Not completely, but yes...the number of voters is a big contributor but how those votes are applied are also very important.
The application of votes matters only in the ways that I mentioned previously. In particular, the number of votes in any sequence that ends the contest is precisely
(Initial # of contestants * 20) - (# of points the winner has at the end)
+ (# of times a downvote is applied to a contestant with 1 point remaining)
No other factors matter.
If we generalize the formula to
(Initial total # of points for all contestants) - (# of points the winner has at the end)
+ (# of times a downvote is applied to a contestant with 1 point remaining)
then it is (fairly) easy to prove this by induction on the length of the voting sequence (number of votes).
One person has a +1/-2 (net -1) per day, but 2 people can have +2/-4 or +1/-3 per day, if their votes overlap. It's still a net -1 per voter,
Yes, that's what matters.
but the difference is whether two separate options get reduced by 2, or one single option gets reduced by 4. Three voters? Three options reduced by 2 each, a single option reduced by 6, or some combination thereof (and the possibility of overlapping).
Which makes no difference in the end.
And the contest ends when a certain number of options are eliminated, not when a certain number of downvotes are made.
Yes, but those two quantities are related through the number of points.
I'm bad at explaining this stuff. :-/ But you can model this stuff in Excel pretty easily;
I'm not sure what you mean by 'model' - you could be talking about either some sort of sampling or some procedural way to generate all possible voting sequences from some initial state. The former could be misleading; the latter, if done correctly, should produce results in line with the formula I gave above.
I've found that the contest ends quickest when everyone "focus fires" their upvotes on a single option
Yes, each vote that upvotes the eventual winner reduces the number of votes needed by 1.
while downvoting the lowest-scoring option.
Nope, doesn't matter. Eliminating all the others takes a fixed number of downvotes regardless of order. Suppose, for example, we start with 5 contestants A, B, C, D, and E, and each has 20 points to start, and everyone always upvotes A. It will take exactly 4 * 10 = 40 votes to eliminate B, C, D, and E regardless of the order in which downvotes are applied to B, C, D, and E. This seems kind of obvious.
Which will never happen in practice, but the theory is sound.
AAANNNYYYway, back to the contest. Long live the Crow!
The crow suffered a slow and painful death after you cut off his bill to make that silly weapon.
DISCLAIMER: Yeah, I said all that like I am infallible. I have overlooked things on occasion, but at least for now, I'm convinced I have this right.