We've known about the 'Sandbox Problem' for a long time. We have a post every six months or so here on Meta complaining about it, and yet, there have been few constructive suggestions for answers: in fact, I've only found two actually practical solutions:

  • Put the sandbox onto the main site with a tag, allowing people to answer if they wish, but also alerting people that they should downvote if it's unclear at its current point
  • Creating a new, separate site where people can post challenges as questions, rather than as answers to one question on this site.

There's also the camp that believe that the sandbox should be nuked altogether, but by my definition at least, that's not exactly 'constructive'.

So, I propose a third solution to fixing the problem, which I believe has the potential to be implemented in a matter of minutes, and solve most of the problems encountered with the current solution.

What's the problem, anyway?

The sandbox, a thread on Code Golf Meta, was first created in 2014, while this site was still in beta, and would remain as such for at least the next couple of years. Since then, however, not much has changed - most of the differences that you can see in the edit history are simply moving the Sandbox FAQ to be a part of the actual question body, and the next substantial edit was the addition of the Sandbox Viewer.

The issue is, as the site has grown, so has the sandbox. This has resulted in a number of adverse effects:

  1. Many posts get unnoticed, since there are so many new entries to the sandbox.
  2. Not enough people are actively reviewing. In my opinion, this isn't a flaw with the users of the site, but rather, a side effect of the fact that there is enough content on the main site, with about 4-5 new challenges each day, that there's no need for people to resort to checking through the sandbox, or to think "there's not enough content, let me help make some". This bystander effect essentially nullifies the purpose of posting challenges there
  3. It's too much work to actually help out. I've never given more than one sandbox suggestion in one go, since it's a lot of work to fully evaluate all potential problems, and explain them in comments. There's just too many steps involved, having to research questions, etc., just for the chance of being told 'no, I disagree with your suggestion'.
  4. There's far too much clutter. Many people either don't delete their Sandbox answer after they've asked it on the main site, but a bigger problem is people waiting for feedback, but not getting it, but also never deleting their posts, because "I just need to be patient, someone will answer me soon.". These answers get buried, though, so no-one notices, and they just remain there, forever, untouched...

The A solution (I think)

So, my suggestion is to implement a system that aims to deal with most of these problems by aiming to make the Sandbox a more algorithmic system, that's easier to deal with and in the long term, more effective. It goes as follows:

  • Split the sandbox up into 4 different 'steps', each an individual question on Meta:
    1. Writing & formatting - actually writing up the challenge, from a rough idea or a few lines.
    2. Clarify details - add a 'Notes' section, or embed more detail/make what there is easier for the general public to understand
    3. Rules & loopholes - add the 'Rules', or, if this has been done already, check for any potential loopholes that need clearing up.
    4. Final polishing - a final layer to get through.
  • Posters can post their answer into any starting category they deem appropriate.
  • Reviewers can add comments with suggestions. Up/downvotes signify whether the challenge is good or not.
    • Now, here's the icing on the cake: if an answer gets a score of -2 or less in any stage, or has a score of -1 72 hours after posting (a 'negative result' in that stage), it moves down a stage to the previous step. If an answer gets a score of +2 or more, or has a score of either 0 or 1 after 72 hours (a 'positive result' in that stage), it moves on to the next step.
    • If a question in the 'Writing' stage gets a score of -2 or less, that's clearly because the question itself is bad: entries into the Writing stage are only challenge stubs, e.g like my answer to Secret Santa's Sandbox. If these get downvoted, obviously, it's because even at the roughest level the idea is bad. Therefore, the question should be removed, and this is enforced through the rules.
  • If the answer has a positive result in the final stage, it can be posted.
  • Answers can stay for indefinite periods in any stage, although when they are moved on to the next/previous stage, the current version of the challenge is deleted, and is posted in the thread for the other stage. This means that if a challenge gets a positive result in stage 2, and while 1-2 suggestions are made in stage 3 it gets an overall negative result, it is moved back in its current form, stage 3 edits and all, to stage 2. This avoids questions just falling into cycles.

How does this solve the problems?

In order, here are my proposed ways this solves the problems listed above:

  1. This is the hardest issue to solve, in my opinion. It's just engraved within the concept of the sandbox, and the way the site works. It's a negative feedback loop: the more people use our site, the more the sandbox suffers from this issue. However, with this new system, posts will be filtered into categories. There will, ideally, be at most a third of the posts in any place at one time, so it's never overwhelming. Plus, since posts can start at any stage the posters feel, all the low-quality challenges won't just be dumped into one queue, but spread out across the threads.
  2. Many complicated reputation- and privilege-based systems have been proposed on Meta, in various places, to increase participation in reviewing. Since it's so simple here, including a simple upvote being an influencing factor in whether a challenge gets posted or not. Further, people will become more familiar with the different categories, and therefore will be exposed to a higher percentage of total sandbox posts. As once mentioned somewhere on Meta (appreciate a link if someone can find it), 'people post their question in the sandbox, then maybe scroll to look at a submission or two before leaving' (paraphrasing). If people view challenges in lots of different places, there's a higher chance of them experiencing more variety in these 'one or two' submissions they look at before leaving. One point is important to remember: you can't upvote twice! If the same challenge is at the top of the sandbox when you post your answer, it won't matter: you've already given feedback on it, and of course you don't have the patience to scroll and look at other challenges.
  3. This leads on to the next point: there's an issue with the fact that there's simply too much effort that goes into reviewing. Here, however, since it's split up into steps, you are only responsible for a fraction of the changes that need to be made whenever you are giving feedback. This system, at least its Utopian implementation, would increase collaboration, also resulting in a higher rate of non-closed posts as more unbiased viewpoints are considered.
  4. The algorithmic nature would decrease clutter, and strict rules on when to move or delete posts would be enforceable by the community. Posts can't go stale from being left in the sandbox for years on end, like in the current sandbox where the oldest post is over 8 years old (I mean, why would you delete it? It's just waiting for more feedback... 8 years later), because there are exact requirements on where and when to move posts. Even if there's no feedback, it will be moved after a week.

Just a little thought. Please don't use up/downvotes to this answer to signify agreement, rather, post as answers with your comments. I'd be interested to hear whether you think my idea has potential, or is just another proposition that's destined to fail.


This section is for possible questions to discuss in the answers. Please do give your thoughts!

  • @Adám - Who would manage the movements of posts? In my opinion, we should set a deadline of 1 week to make the transfer, and make the poster of the question responsible. After all, they are also the ones responsible for implementing change suggestions from the comments. However, a moderator/someone with deletion privileges can delete the post if the deadline is not met, since clearly, they don't care enough about their post. We could potentially even add a flag category to reduce the responsibility of the mods.
  • 1
    \$\begingroup\$ Who will manage moving answers between sandboxes? \$\endgroup\$ – Adám Aug 13 at 20:05
  • \$\begingroup\$ I'm not sure, maybe I'll start a section in the main post for 'Areas to be cleared up'. I'd say, however, they should be moved by OP, but if OP doesn't act in time (maybe within a deadline of 1 week), a mod can delete their post. \$\endgroup\$ – Geza Kerecsenyi Aug 13 at 20:08
  • 3
    \$\begingroup\$ Sounds like a huge amount of work for the mods. \$\endgroup\$ – Adám Aug 13 at 20:09
  • \$\begingroup\$ @Adám maybe we could add a flag category for it then? \$\endgroup\$ – Geza Kerecsenyi Aug 13 at 20:12
  • 4
    \$\begingroup\$ Agreed with Adám. The main problem I have with this is that it creates way more work, which kinda exacerbates the problem of people not paying enough attention to the sandbox. And also that it creates a pretty sizeable amount of work that only the 4 mods could handle (moving/deleting posts). Not to mention that requiring a challenge to go through so many steps creates more clutter, not less. \$\endgroup\$ – DJMcMayhem Aug 13 at 21:10
  • \$\begingroup\$ @DJMcMayhem The way I see it, instead of having it filter through more posts on the main site, your workload will be shifted to the sandboxes, since it will create higher-quality posts. Plus, it won't create more clutter: if more people use/comment on posts in the sandbox, for the reasons I described before, it would result in more community peer-moderation about these sorts of things. Introducing a stricter, more algorithmic sandbox would mean changing the 'tone' of the sandbox, and in turn changing people's attitudes towards just dumping and leaving content there. \$\endgroup\$ – Geza Kerecsenyi Aug 13 at 21:14
  • 1
    \$\begingroup\$ I thinking making it more algorithmic would make people more likely to skip it to avoid the hassle. Especially since the sandbox isn't mandatory (nor should it be) \$\endgroup\$ – DJMcMayhem Aug 13 at 21:31
  • \$\begingroup\$ @DJMcMayhem if you look at similar questions, especially [this] (codegolf.meta.stackexchange.com/a/1791/85546) answer, I think we can see that people currently skip the Sandbox exactly for the 4 reasons I gave in my question body. Making it more of a 'structure' than merely a, well, sandbox, would help more people trust the result. Right now, for instance, let's say I don't get any comments on a post in the Sandbox, or any votes. After 72 hours, can I post it? How about after 1 week? This structure will address the uncertainty, and help people better gauge the interest in their posts. \$\endgroup\$ – Geza Kerecsenyi Aug 13 at 21:40
  • 4
    \$\begingroup\$ I'm not convinced that splitting the sandbox process into stages will lead to less work for reviewers -- I think it will actually makes for more. It's not actually less effort for me to consider only a certain kind of problem for a given step, since I'd need to read through the whole post anyway, and all kinds of issues would come to mind. I wouldn't know if these issues are things that the poster was going to clarify anyway at later stages. And then for later stages, I or someone else would need to read the fleshed out post, duplicating work. \$\endgroup\$ – xnor Aug 14 at 6:41
  • \$\begingroup\$ @xnor you're ignoring the current problem: there's only about three people like you, who actually do reviews currently, on the platform. If we could reduce the stress on each individual reviewer, and get them exposed to more questions, that would be a win in my book. Currently, people aren't writing any comments. Plus, with the algorithmic system, the overall *result*/success rate will be better, as it works for the interest of the greater good. - maybe, you could write the comment most applicable to the current section, and then use up/downvotes to signify the intensity of other problems. \$\endgroup\$ – Geza Kerecsenyi Aug 14 at 10:29
  • 3
    \$\begingroup\$ @geza Adding more steps to the process won't make more people magically start reviewing. It just means the people who are reviewing are more likely to miss any given post. \$\endgroup\$ – DJMcMayhem Aug 14 at 12:46
  • \$\begingroup\$ @DJMcMayhem Simply no. People won't miss more posts, but instead will be exposed to more posts. Instead of looking from your 42K-rep, moderator, 5-year usage of the platform, you need to look at it from the POV of an average user. A user who isn't stupid, or lazy, but just doesn't want to spend time on researching, thinking, testing and writing answers to the sandbox. I could easily make one suggestion to every item in the sandbox, but if I only make that one suggestion, I fear that the OP will think that I can't see anything else wrong with their question, and post it with just that one fix \$\endgroup\$ – Geza Kerecsenyi Aug 14 at 12:51
  • \$\begingroup\$ @DJMcMayhem Would it not be possible to run a test, and maybe work out the amount of questions that get upvoted/not closed after coming from the old sandbox, vs this sandbox? It's worth a try, in my opinion, since currently all we have is a bunch of complaints (literally over a dozen) and very few actual suggestions for solutions. \$\endgroup\$ – Geza Kerecsenyi Aug 14 at 13:13
  • \$\begingroup\$ @GezaKerecsenyi punishing sandbox users (who come to the sandbox just to improve their idea) and making the "legislative" complexiness so high will distract bascially everyone from the sandbox and shitty ideas will spill all over the codegolf instead of staying in the sandbox. \$\endgroup\$ – Krzysztof Szewczyk Aug 14 at 17:43
  • \$\begingroup\$ @KrzysztofSzewczyk in what way are we 'punishing' them? \$\endgroup\$ – Geza Kerecsenyi Aug 14 at 20:49

You must log in to answer this question.

Browse other questions tagged .