[SETUP] The One Day Scum Hunt
- callforjudgement
-
callforjudgement Microprocessor
- callforjudgement
- Microprocessor
- Microprocessor
- Posts: 3972
- Joined: September 1, 2011
[SETUP] The One Day Scum Hunt
Inspired by mith's Red Flag and zoraster's The Fortnight. The basic aim here is to recreate a really high-stakes D1 in a way that gives town a decent shot at actually winning it.
EV is 10/21, or around 47.6%.
The voting mechanic of "two lynches but votes for one don't count against the other" is something I've been thinking about for a while, and works well in this style of setup. Unfortunately, the issue is that you need >⅔ of the playerlist to be voting to lock in a lynch, not >½, which makes the setup vulnerable to "scum refuse to bus even if it makes them obvious and deprive the town of their chance to hammer" strategies. With only 11 townies (this number is chosen for EV purposes; I wanted 4 scum), 6 votes to lynch would be too many as the townies only have 11 votes total, so I had to knock it down to 5½ (which also gives Doublet Voters a purpose to exist, making the half-vote possible, while also perhaps leading to more interesting VCA; it also increases the chance that town can arrange themselves onto the wagons in a way that works). Disallowing unvotes means that if a player becomes fairly obviously scum, you can pile 5½ town votes onto them and 4½ onto a random townie that scum is currently voting, meaning that town can get a hammer even with one townie inactive. However, I'd recommend using plurality lynches just in case town who have identified scum have trouble getting over the finish line.
I'd recommend a deadline of 15 days (matching the player count of 15 is aesthetic) and a pretty short prod timer, as almost everyone will need to be active to make this work.scum· scam · seam · team · term · tern · torn ·town- mith
-
mith Godfather
- mith
- Godfather
- Godfather
- Posts: 9267
- Joined: March 27, 2002
- Location: McKinney, TX
- Contact:
I like the idea, but needing such a high ratio of town votes for a hammer may make things impractical. Wouldn't the idea work with 8 to lynch and each player having two totally distinct votes (a "Lynch A" and "Lynch B")? Players could still double up, but if there is a clear leader they could then choose to all move their Lynch B votes to someone else. (Add the condition that a hammer consists of A and B reaching 8 votes simultaneously on two different players, to prevent scum from pushing a B lynch on the same town player once A hit 8.) This would also fix the potential issue of *three* players hitting 5.5 at the same time (A is on 5.5, B and C are on 5, someone switches from D/E to B/C).
(And FWIW, it's more similar to The Bus Stop 2 - town wins if one of the first two lynches is scum - than Red Flag. )- callforjudgement
-
callforjudgement Microprocessor
- callforjudgement
- Microprocessor
- Microprocessor
- Posts: 3972
- Joined: September 1, 2011
It was inspired by the gameplay of Red Flag, rather than so much the mechanics. The Bus Stop makes it hard to put a high scum density in the game, whereas I feel that that's required here. (In case you missed it, Red Flag just had a test run, with fairly mixed reactions to the setup. My reaction was that Day 1 was a lot of fun but the rest of the game wasn't really, so I tried to preserve the feeling of that part of the setup.)
I tried to make a Doublevoter version work (i.e. equivalent to Doublet Voters but with the numbers doubled and without the ability to contribute twice to the lynch of the same player) but it'd require 11 to lynch to prevent a hammer contradiction. I guess you could just reduce the lynch threshold while not worrying about the "three wagons can viably hit 8 at the same time" problem. It's possible in your version of the setup to have four players lynching player X with vote A and player Y with vote B, four players lynching player Y with vote A and player Z with vote B, and four players lynching player Z with vote A and player X with vote B, meaning that three players are in a hammered state simultaneously. That's normally undesirable, although you can make it work.
Three 5½-votes at the same time are impossible; there are 15 players so if 11 votes are tied up on two of the players, a third player can have at most 4.scum· scam · seam · team · term · tern · torn ·town- Mathdino
-
Mathdino Survivor
- Mathdino
- Survivor
- Survivor
- Posts: 14337
- Joined: February 24, 2013
- Location: Right Behind You
I might make the Mafia titled Mafia Doublet Voter Lovers, since the game ends if any are lynched. Cuts down on the extra wincon descriptions. But that's not really relevant.
I like it!
I think it's scumsided though :/
An EV of under 50% is a red flag (no pun intended) for an all vanilla setup, which MS consistently performs incredibly poorly on. White Flag's actual winrate is half its expected winrate.
Add to the fact that D1 lynches onsite are essentially random to the fact that mafia is incentivised to not actually let mafia get lynched, and we get something similar to Polygamist. Where the EV is much MUCH higher than the actual winrate.
It's actually pretty similar to the proposed revisions to Polygamist in the Autopsies thread.
Are there figures that raise the EV above 50%?- callforjudgement
-
callforjudgement Microprocessor
- callforjudgement
- Microprocessor
- Microprocessor
- Posts: 3972
- Joined: September 1, 2011
OTOH, an EV of almost 50% is a red flag for a Nightless setup, which MS consistently performs incredibly well on. I think the lack of nightkill is probably more important than the lack of power roles.
You can increase the EV in a fairly simple way by removing VTs, although that makes it even harder to give town enough control over the lynch vote.
I don't agree that D1 lynches onsite are essentially random. Don't they hit scum more often than EV would suggest? (It's just that the EV for hitting scum D1 is very low in most setups.)scum· scam · seam · team · term · tern · torn ·town- callforjudgement
-
callforjudgement Microprocessor
- callforjudgement
- Microprocessor
- Microprocessor
- Posts: 3972
- Joined: September 1, 2011
Actually, I just checked the D1 lynch stats for Matrix6 (the most readily available source of information on the matter). The odds of a town lynch there were almost identical to that gained by random lynching (which is a bad sign, as you'd assume that in Newbies some players would be able to save themselves with a claim). The odds of a scum lynch were a little lower than would be expected by chance (being a Newbie setup, there were a few D1 no-lynches).
So the EV here probably wants to be around 50%, based on those statistics. I think the mechanics in this setup will make scum easier to hit, though, as the relational tells should be much stronger than usual.scum· scam · seam · team · term · tern · torn ·town- Mathdino
-
Mathdino Survivor
- Mathdino
- Survivor
- Survivor
- Posts: 14337
- Joined: February 24, 2013
- Location: Right Behind You
Both of those arguments have been made for Polygamist and Lovers Mafia. Nightless and should in theory produce very good associations. The result was that the meta shifted toward D1 running each other up to L-1 until towns got cold feet. Polygamist, with a 4:8 setup and similarly 2 nightless day cycles, lynches scum significantly less than random.
tl;dr scum are better at faking associations for a single day than town are at catching associatives.
And unfortunately, you're correct that removing VTs gives scum more control over the game.
Also notable that the voting mechanics are going to scumside it. Nonstandard voting tends to. Lack of unvotes encourages rash decisions.
Ughhhh I really want this setup to work but towns are so bad at all vanilla -_-
Why 4 scum again? Could we try with only 3?- callforjudgement
-
callforjudgement Microprocessor
- callforjudgement
- Microprocessor
- Microprocessor
- Posts: 3972
- Joined: September 1, 2011
Changing the number of scum doesn't actually seem to significantly change the EV if you adjust the number of townies down to compensate for them not needing as many votes, likely because the setup is about ⅔ townies at any size. (If you reduce the number of scum while keeping the number of townies the same, EV goes down.)
I thought 4 scum would be necessary to give town a reasonable chance of catching D1 associatives and/or finding a weak scum to press for info.scum· scam · seam · team · term · tern · torn ·town- mith
-
mith Godfather
- mith
- Godfather
- Godfather
- Posts: 9267
- Joined: March 27, 2002
- Location: McKinney, TX
- Contact:
I think you're misunderstanding my proposal. When I say A and B lynches, these are totally distinct. You don't add the votes on A with the votes on B to reach 8. At most one player can hit 8 votes on A, and one player on B, and as soon as there are two different players hammered simultaneously they are both lynched and the game ends.I tried to make a Doublevoter version work (i.e. equivalent to Doublet Voters but with the numbers doubled and without the ability to contribute twice to the lynch of the same player) but it'd require 11 to lynch to prevent a hammer contradiction. I guess you could just reduce the lynch threshold while not worrying about the "three wagons can viably hit 8 at the same time" problem. It's possible in your version of the setup to have four players lynching player X with vote A and player Y with vote B, four players lynching player Y with vote A and player Z with vote B, and four players lynching player Z with vote A and player X with vote B, meaning that three players are in a hammered state simultaneously. That's normally undesirable, although you can make it work.
The idea is that a coalition of 8 townies can force a lynch, and both lynches can be truly controlled by the town, rather than either needing all 11 townies voting together or else relying on the first target's vote to help lynch the second target (which is fine if you are lynching scum, but not so fine if you are lynching town - might as well make it a normal lynch plus a vengekill if it hits town).
Some days I can math, some days I can't.Three 5½-votes at the same time are impossible; there are 15 players so if 11 votes are tied up on two of the players, a third player can have at most 4.- mith
-
mith Godfather
- mith
- Godfather
- Godfather
- Posts: 9267
- Joined: March 27, 2002
- Location: McKinney, TX
- Contact:
I was aware it was being run, but not that it was finished. I'll have to flip through it at some point.(In case you missed it, Red Flag just had a test run, with fairly mixed reactions to the setup. My reaction was that Day 1 was a lot of fun but the rest of the game wasn't really, so I tried to preserve the feeling of that part of the setup.)
At some point I'd like to run some actual numbers for the "should scum kill their own" question. I mentioned it as a possibility about a week after suggesting the setup (and even suggested that maybe the wincon should just be 2 Mafia dead rather than 2 Mafia lynched), but never really explored the ramifications of leaving it as is. In the actual game, not killing Mafia N1 was clearly a mistake; in fact, with a 4:9 setup I would hazard a guess that scum should *always* kill two members, regardless of how the lynches go. Town does still get a benefit over a comparable Vanilla setup though, in the form of higher odds on the first couple days. In 4:7, things are probably different if the first lynch is town (since town is immediately in a double lylo if scum kill town).
It's an interesting problem though, and worth exploring to see how that strategy affects the EV - the EVs I gave originally were assuming they would kill town.- mith
-
mith Godfather
- mith
- Godfather
- Godfather
- Posts: 9267
- Joined: March 27, 2002
- Location: McKinney, TX
- Contact:
So, if my numbers are correct, after a day 1 scum lynch in the 4:9 setup, town EV is ~72% if scum kill town N1 and ~67% if scum kill scum. Not as big of a difference as I was expecting, but still a clear blunder in that game. After a town lynch, scum should kill town (~35% vs ~38%). Overall I get a 44.5% EV for 4:9. (It also has a pleasing feature that after day 1, EV is close to 1/3 if town was lynched and close to 2/3 if scum was lynched.)
At 4:7, scum should kill town either way - there should always be some point where you're close enough to Lylo to just go for that. Curious where the boundaries are for other counts - there is one boundary to cross where scum should kill scum after a scum lynch, and another where scum should kill scum after a town lynch. For 4 Mafia, you have:
4:5 - kill town either way
4:7 - kill town either way
4:9 - kill town after town lynch, scum after scum lynch
4:11 - kill scum either way
In that sense, 4:9 may be the most interesting count.
At some point I'll set up my script to handle this and start a new thread. It's a pretty interesting setup theoretically, even if it doesn't play that well.
[edit]Just saw that you came up with something different in the post-game. I did find one error in my spreadsheet, but not something that affected 4:9. Still checking.[/edit]- callforjudgement
-
callforjudgement Microprocessor
- callforjudgement
- Microprocessor
- Microprocessor
- Posts: 3972
- Joined: September 1, 2011
- Gamma Emerald
-
Gamma Emerald AnySurvivor
- Gamma Emerald
Any- Survivor
- Survivor
- Posts: 69101
- Joined: August 9, 2016
- Pronoun: Any
- Location: Hell on Earth (aka Texas)
This isn't functionally nightless though: you don't have the carry-over of trusted towniesIn post 4, callforjudgement wrote:OTOH, an EV of almost 50% is a red flag for a Nightless setup, which MS consistently performs incredibly well on. I think the lack of nightkill is probably more important than the lack of power roles.
You can increase the EV in a fairly simple way by removing VTs, although that makes it even harder to give town enough control over the lynch vote.
I don't agree that D1 lynches onsite are essentially random. Don't they hit scum more often than EV would suggest? (It's just that the EV for hitting scum D1 is very low in most setups.)<Embrace The Void>
“A flipped coin doesn't always land heads or tails. Sometimes it may never land at all...”- mutantdevle
-
mutantdevle Mafia Scum
- mutantdevle
- Mafia Scum
- Mafia Scum
- Posts: 3731
- Joined: October 21, 2017
- Location: Hell
- callforjudgement
-
callforjudgement Microprocessor
- callforjudgement
- Microprocessor
- Microprocessor
- Posts: 3972
- Joined: September 1, 2011
Reviving this thread because I thought of a simple way around the vote density problem (this also gives a lot more scope to change balance by changing the numbers):
The EV is going to be higher than the 47.6% I placed above, because there is going to be some scope for minor theoretical breaks by relying on the fact that scum votes don't count towards the wagon. (My best attempt to break this involves splitting the wagons 7:4:4, with the small wagons consisting of two scummy players crossvoting plus the six towniest players providing the votes needed to make the wagons valid, but I think it probably works out worse in practice than just playing the setup normally and not trying to break it.)scum· scam · seam · team · term · tern · torn ·town - callforjudgement
Copyright © MafiaScum. All rights reserved.
- mutantdevle
- Gamma Emerald
- callforjudgement
- mith
- mith
- mith
- callforjudgement
- Mathdino
- callforjudgement
- callforjudgement
- Mathdino
- callforjudgement
- mith
- callforjudgement