Send each thin page down one of three branches, and let a clear test pick which. The flattening move is to treat thin content as a single problem with a single answer, usually deleting it or padding it with words, when the right action depends on what the page is and what it could become. Merge, rewrite, or delete are genuinely different calls, and each has its own pivot.
Merge when the page duplicates a sibling. If two or three thin pages cover the same topic and compete with each other, the fix is to combine them into one stronger page, redirect the retired URLs to it, and update the internal links that pointed at them. This is the right branch when each page has something worth keeping but none is strong enough to rank on its own; consolidation concentrates what was scattered.
Rewrite when the topic has real demand and the page can be made to deserve it. A thin page on a subject people actually search for, one that ranks on the second or third page or carries backlinks worth preserving, is an investment rather than a deletion. Here you add the depth and substance the page was missing, because the potential is real and the page can be brought up to it.
Delete or noindex when the page serves no one. No traffic, no links, no demand, and no path to becoming useful means there is nothing to preserve. The one caution is backlinks: a page with links worth keeping should be improved or redirected rather than removed outright, so the equity is not thrown away.
So the editor assigns each thin page to merge, rewrite, or delete, using the per-branch test to make the call and never parking it in an undecided pile where it weighs down everything around it.