It depends on how recent the words are. If you are concerned that your meaning will be unclear, by all means use the hyphen. Words like reentry and reelect have been in usage for a long time and pretty much no one has a problem with them. Reexcite has not, so you would do better to stick with re-excites. If you want to edit something again, you are probably better off to re-edit it if you are worried that reedit will cause the reader to stumble.
Where the prefix is not followed by a vowel, however, you are not honor-bound to add the hyphen. If you want to repolarize something, go right ahead. According to Etymonline, the word repurpose is less than 30 years old, having dropped its hyphen somewhere in the early 1980s. So if you are the first to repolarize something, you are not likely to be the last, and you will probably start a trend. In any case, your meaning will be clear.
This is hardly some sort of unique New Yorker idiosynchrasy. Rather, it’s merely an older tradition.
Writing Zoë, Noël, reëlect, coöperate, zoölogical, mosaïc, aïoli, cacoëpistic, hyperoödon, haliæëtos, naïve, Moët, naïveté, Thaïs, monoïdeism, panzoöty, Laocoön, langue d’oïl, Boötes, faröelite, caïque, &c is actually an older orthographic style.
This style fell into disuse as people subjected to the infinite tyranny of typewriters (and their ASCII derivatives) bereft of proper diacritics got out of the habit of using things they didn’t know how to type. I rather hate seeing *Zoe written by people who are too lazy to write Zoë properly.
On the other hand, English has never used a diaeresis in words like coalesce, where one might expect *coälesce to differentiate from something related to coal. Similarly, we’ve never used *reäct for react. There is also the tendency to drop hyphens and diacritics from assimilated words. Notice it’s just zoology in modern writing, not the original zoölogy, and you almost always see dichroic and dichroous instead of dichroïc and dichroöus.
I’d sometimes rather see a diaeresis than a hyphen myself if it comes to that. I actually prefer writing coöccurrence, coössify, demosaïcking, reënact, reïnvent (not a rein-vent, whatever that might be), reëxamine, intraätomic, pal(a)eoöceanography, proöxidant, &c to make it clear what’s afoot, where others might use a hyphen or nothing at all. I’m trying to be considerate for the reader, and consistent.
And yes, coopering is a real word, one with three syllables, though, not four. Hence the preference of writing coöperate to show that it has four syllables instead of just three, and is unrelated to coopers.
J.R.R. Tolkien, a very careful writer of English from a slightly earlier era than our own, uses the diaeresis a fair bit in his writings. Consider the raven Röac from The Hobbit; its name is meant to be bisyllabic onomatopoeia for its croaking cry (or is that croäking? :). Or namarië, Eärendel, or Manwë; written using Tolkien’s own phonetic tengwar script, these need no markings to indicate those vowels should be pronounced, but as an aid to English-speaking readers using the Latin script and used to “silent” letters, it helps to mark these explicitly.
By the way, the word diaeresis is so spelled in the Unicode Standard. You will find older books that spell it diæresis, which uses yet another non-typewriter letter. And it is not uncommon to find it spelled dieresis in American writing.
Best Answer
Neither is inherently "better". Take, for example, crab-wise, which appears thus in A thesaurus of traditional English metaphors (published only 20 years ago), despite the fact that most dictionaries discarded the hyphen long ago.
As a general principle I think it's better to avoid unnecessary hyphens, so I'd say if you see the unhyphenated version of anything in any credible dictionaries, just copy them.
I shan't feel hard done by if this question is closed as a duplicate of When is it necessary to use a hyphen in writing a compound word?, but I do think the specific issue of "What if it's not in the dictionary?" isn't really covered there.
Even more specifically, "What if the dictionaries disagree?" isn't covered. For example, most dictionaries hyphenate hard done-by, but I'm not going to take issue with Collins in my first link (who didn't). In that particular case one would avoid collapsing it into hard doneby because it would be tricky for the average reader to recognise the component words.
EDIT: I can't believe nobody has yet voted to close this as a dup. But here's another candidate - my own Can word-hyphenation ever be semantically significant?.
What I find fascinating about that one is most votes support answers saying the hyphen can be significant. After thinking about it, I've decided it probably can't, really. If context doesn't already imply the relevant word associations, it's some kind of orthographic game, not real language.