Why is it needed to vote for each rule separately?
Actually you can have an advantage when you're in a rule with a higher survival rate than the others, which doesn't seem to be justified.
Why not make one big mixed round with all the levels and then best 5 levels or so go in the next round?
A while ago, I submitted a level for a rule that had only one other submission, and both levels pretty much got an automatic pass into the next round. It is a little unfair, but I'm not sure what the alternative is. Trying to put the survival rate above 66% or so is about as fair as possible for everyone, unless it's changed so each rule is judged separately and there's no mixed group, but I don't think that will go down well and the competition will return to just a single rule per contest.
I set the survival rates for the rules as close together as possible
while also trying to result in a reasonable sized mixed group, so no big advantage spawns from this.
This was the very first thing I minimized when taking over the contests! It could be that before my time there was a more unfair rate though.
Under me the highest diffrence in survival rates was/is 16,666666 %:
Biggest differences I had:
This contest: 50%-66,666666%.
Contest 15: 33%-37%
Contest 14: 50%-66,66666%
Contest 13: 40%-50%
1 big mixed round would be a bad thing. Large groups tend to deliver inaccurate results with the current voting poll way. The results from smaller groups are more accurate with polls as putting 1 level angainst 1 other is the clearest stand-off.
For very large groups a complete different voting scheme would be needed without polls, where the users not just vote for a few favorites from a selection, but deliver a complete ranking of all levels from "most favorite" to "least favorite".
The result would be a median of all rankings sent in.
Only 1 round would be hold this way with an immidiate overall result.
But I must say I don't really like this kind of voting.
I want to make the groups rather smaller than larger in general as a result.
Although I could distribute the levels from all rules together in ~3 groups, with near identical size. Mixing itself is no problem.
All in all I decided that after this contest's voting I will make a poll for 2 things:1.) Shorten voting time from 3 to 2 days.
2.) Distributing all levels from all rules over a set number of groups with ~ the same size for the voting phase, so rules can get mixed. And survival rates are very colse together or identical, the rates themself will also depend on the number of entries.