When each tie results in 1 extra level going through as tiebreakers are not being run, a doubling of the survival rate to 2 would equal at least in 8 survivors.
In the last phase we had a 60% increase in survivors due to ties. Let's estimate low and say only 2 extra levels would go though (25% increase). This would lead to 10 survivors here.
A single group of 10 to vote on I would call too large and I was only doing it once in the past extreme case of ~ 90 nominations in last years LOTY 1st phase. (also wasn't too keen on it, but even the group count was very high)
So we would have an extra phase of ~ 10 levels to vote down to the final group of ~ 4-6. The point of no tiebreakers was to make the voting process faster. So in this case we implemented no tiebreakers, only to end up with more voting rounds because of it. This is counterproductive.
The likelyhood of ties may need some more data points, but this should influence the survival rate.
The low survival rate also is in line with statements from the
poll thread:
From other users:
...
Secondly, reduce the survival rate.
...
From my new guideline:
This way I will cover the following critiques:....
...
- Reducing of the survival rate
...
In user feedback:
...
Fewer rounds with long voting times is good. Reducing survival rate is good.
...
So whenever I have to choose between a higher and a lower survival rate, I will tend to the lower one.
"if second place is only behind by one vote"
I want to keep the voting free from extra clutter rules as well. Extra "Ifs" should be avoided if possible.