This is already covered within the help center.

Deleting questions

Users with this reputation level can delete closed questions.

When should I delete questions?

Closed questions that are of no lasting value whatsoever should be deleted.

Before voting to delete, please check whether there are any good answers; if so, then the question should be flagged for moderator attention as a potential merge candidate. We don't like to lose great answers!

Also, be cautious when deleting questions closed as duplicates; they can serve as a signpost, directing users to useful answers on another question.

It takes 3 votes, minimum, to delete a closed question. However, the number of delete votes required scales to the number of votes on the question and all its answers.

You must wait for a question to be closed for 2 days before you can vote for deletion. This restriction is removed for trusted users when a post scores -3 or lower. If you feel a post should be deleted despite having lots of votes or for being new, please flag it for community moderator attention.

You can view a list of posts with outstanding delete votes via the moderator tools delete tab.

My understanding of the disagreement is what "no lasting value whatsoever" means... however I don't understand the objections to that.

If something is of "no lasting value whatsoever", that means it is unsalvageable; it cannot be edited by anyone including the op in such a way that it'd have lasting value.

What does "lasting value" mean?

It certainly isn't a higher bar than any of our existing close reasons, so really it just means can it be edited to no longer fit within any of our close reasons. It doesn't mean we, as users with the ability to delete vote, are suddenly the arbiters of long-term usefulness and should use these delete votes as extra downvotes. If it can be edited such that it'd be valid to reopen it, then it certainly doesn't rise to the level of hasty deletion.

Notwithstanding what the Help Center says, I want to try to derive an up-to-date guideline logically, and offer some more precise guidelines.

When is it appropriate (or inappropriate) for high reputation users to cast delete votes...?

When the question cannot reasonably be salvaged. This entails:

• Others cannot edit it into a form worthy of being reopened, while being confident of preserving authorial intent (but if you can, you should just do so instead of voting to close); and

• OP cannot reasonably be expected to edit it into a form worthy of being reopened, even with clear direction from others, without fundamentally altering the question; and

• The question does not contribute to a searchable, high-quality library in spite of being closed.

Let's look at the close reasons:

• Duplicate: Deleting these is low priority, because even when they do not function as useful signposts, they help indicate the importance of canonicals. Delete a duplicate if:

  • The title is misleading or not useful for search, and cannot effectively be fixed without more or less duplicating the title of another signpost; and

  • The question did not already attract useful (i.e.: that don't duplicate answers on the canonical) answers prior to closure; and

  • OP is not contesting the closure; and

  • a reasonable amount of time has passed (as a rule of thumb: enough that the question would be eligible for the Roomba if it weren't a duplicate).

• Needs details or clarity / Not written in English: This can almost always be salvaged in principle, even though OP might not be capable of it. Give OP a chance to communicate the same idea properly. It rarely happens, but there is not really any rush. These questions should realistically never be upvoted or locked. If they get reopen votes or an edit, they should not be deleted. So the only reason to delete them manually is if they received an upvoted or accepted answer before closure - otherwise, the Roomba will get them within 9 days anyway.

• Needs more focus: If this is along the lines of "how do I make a MMORPG?" then go ahead and delete it. The scope fundamentally can't be narrowed without altering the question. On the other hand, if it needs more focus because it's a multi-part homework question, just try to get OP to pick one part to ask about.

• Opinion based / seeking resources: The subjective stuff. Consider whether an objective metric can be extracted, whether the list of potential resources can be narrowed, whether it's actually necessary to ask for a resource in order to make sense of the question (e.g., just edit "what library can I use to do X?" into "how can I do X?").

• Not about programming / belongs on another site: This will rarely if ever be salvageable. Someone who really has a pure math question isn't going to be able to make a programming question out of it again - at least not until after getting an answer to the math question, trying to translate that into code, iterating the "identify an issue -> do debugging -> isolate MRE" process, etc. The end result of this is a new programming question that should be asked separately, not in the original spot. Do use your judgment, though.

• Needs debugging details: By definition, choosing this closure reason entails a belief that debugging details that would make the question acceptable could exist. If they couldn't, you would have voted "not reproducible" instead. Don't delete these until the 9-day mark. Try to get answers deleted first, since they're (in your assessment) trying to do the impossible, and thus can't possibly be useful.

• Not reproducible or caused by a typo: This is the counterpart that you would choose if the question can't be salvaged. Obviously we don't want to keep around questions where OP simply fat-fingered something; clearly there is no actual question underlying that. Similarly, if OP is sincerely asking why two and two don't make five (idiosyncratic logic error), there's no realistic hope of editing that into a question that could be helpful to others (even if someone else had the same bizarre thought pattern, it wouldn't be searchable).

  However, there are many simple errors that follow a common pattern, that experts will recognize over time (e.g., failing to account for De Morgan's Laws, or getting the indentation wrong for languages where that matters, etc. It's important to recognize that these sorts of things are generally not actually typos; OP has misunderstood (or neglected) something vital, easily described and explained, and reproducible. Find or make duplicates for these instead.

• Custom reason: Uh... tell you what, if you can get two other people to agree that the question should be closed for a reason that isn't covered by any of the standard reasons, and it doesn't lead to a reopen war / Meta drama etc., maybe we'll cross that bridge when we get to it.