We've recently seen a number of incidents where a user posts essentially the same problem several times.
Sometimes, they delete the old question and post [almost] exactly the same thing again, essentially rolling the dice and hoping for a better result.
Other times the existing question remains, and a new one makes only trivial changes
Both of these practices seem inefficient and unhealthy for the site.
Deleting questions deletes the responses and analysis which community members have already put into the problem, wasting effort by forcing the process to start anew
Even if the original isn't deleted, a new posting still splits the context and collected knowledge of the problem, especially if the old isn't linked
Often what are intended to be "new-strategy" re-postings of a problem actually still preserve the same technical misunderstanding at the heart of the original difficulty
It is, however, true that:
getting a question originally stated in a poor form taken back off hold can take time (or simply not happen, if edits do not actually resolve the reason it is on hold)
poorly stated problems may receive a frustrating lack of response, or only responses which explain what is wrong with the question
Stack exchange isn't meant for evolving discussions and so cannot handle evolving problem statements; at the same time, it's really meant for questions which can be concisely answered, not project-arcs which need the evolving discussion of a traditional discussion forum.
How do we deal with these? To a large extent, finding duplicates (and especially re-posts of deleted questions) depends on site users simply remembering that we've interacted with unique aspects of a problem in recent memory.
To be clear, this is a question for the community as as whole. It's our site, and up to us to figure how to make it work well.