Day 1 Lynch Policy Analysed

[Iced_Monopoly]

So recently I decided to review our S-FM games to determine what the outcomes of the D1 lynch policy typically are, as you can see in the table below, I went through 27 games, only choosing games that had 2 distinct factions (Scum Vs Town); taking note of whether there was a d1 lynch or not, and what the outcome of that lynch was, as well as the game end outcome. Also noting here I excluded games with mandatory lynch/nolynch policies on D1

Screenshot_1.png

So from this we can see:
64% of games are D1 lynch games
On a D1 Lynch, A town member is killed 69% of the time
On a D1 Lynch, Town wins the game 50% of the time
On a D1 No Lynch, Town wins 64% of the time

My thoughts on this are:
Wow, I wasn't expecting the no lynch option to be so prevalent, I was expecting it to be as low as 20%, not closer to 40
We really suck at choosing lynch candidates

So in regards to the last two points, I understand that there are a lot more variables in play and that it doesnt come down to a D1 lynch to determine the game outcome, however, there clearly is some association with the two though even if it is minor.

What is there to take away from this? Well as someone addressed in a previous post (or end game chat, Ive forgotten), we are terrible at scum hunting without power role leads or associations. Do I know how to resolve this? No fucking clue, but it exists so someone else should get onto that. There is almost a better chance to random a slot for lynching and hitting a scum than there is in the town 'scum hunting' on D1 (there actually were better odds before I expanded the sample size).

Whilst I am in support of D1 lynches (Town controlling the kill blah blah blah), the town at this point might as well be a child with a gun in a blindfold. I would temporarily support D1 No-Lynches until town improves D1 scum hunting techniques (someone suggested a game with no PRs, that might help educate and allow players to experiment) because there is some mild correlation that a D1 no-lynch improves odds of a town victory.

Anyways im sure plenty in the post will be disagreed with, but im just providing some numbers and a platform for discussion based on previous games. I will continue to add to this table over time and re-adjust percentages as required, I will probably only work my way down to the start of 2015 though, as the player base changes enough to affect the data by then.

[Firebringer]

That doesn't seem like a huge difference, yes town wins more games they no lynch. But does that have to do with game mechanics at all? Like, some games mechanics are much different than all. So this data isn't as reliable as you may expect.

Though I agree, we tend to lynch more town than scum day 1.
If you townread everyone in the game is the only reason you should not lynch someone. Though that is often rare. I don't think players should compromise onto lynching a townread of theirs.

[Quick]

I don't think players should compromise onto lynching a townread of theirs.

This.

[Calix]

This is a decent post that can be taken into consideration if the leading one or two trains at EOD are on town-reads.

The problem with mandatory No Lynch days, however, is that most players resort to trolling because there is "no pressure" via votes. Case in point, RLVG's Hidden Setup; the first 200 posts are almost completely worthless.

If people stopped fucking around so much just because they can't be lynched for it, I'd probably appreciate the benefits of a No Lynch more. As site meta stands, it's just a pain in the arse.

[PLZLEAVEDUCKK]

I don't think sleeping is ever better than a lynch. The information from the vote logic and the flip are crucial for night actions and day 2. As players improve the accuracy of day 1 lynches should improve. Honestly, lynching a town day 1 is fine if town is able to get the correct information out of it.

[Iced_Monopoly]

I don't think sleeping is ever better than a lynch. The information from the vote logic and the flip are crucial for night actions and day 2.

Im not arguing against that, Im saying that the data shows that d1 lynches are more likely to kill town than a scum. Those methods to find scum are fine, but I want to focus on improving scum hunting WITHOUT those techniques.

As players improve the accuracy of day 1 lynches should improve.

The data suggests the opposite is happening. More D1 lynches are occurring in recent games, town are being lynched more often in them, and mafia are winning more games.

The point of this was not to debate the Lynch/No Lynch agenda, but rather to discuss strategies to improve scum hunting without the reliance on flips or associations so that D1 lynches result in less town kills and increase the likelihood of a town faction win.

[PLZLEAVEDUCKK]

Well players need to learn to look at others play from their point of view and look at what agenda they are pushing and what their intent is. People tend to focus on random things and not look at it from if this person is mafia what are they trying to achieve with this play.

I think unless every player in the game is of equal skill, a town will usually be the mislynch day 1. It is too easy for mafia to frame a weaker town when they say things like "I am going to ignore you".

[MattZed]

The numbers here are interesting, although I do worry about drawing conclusions from them. If we were treating these like random trials in a statistical test, (which I'm not really comfortable doing because the setups and players are different from game to game) the margins of error would be something on the order of ±20%, which makes it hard to tell if no-lynching is really better or worse than lynching.

[Iced_Monopoly]

The numbers here are interesting, although I do worry about drawing conclusions from them. If we were treating these like random trials in a statistical test, (which I'm not really comfortable doing because the setups and players are different from game to game) the margins of error would be something on the order of ±20%, which makes it hard to tell if no-lynching is really better or worse than lynching.

This too was my concern, which is why I tried to limit the specifications for a game to make the list. Obviously there is no complete control over hosts/setups/players/play styles etc, so it is difficult to put an accurate figure on it all, especially regarding game outcomes. The main focus is on the flip of the d1 lynch though, which is easier to control for, and should become more accurate as more games are fed into the pool.

[MattZed]

This too was my concern, which is why I tried to limit the specifications for a game to make the list. Obviously there is no complete control over hosts/setups/players/play styles etc, so it is difficult to put an accurate figure on it all, especially regarding game outcomes. The main focus is on the flip of the d1 lynch though, which is easier to control for, and should become more accurate as more games are fed into the pool.

The thing I find most interesting from the data is that it goes along with something I think we've all thought: The D1 lynch can be fairly predictive of the outcome; if mafia is lynched D1, town tends to win, and vice versa.

[Calix]

Well players need to learn to look at others play from their point of view and look at what agenda they are pushing and what their intent is. People tend to focus on random things and not look at it from if this person is mafia what are they trying to achieve with this play.

This.

Intent > Random list of accepted 'scum tells'

Scum don't act like scum for the hell of it. Most things they post have some purpose behind it that furthers their agenda.

[Quick]

Honestly I don't think I've ever played on a site with such a high day 1 no lynch ratio. I don't really know why exactly. Most sites make it almost mandatory to lynch day 1. Like they view it as a really big deal if a lynch isn't reached day one. One reason I can see for this is that this site primarily uses majority lynches and days are 48 hours. If days were longer this might change, or if games were plurality lynch this might change. Most sites I've been on the pattern I see is that, in general, the shorter the day phase the more posts per hour you get by a significant margin, even accounting for ratio of length of day to how active the games are. *shrug* I can't say I know of a solution based on that, getting more scum lynched day 1 I mean. I think what really matters here is what the win rate is for town vs. scum. Taking a look at that might even things out a bit, IDK.

[DarknessB]

The thing I find most interesting from the data is that it goes along with something I think we've all thought: The D1 lynch can be fairly predictive of the outcome; if mafia is lynched D1, town tends to win, and vice versa.

This. Also makes sense given most S-FMs only have 2-3 groupscum. If you can knock one out on Day 1, Town really has no business losing the game for the most part, especially given the associations you can make from the successful lynch.

People should take Day 1 far more seriously than they do for this very reason. The tone of the game (and often the outcome) are heavily influenced by what goes on there, for better or worse.

[DarknessB]

This.

Intent > Random list of accepted 'scum tells'

Scum don't act like scum for the hell of it. Most things they post have some purpose behind it that furthers their agenda.

Also, I hate to go meta here, but a number of players deliberately / unintentionally act off the charts scummy during Day 1. The real issue is teasing out TPRs hiding themselves (I find that annoying and overdone usually) from underperforming players from actual scum.

[MattZed]

Also, I hate to go meta here, but a number of players deliberately / unintentionally act off the charts scummy during Day 1. The real issue is teasing out TPRs hiding themselves (I find that annoying and overdone usually) from underperforming players from actual scum.

I think we'd be better off as a site if TPRs didn't give a shit if they got night killed.

[Kovath]

I think we'd be better off as a site if TPRs didn't give a shit if they got night killed.

Probably better off as a site, but then who would guide the reliant on night actions to find scum!?

[Quick]

I should prolly explain why I agreed so much with that quote I had of fires, feel free to disagree. It comes from a perspective not knowing the sites meta so it may not work here IDK. Basically If Town doesn't compromise on lynching Town reads then it will ensure that more overall scummy people are lynched instead of Townie looking people. I understand this cannot always be done, but it should be the goal. If Town can find common ground on who they absolutely do not want to lynch, than town has a better chance to vote off scum that day. I am just banking on Town overall will be right more often then they will be wrong. If this isn't true then its a moot point. The idea is to condense a lynch pile rather than having the attitude to settle. I think it could also help discussion for town to trust their town reads better. The thinking here is that its easier to get solid Town reads before solid Scum reads.

If someone can explain why this will or will not work on this site it will help me understand this game a little better I think.

[DarknessB]

I think we'd be better off as a site if TPRs didn't give a shit if they got night killed.

Agreed -- I would rather have productive and scumhunting TPRs at a higher risk of being night killed versus 1/3 of the Town being useless and unhelpful to avoid drawing the attention of scum. This is one of the reasons our site's Day 1 needs major improvement.

If everyone tries to be productive, then there's more information / posts to work with, and TPRs might not stick out so much as hardcore lurkers / fence sitters.

[Calix]

I should prolly explain why I agreed so much with that quote I had of fires, feel free to disagree. It comes from a perspective not knowing the sites meta so it may not work here IDK. Basically If Town doesn't compromise on lynching Town reads then it will ensure that more overall scummy people are lynched instead of Townie looking people. I understand this cannot always be done, but it should be the goal. If Town can find common ground on who they absolutely do not want to lynch, than town has a better chance to vote off scum that day. I am just banking on Town overall will be right more often then they will be wrong. If this isn't true then its a moot point. The idea is to condense a lynch pile rather than having the attitude to settle. I think it could also help discussion for town to trust their town reads better. The thinking here is that its easier to get solid Town reads before solid Scum reads.

If someone can explain why this will or will not work on this site it will help me understand this game a little better I think.

Nothing clashing with site meta here. It's perfectly sound to attempt to agree on players to NOT be on the table for a lynch.

If anything, we generally have around 5-7 people voted on D1, so the wagons are spread out. We could do with a way of preventing that, given that lots of small trains rarely leads to an organised D1 atmosphere.

[Kovath]

Agreed -- I would rather have productive and scumhunting TPRs at a higher risk of being night killed versus 1/3 of the Town being useless and unhelpful to avoid drawing the attention of scum. This is one of the reasons our site's Day 1 needs major improvement.

If everyone tries to be productive, then there's more information / posts to work with, and TPRs might not stick out so much as hardcore lurkers / fence sitters.

This is the kind of thing that would make everyone better off if they did it, but if not (or in the meantime) would only serve to further draw night kills onto active / skilled players and kill more TPRs in the process. So kinda like Stag Hunt from game theory.

[Quick]

Nothing clashing with site meta here. It's perfectly sound to attempt to agree on players to NOT be on the table for a lynch.

If anything, we generally have around 5-7 people voted on D1, so the wagons are spread out. We could do with a way of preventing that, given that lots of small trains rarely leads to an organised D1 atmosphere.

What about having 2 solid Bandwagons throughout the day(s)?

[AnassRhamur]

LETS REPORT THIS FUCKING BITCH LIKE ITS A SCHOLARLY ARTICLE. ITS FUCKING 6 AM AND I STILL HAVE SO MUCH OF MY FUCKING RESEARCH TO INPUT.

A recent study done by Iced_Monoply at Trump University has shown that a Day 1 lynch is harmful for the town's chances to win (n = 27, DF = 1, p > .05). Analyzing the data yielded a pearson correlation of -.34 with a significance of .086 ( r = -.34, p = .086). These results indicate that randomly lynching had an inverse relationship with the likelihood of Town winning the game as a member of town was most likely to be selected for the lynch on D1. Unfortunately the p-value calculated from this survey is too high to be statistically significant but this is highly contributed by the low sample size.

[Cryptonic]

Depends on the setup, but d1 lynch is beneficial most of the time.

The most town-controlled deaths in a game, the better Town's chances.

[Helz]

LETS REPORT THIS FUCKING BITCH LIKE ITS A SCHOLARLY ARTICLE. ITS FUCKING 6 AM AND I STILL HAVE SO MUCH OF MY FUCKING RESEARCH TO INPUT.

A recent study done by Iced_Monoply at Trump University has shown that a Day 1 lynch is harmful for the town's chances to win (n = 27, DF = 1, p > .05). Analyzing the data yielded a pearson correlation of -.34 with a significance of .086 ( r = -.34, p = .086). These results indicate that randomly lynching had an inverse relationship with the likelihood of Town winning the game as a member of town was most likely to be selected for the lynch on D1. Unfortunately the p-value calculated from this survey is too high to be statistically significant but this is highly contributed by the low sample size.

This is actually very much in line with formal statistics.

One point I would like to present is that the lynch rate does not look very far off from what I think the ratio of scum/town would be. I would draw the conclusion that 'on average from this pool of data' our site is not much better at D1 lynching than grabbing a name at random.

Although I am with Crypt and a hardcore advocate of D1 lynching. Its really my favorite day to scum hunt because the perspective gap between mafia as informed and town as uninformed is the largest