It took me longer than normal to type my post because I was tabbing out in the middle of a raid to do it. I essentially had the same train of thought as you.

We are two birds of a feather!

@JD:

Innistrad: Day 1 Mafia Lynch
Bolas' Birthday: Day 1 Town Lynch
Mafia the Gathering: Day 1 Town Lynch
Old Port: Day 1 Town Lynch
Chocolate Mafia: Day 1 Town Lynch
Pick your Role: Day 1 Town Lynch
Touhou Double Dealing Characters: Day 1 Town Lynch
Intervals Mafia: Day 1 Town Lynch
Werewolves invade NGA: Day 1 Town Lynch
Curse of the Blood Witch: Day 1 Mafia Lynch
MTG upick: Day 1 Town Lynch
I excluded both Dethy's, ultimate mafia, epistomology, and Dante's inferno since they seemed to operate under significantly different rules.

Chance of Lynching Mafia on Day 1: 18%

This seems high to me, and I imagine it will go down as more games are played. I've played on a few different sites and it is normal for town to be lynched day 1 more often than a random lynch would result in. If you're still doubtful I can explain to you why I believe this trend exists.

As for what it provides on concurrent days: The answer is roughly the same thing a directed lynch would, albeit a little less. Players are still forced to take stances and argue as we are doing now. They'll be a little less to analyze in the way of voting records than there would be otherwise, with two player wagons competing over a rynch and a player wagon competing, but the increased chance in killing scum more than makes up for that small loss in information.

EBWOPreview: Will respond to KoD's wall of text shortly.

@Rag:

You think that those past games is an indication that a random lynch will do better? Do you know how probabilities work?

Let's do a little experiment for fun. You have listed there 11 games. So 11 will be our number of trials.

We're going to consider a d8 (an eight-sided die). Each of the numbers has a 1/8 chance of occurring. So, for this experiment, we'll say that on a 1 or 2 a mafia aligned person has been lynched. 3-8 town. That means 2/8 is the chance of a mafiate being lynched (which is 25%) and the chance of a town being lynched is 6/8 (which is 75%).

Now, let the dice fall:













11 rolls to simulate 11 games. Depending on the outcome, we'll have an idea of whether or not a D1 lynch of a town or scum occurred. On to my next post to see.

Ok, let's see what we have:

1 or 2: 2

3-5: 9

As you can see, we have here a situation where the random number generator has replicated what you used as your example to backup your side of your argument. And, as the random number generator has faithfully shown, it is just as bad like your examples.

So, tell me, what is your point in listing those example? That town sucks at lynching scum? Well so does the rng apparently. What do we have over a rng though? Interactions. You don't get that with a rng.

*Drops mic.*

All the BBCode i touch turns to dust so I'll be using numbers to indicate my responses rather than quotes.

1) My claim is that random lynching will be more likely to result in a lynch than the traditional lynching strategy. 25% is greater than 18%, to use the numbers I just produced.

2) I do concede that some degree of interaction is lost, but it is not a great quantity. There was never a great quantity of information to be gained on day 1 to begin with. A 72% increased chance of lynching scum is well worth what is lost in information.

3) When I relate random lynching to cop results, what I mean is that they both represent strong mechanical plays rather than the purely psychological plays you would find in purely vanilla mafia. Whether or not to use them is a question of hard math rather than a more subjective discussion over what tells are legitimate, and similar veins of thought.

4) When i say "Voting records, psychological analysis, and the like can still be used to strenghten decisions", I mean that things like cops, vigs, and rynches will never win us the game on their own. The mechanics of a game might confirm that a given player is town, but we still need to use our reads to decide which of the remaining players are most likely to be scum.

Edit: Nice dice. Are you arguing that my 18% figure is flawed or that my 25% figure is flawed?

I'm arguing that your entire argument is flawed.

Or, to be a little more detailed, your entire idea that a random lynch will do better is entirely untrue. After all, a random number generator will only have as much of a chance at lynching scum as we do except that we can add in interactions (reads) to further direct our decision whereas a rng cannot.

Let me take approach this one point at a time.

A random lynch will have a 25% chance of lynching scum, do you agree with this?

Probability isn't determined by the number of successful times you've done something in the past, Rag. At least not in the numerical instance you're trying to apply in that post. Your argument is pretty much "Town sucks at this game. F'it let's just random." And that isn't a strong argument in favor of the town.

The use of the dice was just a nice way to represent your argument to Numbers. You put forth 11 games and colored them in a way where, I assume, red was a day 1 scum lynch and green was a day 1 town lynch.

I put forth a similar scenario of 11 games (11 trials) represented by 8 sided dice that helped replicate the % that we have here in this game (the % for lynching scum or town). And as the rng showed, the dice showed, we had 2 out of the 11 "games" where a scum was lynched D1 and 9 out of 11 where town was lynched first. Exactly like your example.

And the reason for that is because of the probability. A 75% chance to lynch town is greater than a 25% to lynch scum. 3 out of 4 times you will more than likely have a town lynched. Can't be helped if left strictly to numbers. Thus why interactions matter.

Or just accept that what Numbers put forth is accurate down to the letter.

A thousand times what Numbers said.

Anyway, let's see how you guide me. Yes, a random lynch has a 25% of lynching scum in this game with 3 scum out of 12 players.

@KoD: Do you believe that outside of random lynching, the chance of a day 1 scum lynch is approximately 18%?

@JD: Town doesn't suck at the game, in my opinion. At least at lynching scum on day 1. 18% is a higher success rate than I would have expected. Regardless, the fact is that the random lynch is stronger. Whether or not 18% is good or bad doesn't matter as long as its less than 25%.

No. First off, you obtained that % from a sample of only 11 games. Secondly, you are taking that % and applying it to other games even though each game's probability of lynching scum is independent of each other. That is, your 18% is not the correct probability for lynching scum day 1. In this game, it is 25%. In another game, it may be higher or lower depending on any number of factors. But to say that the chance of lynching scum day 1, in this game, is 18% is ludicrous.

Bear in mind that your 18% can change by adding more games to the pool. And those additional games may, in themselves, have different probabilities for lynching a scum D1.


Anyway, no is my answer to your question. I do not believe that, outside of random lynching, the chance of a day 1 scum lynch is 18%. Adding the word approximately changes the response though since 18% is close to 25%. So I could also say yes.

So I will answer yes, I believe that, outside of random lynching, the chance of a day 1 scum lynch is approximately 18%.

Now, what is your point? Because you're not doing a good job of being clear with your line of questioning.

Except that if you refer back to my example, the rng is at 18% as well. Same as us.

Not only is rng just as bad at this as we are, you're misrepresenting probabilities. And that is bad as you are effectively misinforming people based on faulty information.

It is true that the basic probability of the lynch differs from game to game, although most games use a roughly 25% mafia and 75% town split and I would expect the differences to average out. My 18% may be a little off because of this, but not by 7 percentage points unless there was a big outlier in the batch that failed to notice. If you really want me to i will go back and get all the town to mafia ratios for you, but I suspect you already know that mafia will occupy more than 18% of the game on average.

On the topic of a small sample size, it is true that my number will have a relatively large margin of error because of it. That goes both ways though. The number could be lower than 18%, or it could be higher than 18%. Unless we have more information that points to what the number actually is though, then treating it as 18% will yield the most corruct deductions given the information available to us.

Saying that a non-random lynch today has a 25% chance of hitting scum defies what has been demonstrated to occur in the past. Certainly your number could be correct, but as far as I can see you have no reason to believe it is.

Random sucks. Vote info is important in later days.

Wait 15377 is JayDreven?

DUUUUUUUUUUUUDE.

Tried to make this shorter and to the point.

Regardless of what are actual percentage of lynching scum on day 1 is, IMO it's still preferable than random lynching, as long as our chance of lynching scum is "non-zero" (which is obviously true). That is because by making people commit to a vote, they reveal personal preferences.

It may be worse for day 1, but it is infinitely better for future days. Voting patterns matter a lot in the future, especially when you have already found one or two mafia members and are looking for the last ones. If we random lynch forever, then there are no voting patterns to base anything off of, and every day results in town operating under almost zero information and puts us back into yet another "day 1" scenario. Basically random lynching results in an endless cycle of random lynching, which I am not okay with.

I've played quite a bit of mafia recently in person. Even with skillful play, town often kills the wrong person on the first day. But the results of the first day are always a huge catalyst in finding scum. Obviously it's easier to read people in person (at least in my opinion), but the same principles apply.

@JaC/Silly: Information is generated regardless of whether or not a random lynch or a directed lynch ultimately occurs. There will be competing wagons, people will defend themselves and their philosophies. People will argue over meta, they will argue over arguments, and they will argue over nothing, and that arguing will be reviewed on later days to scumhunt. The information we have going into day 2 will not significantly differ even if a random lynch ultimately gains the most votes.

I also hadn't mentioned this earlier, but random lynching stops being effective as the town gains access to more information and can make more informed lynches, which inevitably happens over the course of any game. I would only advocate a random lynch on day 1.

Not actually exactly true. Using statistics, we cannot reject the possibility that the chance of a day 1 scum lynch isn't 25%.

Knowing the last 11 games have 2 successful day 1 scum lynches, then there is a roughly 45.5% chance the underlying probability of a day 1 scum lynch could actually be 25% (using a simple binomial distribution).

Votes are meaningless unless there is some sort of ramification to them.

Town gets no vote information if we vote for no lynch. Cop information is completely useless in catching additional scum by association if we have no vote information.