"I would also like to reiterate my claim that Osuka is sort of obviously town" - rc about scumsuka
"don't tell anyone, but there's a reason why you're one of my favourite people to mod for" - datisi
"I would also like to reiterate my claim that Osuka is sort of obviously town" - rc about scumsuka
"don't tell anyone, but there's a reason why you're one of my favourite people to mod for" - datisi
In post 444, Aureal wrote:
That vote and Python's willingness to sheep Kokichi's meta read after remembering the one game we played with Black had her as mafia are both sticking out to me as particularly fishy.
Also in case it wasn't blatantly obvious when I meta'd Dragon despite never playing with him; I spectate the newbie queue, so I have more than just the one scum!Black game to go off of
My thoughts:
-push on me felt a bit scummy.
-flip flop on Black could be scummy.
-HOWEVER, their explanation of the flip-flop and their subsequent push on Black felt quite townie on their mindset and tone.
Overall read is null-town.
In post 447, IceDragon70 wrote:
Okay I tried to post in catchup style but I feel it's gonna take too long to cover over 10 pages.
So I am just gonna ISO the 3-4 players that stuck out the most for me and see whichever of them I wanna vote now.
If ya'll want to point me to a specific development or post feel free to do it though.
208 (explained in 268) and 294 (for the start of the black push) are the big stuff that happened since then
It's actually optimal to lim every day. Since scum targetting the same person stops the nightkill that leaves a
G
/(
T
+
G
) chance of shooting the other groupscum, and a 1/(
T
2 + 2
T
) chance of both targetting the same townie if a townie is getting shot
The chance of limming scum is
S
/(P - 1)
Plugging in the numbers that means that for now a lim gives an expected result of 8/12 chance of limming town whereas the chance of scum nightkilling town (Ignoring any doctor plays) is 9/11 - 1/99 = 80/99
Spoiler: What the letters mean
P - number of players alive
T
- number of town alive
S
- number of scum alive
G
- number of enemy groupscum alive
-Nameless
Ugh, don't do mech while tired; the math for the nightkills is off by like 0.5% because I didn't properly adjust the "same target = no nk" and didn't account for the doctor. Just listing out every possible action assuming a No Lim gives the following nightkill chances:
Town Doctor
: 240/2484 (9.66%)
Vanilla Townie
: 1760/2484 (70.85%)
Fire Mafia Goon
: 242/2484 (9.74%)
Ice Mafia Goon
: 242/2484 (9.74%)
Which when combined gives the chances of:
Town
: 2000/2484 (80.52%)
Mafia
: 484/2484 (19.48%)
Compared to the elim chances (
Town Doctor
can't get elimed because they just claim):
Vanilla Townie
: 8/12
Fire Mafia Goon
: 2/12
Ice Mafia Goon
: 2/12
Which when combined gives the chances of
Town
: 8/12 (66.67%)
Mafia
: 4/12 (33.33%)
In post 175, Black wrote:
Whatever. Now it's your responsibility to translate the mech for us >:(
It's more optimal to get scum killed through elims instead of nightkills and the reason this math needed to be done at all is because the mechanic where scum targeting the same person stops the nightkill will only ever get triggered on a town nightkill (since if it triggers on a scum nightkill it means that scum targeted themselves)
-Nameless
Btw are you a hydra or what? Cause I saw you playing im the Newbie queue where hydras aren't allowed, but you sign your posts.
In post 455, IceDragon70 wrote:
Btw are you a hydra or what? Cause I saw you playing im the Newbie queue where hydras aren't allowed, but you sign your posts.
We're plural, it's basically an irl hydra. There's a plurality thread in the speakeasy if you wanna know more
-A
In post 456, IceDragon70 wrote:
Also isn't it obvious it's optimal to eliminate every night in multiball?
The mechanic where both scum targetting the same person stops the nightkill means that scum are more likely to get shot at night (Because scum aren't gonna target themselves). It's not strong enough to offset the inherently lower chances of scum shooting other groupscum in this case but if the effect was stronger then No Eliming every day could have legitimately been the best strategy
In post 456, IceDragon70 wrote:
Also isn't it obvious it's optimal to eliminate every night in multiball?
The mechanic where both scum targetting the same person stops the nightkill means that scum are more likely to get shot at night (Because scum aren't gonna target themselves). It's not strong enough to offset the inherently lower chances of scum shooting other groupscum in this case but if the effect was stronger then No Eliming every day could have legitimately been the best strategy
-A
To make it more explicit, the "same target = no nightkill" rule is functionally equivalent to having every townie be 7.5% Bulletproof if there's no elim d1. If that 7.5% was instead say 75%, you can start to see how no limming might be more optimal
I have been informed that it's not actually 7.5% bulletproof because the correct way to calculate the bulletproof percent is to compare the number of townkills with and without the nightkill rule. It's apparently 9.42% (208/2208) bulletproof.
In post 257, usesPython wrote:
Also glad to see you're ignoring 223 when you're just straight up wrong in 247, explains a lot
-Nameless
223 is entirely based on false premises/nonsensical assumptions (ie, scum shooting at random, doctor healing at random, groupscum have =rand accuracy, whatever youre saying about "full control over the lim"?) and your other posts considering nolims are pointless bc why would we ever do that. these mechposts are a waste of time at best and an active distraction from any actual scumhunting/townplay at worst
In post 257, usesPython wrote:
Also glad to see you're ignoring 223 when you're just straight up wrong in 247, explains a lot
-Nameless
223 is entirely based on false premises/nonsensical assumptions (ie, scum shooting at random, doctor healing at random, groupscum have =rand accuracy, whatever youre saying about "full control over the lim"?) and your other posts considering nolims are pointless bc why would we ever do that. these mechposts are a waste of time at best and an active distraction from any actual scumhunting/townplay at worst
You've never done serious mechanical analysis of setups and it really fucking shows. Random actions and random accuracy is the norm when it comes to actually calculating EV and no shit no limming should be considered when there's a rule that's mechanically equivalent to "All townies are X% Bulletproof"
Actually you know what, since you're apparently so good at mech why don't you tell the rest of us how you'd actually calculate the EV for limming vs no limming if actions aren't random and reads aren't random.
In post 461, The Bulge wrote:
your other posts considering nolims are pointless bc why would we ever do that. these mechposts are a waste of time at best and an active distraction from any actual scumhunting/townplay at worst
In post 162, Black wrote:
I don't really think python "main hunting" is AI though
In post 326, Black wrote:
Convenient of you to ignore my first point. Using meta as the foundation of your vote is not town behavior
Python's mainhunting was explicitly used to provide meta to base their Dragon vote on. If using meta as the foundation of a vote is not town behaviour, then you should think it's AI, isn't it?
If actions and reads aren't random then the double target = no nightkill rule is even more likely to go into effect, go do the damn math if you're gonna hold these positions
In post 332, HighPrincessErinys wrote:
This one is going to hit up Black's ISO in DN shortly because ok yeah a bit of an actual scumread is starting to form, but we'll see.
In post 335, HighPrincessErinys wrote:
Okay first big red flag on Black (or black flag, it guesses) is that none of her DN solving effort really seems to be present. Maybe that's just a consequence of how DN's setup was, but... This setup is kinda similar in it's extra emphasis on day-time analysis to find scum since there's no investigatives, which even
In post 466, usesPython wrote:
If actions and reads aren't random then the double target = no nightkill rule is even more likely to go into effect, go do the damn math if you're gonna hold these positions
-Nameless
my position is that your math is a waste of time and a distraction from actually doing anything useful, and also that no limming is a stupid idea
Cause like it's pretty blatantly obvious you're still scumhunting like it's regular mafia without ever reconsidering why the tells you're using worked there in the first place.
In post 468, The Bulge wrote:
my position is that your math is a waste of time and a distraction from actually doing anything useful, and also that no limming is a stupid idea
Great cool you think this, how about you actually run it to the logical conclusion. It's +EV for everyone but the groupscum being limmed to lim scum, the only scum motivation to "waste time and distract from actually doing anything useful" is if it stops groupscum being limmed. How about you actually go back to those posts and see the only person being pushed then was Drew who's wagon I immediately got on after the posts and get out of your shit tunnel