MM's Hypothesis

[Machiavellian-Mafia]

I felt the need to contribute to the recent intellectual discussions here, so I have formulated my own hypothesis regarding mafia:

The Activity Constant for a game in forum Mafia is equal to the quotient of the number of posts squared and the product of the number of real-life days the game has elapsed and the number of views, or:
AC = posts^2 / (rl-days*views).

An AC of below 1 indicates a below average activity level
An AC of above 1 indicates an above average activity level
An AC of exactly 1 indicates an ideal activity level, also known as activity equilibrium

Examples of the AC and how it's a good indicator of the activity level or pace in games:

Mafia 49 (slow game) -> AC = 1622^2 / (176 * 31574) = 0.47
Mafia 57 (fast game) -> AC = 1119^2 / (60 * 10429) = 2.00

Mini 388 (slow game) -> AC = 907^2 / (141 * 8444) = 0.69
Mini 361 (fast game) -> AC = 1187^2 / (77 * 11957) = 1.53

Chrono Trigger Mafia (slow game) -> AC = 2504^2 / (261 * 38962) = 0.62
Lights Out Mafia 2 (very fast game) -> AC = 2503^2 / (68 * 24360) = 3.78


So as you can see from the sample calculations, all the slow games had an AC value of below one while all the fast games had an AC value of above 1.

I think the AC value can be most practical to mods who want to monitor the activity of their games by periodically recording data about posts, views, and days. I will elaborate more later.

Now discuss.

[Kelly Chen]

So the hypothesis is that a value of 1 (not higher) is ideal?

[Mr Stoofer]

Well Lights Out 2 (AC=3.78) went too fast for me while I was alive. I suppose individual players will have their own views about what is an ideal AC.

[Zindaras]

Lights Out 2 was insane.

[Machiavellian-Mafia]

So the hypothesis is that a value of 1 (not higher) is ideal?

The ideal value would be 1 for a game with the players being overall "normal" and the mod having average expections regarding activity. However like Stoofer mentioned, different players and mods have different tastes, some like it slow, some like it fast.

[Mr. Flay]

My ideal is that Posts Per Day = Number of Players Alive, on average. I'm not sure your conceptualization here scales for different sized games.

[ubertimmy]

Now, the real question is, what does this end up being for games like Bad Idea II?

[IH]

Wait, let's try and use this...

If this could be used for a day... could we set a set point to which it drops where a deadline rule could be set?

[Mr. Flay]

Problem with set points (even my guideline I use above) is that it doesn't account for absent/vacationing players very well. I tend not to deadline unless the thread is lagging for 2-3 days AND no one needs to be prodded (3-4 days w/o posting, depending on the size of game).

[mith]

My ideal is that Posts Per Day = Number of Players Alive, on average. I'm not sure your conceptualization here scales for different sized games.

If you assume that Views = Posts*Players (that is, each player reads each the thread once for each new post) you get something similar to Posts/Player-Day (not exactly the same, because you're averaging over the whole game, and there's a different number of players alive at different times in the game).

It's probably not even that terrible an assumption for most games (within some proportionality constant), but if anyone bases a deadline rule on a dubious formula involving the number of views a thread gets, I will personally hunt them down and throw them into the darkness, where there is weeping and biting of tigers.

(I need to formulate "mith's Law", which involves dubious mathematical approximations for Mafia and how they suck.)

[Kelly Chen]

(I need to formulate "mith's Law", which involves dubious mathematical approximations for Mafia and how they suck.)

If you're protown this is already part of your role.

[Yosarian2]

Speed mafia games, like Lights out style games, are obveously going to be much faster, that's the point.

[EmpTyger]

<snip>I will personally hunt them down and throw them into the darkness, where there is weeping and biting of tigers.

Biting *by* tigers, I believe you mean.

(I need to formulate "mith's Law", which involves dubious mathematical approximations for Mafia and how they suck.)

The simplest formulation for "mith's Law" would be to not interpret the apostrophe as showing possession.

[ShadowLurker]

Bump.

I found this was actually extremely accurate and I am using it as a benchmark to set deadlines by taking constant AC checks and seeing the variance.

For example, LO2 which fell into a huge slump midgame with 2 deadline LOs in a row decreased from 3.8 near the start of the game to a 1.7 which is a huge change percentage wise.

However, its 1.7 still makes it a quick game because it was one. (Note this should be slightly higher as it's counting all the views from after the game was over)

This also seems to follow the trend that games always start quickly as players are still interested.

[pablito]

Since my regular games expect every player to post every 48 hrs (in the rules), I require 6 posts per day if there are 12 players alive. If the average posting pace goes below this simple formula, or if there's 24 hrs without a single post in early game, I give a threat of deadline. If there's no uptick, I deadline.

My current game has auto-deadlines, so it won't figure in anyway.

But with Lepers being at 1.53, I guess it might've still been too quick. With lepers, I think my general accepted pace was ~12 posts a day (2 days to make a new page) until I started going into more meticulous monitoring. Don't even remember if I deadline in lepers, you guys were so good.

I also think that with LO2, because the game such had a high early AC, it made it tougher for replacements to keep up and that helped drop the AC later on.

I think it would be interesting to see if the high AC games had a consistent level or if it was a consistent down-slope.

[Guardian]

I often read games I am not in, as I am sure others do. I check in on them to see how they are going; it's like a cheap novel. That would mess up this .

[gorckat]

Not necessarily, if a game is interesting, it'll get views from watchers. I'd expect these interestign games to have more activity. I doubt watchers are able to heavily skew the number of views.

[yellowbounder]

The problem with this is, that when you're dealing with a small game, it gives high results since less views, and thus even with very slow games it gives results higher than 2. I wouldn't say that 1 is an ideal.

Can someone come up with a modification combining players in the game?

[Brutal Assassin]

Speed Mafia - http://mafiascum.net/forum/viewtopic.php?t=4060&start=0

461^2 / (16 * 8782) = 1.512 AC

This game moved at a BLISTERING pace but your rating doesn't really reflect that. Namely because we managed a ton of views in a short period of time. What is also possible is that the game got a lot of views post-mortem, but that can be said of all of the games mentioned here really.

I think the 100 MPH games would follow this same trend, but I can't find them.

[Thesp]

Speed Mafia - http://mafiascum.net/forum/viewtopic.php?t=4060&start=0

461^2 / (16 * 8782) = 1.512 AC

This game moved at a BLISTERING pace but your rating doesn't really reflect that. Namely because we managed a ton of views in a short period of time. What is also possible is that the game got a lot of views post-mortem, but that can be said of all of the games mentioned here really.

I think the 100 MPH games would follow this same trend, but I can't find them.

What about running a game where the town auto-loses if there aren't X number of posts per day, and double-posts don't count? (The time interval might need to be stretched out, and the numbers tweaked, but it might be doable...)

I'd be intrigued to see the lurker hunts in that game.

[Glork]

Thesp, you'd also need to safeguard against spamfests or intentionally drawn-out "debates" in order to enforce that rule properly. I could make one-liners about you being scum, claim that it's part of my chosen playstyle, and bump up the posts-per-day count to the point where the restriction probably wouldn't matter.


I'd suggest a "Game-relevant content required" clause, as well as a "Minimum Length" clause if you're going to do that. It'd be like a Verbose game, only with higher stakes for nonposters.

[Yosarian2]

Thesp, you'd also need to safeguard against spamfests or intentionally drawn-out "debates" in order to enforce that rule properly. I could make one-liners about you being scum, claim that it's part of my chosen playstyle, and bump up the posts-per-day count to the point where the restriction probably wouldn't matter.

Well, I'd rather see people making one-liners about who they think is scum then see people not posting at all. And if double posts don't count, then one Glork posting lots of one liners can only be at most half of the needed posts.

Besides, long drawn out debates probably don't really help; you get a lot of posts from those two people, but once you get to a certain point they can make other people post less.

[Mastermind of Sin]

So the hypothesis is that a value of 1 (not higher) is ideal?

The ideal value would be 1 for a game with the players being overall "normal" and the mod having average expections regarding activity. However like Stoofer mentioned, different players and mods have different tastes, some like it slow, some like it fast.

and some like it hot.

[Mastermind of Sin]

MM, are you counting posts made before and after the actual game? If so, are you adjusting for dates as well?

Committee mafia was .636

[Mastermind of Sin]

Mafia 64 has 2.278 already in Day 1 omg...