realistically it would be more like.
okay let’s just use a standard, no-conversion singleball 17p list for reference
which would look like this:
1 King
4 BD/HoB (heroine guranteed for bd)
1 Duke
1 Marshal
9-10 Random Unseen
0-1 Neutrals (could do 1-2 but feels a bit excessive with 4 starting scum)
i think the big restrictions you would have to place would look something like:
- prestige capped at 2 (the only reason it isn’t is just for host flexibility, so lmao)
- there can’t be an Evil King and a Neutral in the same game (so ek forces 0 neutrals, not the other way around)
which i think is ~reasonable; you Can meta that the king is good if a neutral flips but it’s not that big of a deal imo
i would also just straight up remove the King for singleball lists but i haven’t checked to see which roles would need adjustments and i kinda don’t want to, so i basically put it at 0-1 neutrals b/c 2 AND ek would be Excessive. if you wanted 1-2 neutrals list that one would be the restriction
if you could figure out what roles to change for no-king modes and all of them worked COUGH COUGH CHANCELLOR (that one’s not even that bad but) then i would opt for a 1-2 neutrals list with no king
in addition, i would place restrictions on Unseen killer roles according to internal evaluations of their strength (arbalist being the strongest at 1); i fear that they would end up becoming tos role-bucket-y but if you weight them you could probably just do 1 < x < 2 and have it be ~fine. the weights would have to be relatively dynamic (as in, open to adjusting them in balance patches) especially given the dubiousness of roles like Fanatic but it works better than having no restrictions
i would probably just say that there can’t be more than 4 protectives as well, or 5-6 [offensives + protectives], not sure which. most of the roles once you factor out killers tend to be investigative (which, if we’re doing automation like this, can probably just be unrestricted), offensive or protective (which serve ~similar functions i think).
capping at those threshholds seems reasonable. you would probably have to weight roles here as well, but to a much lesser degree (e.g. minstrel is 0.5, plague doctor is 2 (and probably unique tbh), alcoholic is 0.5, bureaucrat is 2 or 1.5 or smth). so i think putting the cap at 5 [offensives + protectives] is reasonable. idk if it would actually effect anything very often given the fact there’s only 10 max random unseen slots but idk
investigatives are ~reasonable enough to just put no restrictions on so. and you also gotta fill up the list somewhere
this all gets kinda fucked up once you rand roles like Plaguebearer and Sibyl (since those basically count 0.33 for each category (sibyl maybe 0.5)) but given the fact this is for theoretical rand automation i think you could probably just ban that pair and get rid of most of the problems. even aristocrat doesn’t really need to be factored in anywhere, since its abilities are “weak investigation” and “very conditional kp”
