I picked a range of versions from almost the earliest to almost the latest. I guessed at what might be chosen such that the poll histogram gave a meaningful result. As I see it, the main point of the poll idea is to tailor the choices based on feedback, prior to the actual poll being published.

I ended up with eight versions which I believe gives suitable coverage. I didn't think listing every base version (i.e. either 5.00[0-5] or 5.x.0 where x >= 6) would have been a good move, or even allowed by the pollsters; similarly, listing only a few versions would not have given a meaningful histogram result.

The short, parenthesised comment after each version is, as I stated, a "potential reason" for making that choice. It's not intended as the main reason, nor does it need to factor into a user's decision at all.

I'm more than happy to make changes (additions or deletions) to the list of options. This could be the result of a strong argument or a consensus of opinion. I'm unlikely to add a version because it introduces an experimental feature; I'd need a better reason.

• lexical_subs — experimental in 5.18 (perl5180delta: Lexical subroutines); stable in 5.26 (perl5260delta: Lexical subroutines are no longer experimental).
• signatures — experimental in 5.20 (perl5200delta: Experimental Subroutine signatures); stable in 5.36 (perl5360delta: use v5.36). Note this extract (with my emphasis): "Introduced in Perl version 5.20.0, and modified several times since, the subroutine signatures feature is now no longer considered experimental.".

I agree with ++hippo's comments regarding your last two dot points.