I'm going to posit an answer that's more or less a combination of what was posted above coupled with recommendations following my own experiences associated with attempting to answer these kinds of questions over the past year:
Demonstrate your familiarity with the system
Our comments, suggestions, answers, etc. fall upon deaf ears if the querent cannot be bothered to learn the system they are proposing to homebrew well enough to ask a question that is coherent to other users of that system. For example, in 5e, if you aren't familiar with what Actions, Reactions, and Bonus Actions are then you should probably hold off on trying to homebrew an entirely new class until you get that figured out.
Your question, as written, should demonstrate knowledge of the system that you're working in. Treat the game's terminology the way you'd want it treated were it to be read by any other user of the system; certain systems the criteria is very well-defined, others, less so. Regardless, if 90% of people playing that system interpret your question in differing ways that's a clear sign that the querent has not posted a useful and answerable question and should be downvoted/closed.
Let it breathe
So many times do I see someone post a homebrew that's going to require more than 10 minutes to knock out an answer for. I don't have time for it right now so I move on. Within a couple days, there's several answers that address multiple different issues; one is deemed accepted; and then there's a second draft issued as a subsequent question. Half the time, the question has also since been closed (probably because each answer seemed to focus on a different issue and thus the question was unclear).
Many times, the answers that did get in do not address some underlying issues and as a result, those issues carry forward into the next iteration. Furthermore, it complicates review of the second draft because you now need to refute the changes that were made since they didn't consider the underlying issue.
In general, I think we need to establish some sort of timeline for iterations as well as asking new iteration questions when an existing question is on hold. I think 1 week is reasonable for new iterations, but others are likely to disagree. I also don't think you should be permitted to post a new iteration if the original question was closed; edit that question so that it can be re-opened and new answers can be provided.
Tell us why
So many homebrew questions are posted without any sort of stated goal. This goes back to my first point, if I don't know what problem with the system (perceived or otherwise) you're trying to fix with this homebrew it's very hard to provide useful comments.
Tell us your process
I'm not saying that you cannot create new mechanics with homebrew, but if you elect to you need to fully explain the process by which you created those mechanics. Want to create power points in a 5e system, explain what logic you're using to balance that (perhaps the DMG's alternate rules on spell points).
If you're electing to create something out of whole cloth, I hope you can demonstrate some results of the playtesting you've done. Because if MacGuffin summonings are based on the phase of the moon in the Realm of Thanus, I can all but guarantee you that nobody else out there has done this already, so we are going to be more reviewing your playtest results and helping you identify ways to improve playtesting.
Comments plus Downvote/Delete answers on poor questions
This may be the most unpopular suggestion since it targets respondents instead of querents, but we get a lot of answers to poorly phrased homebrew questions that are assuming a lot of unstated details. This disincentivizes subsequent and perhaps more thoughtful answers, but it also puts the querent in a weird spot with regards to editing their question to be more thorough. We don't want edits that invalidate existing answers, but if the question was poorly phrased, then it needs to be corrected before useful answers can be posted.
Comments, downvotes and delete votes are the tools we have available and should be using them on answers that may be premature. I believe diamond mods also have the means to denote deficiencies in an answer and should be using them in these scenarios; flags should be considered to help draw diamond mods to those premature answers.