tag -> Post the question when he feels it's ready since AFAIK migration rights are limited to moderators. \$\endgroup\$ – user80551 Jul 17 '14 at 17:08
  • \$\begingroup\$ No, 5 users with 500 rep can migrate between main and meta, and there's a per-site configuration to allow them to migrate to more sites. It's handled as part of the closing mechanism (close reason: belongs on a different site). \$\endgroup\$ – Peter Taylor Jul 17 '14 at 17:17
2
\$\begingroup\$

What are your thoughts on this proposal?

Aren't you just reinventing Stack Exchange?

Let's compare an average post on another SE site, say, Stack Overflow, with the sandbox system you're suggesting.

Normal SE site                                  PPCG with sandbox
Question is posted on main site                 Question is posted in sandbox
Question is downvoted if low quality            Question is downvoted if low quality
Question is edited to fix minor issues          Question is edited to fix minor issues
Question is commented on to address problems    Question is commented on to address problems
Question is closed if major issues need fixing  N/A
Question is reopened when ready                 Question is posted to main site when ready

We're part of the Stack Exchange network, so what's wrong with acting like it? What's wrong with using the voting, commenting, closing, and reopening mechanisms that already exist instead of recreating them?

I've elaborated on this in a separate post, and the main arguments against this that I see there are:

  1. Chris Jester-Young's answer:

    Your proposal could be workable, if we are willing to weed out questions (and answers) as aggressively as is being done on the Software Recommendations site (and if we have enough high-rep users who can nuke inappropriate questions and answers quickly). That doesn't seem to be the current trend, though.

    As long as we're changing things, we might as well change that, which seems like quite a problem regardless. And in any case, the community as a whole has gotten much better at using the moderation facilities that the Stack Exchange platform provides.

  2. Dr. Rebmu's answer:

    But some questions may be in the form that people want to get some review before that.

    and, to an extent, Howard's answer:

    When we encounter a question which we think should and can be enhanced (with some effort)

    This could be a drawback of nuking the sandbox; for example, some King of the Hill challenges are quite complex, and the Stack Exchange format might not scale quite as well to those and other high-effort challenges.

    But that problem still exists in the sandbox: it's still bound by the same limitations (a single unorganized comment thread, no true collaborative editing (as in Github / the new Google docs "suggestions" feature), etc.). Perhaps it would be a better idea to use a tool built for large projects like these, such as Github, instead of trying to wrangle it into the SE format.

So why reinvent the wheel?

\$\endgroup\$
  • \$\begingroup\$ The problem with this answer is that it included plenty of similarities between the two ideas, along with one insignificant dissimilarity, which clearly indicates bias. Some huge dissimilarities were excluded. First, reopened questions on main receive less than average attention. Second, questions can be answered before they are fixed. Third, questions might be temporarily abandoned, leading to accidental deletion. Fourth, questions might be permanently abandoned, leading to clutter. We may be unable to differentiate between the last two cases if the author goes on vacation without notice. \$\endgroup\$ – Rainbolt Jul 16 '14 at 14:43
  • \$\begingroup\$ You did make some good points, and I desperately want just as much as the rest of you for the answer to be "We can already do this if we just use what we have properly." I just think the proposal was treated unfairly. The author listed a few benefits of the proposal that were almost untouched. \$\endgroup\$ – Rainbolt Jul 16 '14 at 14:47
  • \$\begingroup\$ @Rusher 1. Why is that? That hasn't seemed to be the case, at least in my experience. 2. Those answers should be comments about a loophole / deficiency in the question; therefore, they should be flagged as such (not an answer). 3. Why delete those? Just because the OP's not around to make improvements doesn't mean the community can't. 4. (see #3) If the question is so bad that it needs a complete overhaul, and the OP's not there to do it, it can be treated just like any other very low quality question (deleted). \$\endgroup\$ – Doorknob Jul 16 '14 at 15:00
  • \$\begingroup\$ 1. Trying to argue that questions on the front page receive the same attention that question that are not on the front page is silly, and not enough to push me to actually go and find examples of buried questions. 2. If the challenge changes after being posted, then answers that were potentially valid are now invalidated. 3. Good point. You opened my mind to more aggressive editing of challenges, but this also means that posting controller programs pretty much have to be done via github so they can be edited by the community as well. \$\endgroup\$ – Rainbolt Jul 16 '14 at 15:09
  • \$\begingroup\$ @Rusher 1. Reopening questions bumps them up to the front page as well, and so does editing. 2. Instead of posting an answer exploiting a loophole the OP obviously meant to close, a comment would be a better course of action. 3. Posting the controller code in the question itself would also work, if it's not too complex. But yes, in the KotH that I'm currently working on I plan to post the controller on Github. \$\endgroup\$ – Doorknob Jul 16 '14 at 15:17
  • \$\begingroup\$ 1. That's not necessarily true. Questions by default are displayed newest first. One must deliberately change this in order to see active questions. Some users might avoid this because they equate "old" questions with "stale" questions. Obviously there are exceptional questions that shine long after being posted. 2. The question could have simply been improved due to something the author did not foresee when posting, and that change could invalidate a perfectly good answer. This does not necessarily mean it was a loophole. It could just be a good question could getting better. \$\endgroup\$ – Rainbolt Jul 16 '14 at 15:22
  • \$\begingroup\$ @Rusher 1. That's true for the /questions page, but for the main page (codegolf.stackexchange.com) - the one I usually use, the default sort is by active. 2. Hmm, that could be one unfortunate side effect, but I suppose the user who posted the answer could just edit something in along the lines of "answer was posted before rule x was a thing." That could be a problem with the whole "kill the sandbox" thing, though. \$\endgroup\$ – Doorknob Jul 16 '14 at 15:34
  • \$\begingroup\$ I think no. 2 is big issue here. Of course, you could say "this was posted before change x", but then how are you going to objectively compare answers to find a winner? Regaring no. 1, sorting by "active" may at some point become infeasible if activity on this site goes up considerably. Finally, I used to really be on the side of "we shouldn't pretend we're any different from other SE sites and deal with it". But I'm shifting more and more towards "we should admit that there are a few differences". [ctd.] \$\endgroup\$ – Martin Ender Jul 16 '14 at 16:58
  • 2
    \$\begingroup\$ The main difference is that posting a question here is considerably harder than on any other SE site, so providing some additional mechanism to polish questions does make a lot of sense to me (in whatever form). Your (Doorknob's) argument quoted by professorfish doesn't seem to hold for me - on other SE sites questions can be improved, but the difference is that on other SE sites answers can be posted before a question is perfect. It really makes sense to remove clutter of unfinished questions/challenges from main, because what are they doing there if they can't meaningfully be answered yet? \$\endgroup\$ – Martin Ender Jul 16 '14 at 17:01
-3
\$\begingroup\$

I don't think this is necessary. The reasons given in other posts for getting rid of the sandbox in its current form, such as this one, seem good.

But, aren't we a Stack Exchange site? The entire network was designed so that there is no need for a sandbox. Is there not editing and closing? Questions don't have to be perfect immediately on posting; if you see a minor problem, just edit the post! Comment for major problems, and close unrecoverable challenges. Are you starting to notice something? This is exactly what other Stack Exchange sites do!

We are different from other Stack Exchange sites, but not that much. That is no excuse for a separate sandbox.

An alternative

As I suggested in my question here, you could simply post questions on the main site straight away. If it needs a sandboxing period, you could clearly mark it with a tag like or , and tell people not to post answers yet. When sandboxing is done, just remove the tag and allow people to answer.

This worked on one of my questions, Fast Trig Calculation. I posted the question in this state (first revision), and after some comments and some edits, it was good to go. (@Doorknob also closed it while it was being sandboxed - if we could automate that, then the question could de-sandbox automatically when it had enough reopen votes)

\$\endgroup\$
  • 1
    \$\begingroup\$ I didn't think it worked very well with your trig question. It's currently on revision 10: if every new question goes through half as many revisions, bumping every time, the front page is going to lose a lot of utility. \$\endgroup\$ – Peter Taylor Jul 16 '14 at 19:08
  • \$\begingroup\$ @PeterTaylor: The oldest post on the front page right now was modified almost 48 hours ago. Until PPCG gets a little more activity, I don't think that's an issue. \$\endgroup\$ – Dennis Jul 22 '14 at 18:43

You must log in to answer this question.

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