it might be interesting for post analysis and we could define a new metric if something like that happens, which I'm not expecting as the paranormal bias should be evenly distributed for each word guess.
Correct the only way I can pass is to get the OCCASIONAL 3/3 and probably LOTS of 1/3 too.
There are several models to calculate the odds, but monte carlo simulation is quite accurate.
Say for 100:1 odds, 10 batches, you type in 10, 100 SIMULATE
Do that 10 times, write down the 10 high scores.
I got:
BATCHES = 10
SIMULATIONS | ODDS = 100
HIGH SCORES
17
20
18
18
19
18
17
17
18
17
so the middle value is 17 or18
So the pass mark for 10 batches would be 19 words right from 10 batches of 3.
That would prove supernatural with 1 in 100 chance of error!
Let's make 2/3 on average the pass mark for atleast 10 batches!
So you want to set up a 100:1 ODDS test?