So.. What should I actually be doing?

[RadioFM]

Learn about paleography; the different handwriting scripts/styles around the time of the radiocarbon dating for languages such as Latin, Romance and Germanic languages, Greek, maybe English. I'm not sure about other language families.

Why? Voynichese is obviously an invented script, the vellum has been dated, so it's very likely no older than early 1400s. Knowing that the scribe was probably familiar with Humanistic minuscule gives zero new information. I frankly cannot reliably tell Currier A from Currier B, and I'm quite happy knowing that there are people out there who can and do investigate all these hands and scribes, but I won't pretend being able to understand something that I don't.

More generally, I think it makes sense to thoroughly learn things that are relevant to your line of investigation, but it's quite enough just to be aware that some other domains exist without trying to cover everything.

I agree that it being an invented script, it's not strictly necessary to learn these things in order to have a shot at decoding/debunking it. But we do see some symbols and handwriting style that was similar to what was used in plaintexts at the estimated time of writing, and many people, especially those coming from a purely statistics/computational background, may not initially be aware of this.

I think that the symbols themselves could potentially give us an insight into what the scribe(s) were thinking about when coming up with this script. Do they use abbreviation symbols as a subterfuge in order to make us think it's a syllabary? Were they thinking of a certain letter or phoneme when adding a macron over qo or y? I think having at least a superficial knowledge on this topic can only improve the odds at figuring out the nature of script or any encoding mechanism - not to mention that if it turns out to be a cipher involving any transposition at all (which we don't know for a fact it doesn't), any would-be decipherer would find it handy to be able to read some of the characters.

At least if I was tackling some, say, chinese version of a VMs, I would find it productive to learn any cipherglyph that may also be used in plaintext, instead of just treating each glyph as random strokes and potentially missing out on a hint.

I wholly agree that there's probably no need to become a lifelong expert on it, and that eventually you'll just be met with the fact that the text reads like gibberish, but some base knowledge on the topic doesn't hurt

Read most (if not all) threads in the Analysis of the text forum and Marginalia, even the old ones

Read those that you find interesting/relevant 

I mean, on the whole, yes. It's just that sometimes I've come across threads that dont catch my eye at first glance, but then when I delve into the comments (specially when they digress), they casually drift into fruitful exchanges about topics that do interest me.