EPTOn June 04 2012 10:49 Radfield wrote:
Ver is playing next in CSL!
Ver is playing next in CSL!
Thanks for posting this, Radfield. I totally would have missed it otherwise!
Miscellaneous/Multi-Purpose Mafia I - Page 39
| Forum Index > TL Mafia |
|
Qatol
United States3165 Posts
Thanks for posting this, Radfield. I totally would have missed it otherwise! | ||
|
kitaman27
United States9245 Posts
| ||
|
EchelonTee
United States5245 Posts
| ||
|
Probulous
Australia3894 Posts
Possible Newbie Mini Mafia Imbalance? I have been updating the library and collecting some statistics and I believe we have an issue with our newbie games. I have collected data on all of GMarshal's Minis and the other ones that are in the Library. Given we are up to Newbie Mini Mafia XVI there are plenty missing but I am loath to write them all up myself. Anyway, here are the bald stats Played = 18 Mafia Victory = 14 Most commonly used setup = 3/9 (Town won 1 out of 12) Town victory setups are
Perhaps a 3/11 or 3/12 setup which allows for a modkill and an extra round of Day/Night. Thoughts? | ||
|
EchelonTee
United States5245 Posts
I think 3/10 would be a suitable town buff. Gives a free modkill, but maintains the 3Mislynch to scum victory standard. This would probably only be OK if there are few to none town PRs, as the additional night is too strong, probably. Different setups do seriously need to be considered though IMO, because a 22% town win rate (!!) is absurdly low, even for a small sample size of newbie games. Would like people with more experience to chime in~ | ||
GMarshal
United States22154 Posts
Adding one townie adds an extra night to any power roles, because the town will no lynch in MYLO. In the medic/dt setup that can be fatal, and its less than desirable in the plain 1 dt setup as well. There are two key points that are skewing your viewpoint 1.) (not relevant here, but for completeness sake he mentioned the mafiascum winrates) Mafiascum data does not apply to tl- The game they play on there and the one we play are two different beasts, they have 2-3 week days, and can't talk at night. Players on MS are worse too, I've seen newbie games on there where the entire game had to be replaced by the end of day 2. You cannot use data like that and apply it us because its a completely different ballgame 2.) Newbies are bad. Its a harsh way to say it, but the reason scum keep winning so much, 7/8 of my newbie games, even those with 2 of the more powerful blues (vigis/medics) against weaker scumteams (pure vanilla or just framers) have resulted in scum wins. Without horribly unbalancing it for town you cannot really make newbies win more, because they are still grasping at straws and figuring out how to do things like "don't claim cop when you aren't" You aren't going to fix that by just changing blues without throwing a crapton of information roles at the game, which absolutely fucks scum over, especially considering the fact that in slice of pie and 2of4 they are perfect (no countermeasures at all) 2of4 is a sucky setup, the scumteam have no counter to a follow the cop scenario and it is almost always optimal for the cop to claim day 1 since the mafia can't do anything about it out of fear of a medic. It should only be used if you replace the mafia rolecop with a mafia role-blocker, in which case the setup becomes a lot more solid (and entirely suitable for newbie games if you so desire) The thing to keep in mind is that these loss rates for newbie games are due to modkills and towns just sucking, as opposed to game balance (because having 2 bad players is better than having 7 bad players, especially where a couple idiots will ruin your day). I think its good that newbie towns do so bad, it shows them that there is plenty of room to improve and that losing isn't the end of the world. | ||
|
Palmar
Iceland22633 Posts
If the cop claims day 1, it can only be reasonably assumed he'll get medic protected. The day 1 lynch should still be a fairly easy mislynch for scum. Now, the next night they get a 40% chance of finding out the doctor in the game. Remember that statistic. Assume that mafia don't suck and don't get lynched day 1, they have a 40% chance of knowing who the doc is going into day 2. And even if they don't know who it is, they have narrowed it down to three targets. And that's assuming they have only luck to rely on, no analysis as to who could be the doctor. And this only happens in case town actually gets a doctor and a cop, which is 25% of the time. When I ran 2of4 I replaced the cop with a parity cop, to force 1 night delay upon the results. And people were crying imba that game too, even if the real reason mafia lost is that 75% of the team simply didn't play the game. Town had no concrete information at the end to guarantee a win. Towns sucking is not good for the influx of players into the game. I would very much like to discuss how we can increase the town's chances of winning. One of my ideas was to have smurfs or veterans play on the side of town, we attempted that in a game where I played as electricblack and it worked out okay. The problem being the necessity to create a constant stream of smurfs and the obviousness of them being smurfs because of their post count. The obvious benefit of this solution over balance mechanics is that it does tackle the issue at it's core. You are essentially right that the real problem in newbie games is that newbies are bad. It's inherently easier, and arguable more exciting, to start out playing mafia. Coaching is another option, but again, it requires loads of investment from people who might be more interested in playing games themselves. And honestly, the mafiascum newbie towns are no worse than TL Mafia newbie towns. We have a lot of veterans willing to help and we raise the skill pretty steeply here, but we all start out the same. | ||
|
marvellosity
United Kingdom36161 Posts
On June 07 2012 21:50 Palmar wrote: Coaching is another option, but again, it requires loads of investment from people who might be more interested in playing games themselves. I would happily coach in every single newbie game. I'm there as a coach in Newbie XV. But I think all game I've had... 3 PMs? one of which was about setup speculation. Towns perhaps need to be encouraged (like, repeatedly) to avail themselves of the coaches. | ||
|
GMarshal
United States22154 Posts
Coaches are the way to go over vet smurfs in my opinion, but then people don't ever use them for some reason. Hosts might do well to promote the fact that the coaches are there (Maybe a reminder in each day post?). I don't actually think there is a solution to this problem, I think we just have to allow towns to do badly, its how you learn. Its not really an issue as long as the players come out of it and do better the next game. | ||
|
Palmar
Iceland22633 Posts
On June 07 2012 21:57 GMarshal wrote: Any setup where there can be a cop protected by a medic and the mafia team can do diddly squat about it is an inherently bad setup, since it replaces rational thought with "cop!". I'd argue that its optimal for the cop to claim day 1 in 2of4. (this is obviously subjective) Coaches are the way to go over vet smurfs in my opinion, but then people don't ever use them for some reason. Hosts might do well to promote the fact that the coaches are there (Maybe a reminder in each day post?). I don't actually think there is a solution to this problem, I think we just have to allow towns to do badly, its how you learn. Its not really an issue as long as the players come out of it and do better the next game. I don't agree with your perception of the follow the cop problem, but that's not really what we're discussing. What I really would like to know is some statistic on how well these newbie games have worked to bring people into mafia. I would be very interested in seeing these numbers (but I'm too lazy to gather them).
If we find that there is no correlation between losing and leaving, there is no problem to fix, and you're completely correct that we can just let them play balanced games and push for them to play better. However, if we discover that we're losing potential players because their first game or two are too hard, we should perhaps not shy away from using training wheels in the newbie games. | ||
|
Radfield
Canada2720 Posts
| ||
|
Dirkzor
Denmark1944 Posts
I'm not sure they are introduced to the game properly when everyone is new at the game. The newbie games doesn't flow the same way as other games imo. Regarding who stay around to play after their newbie games... Radfield will find out =) | ||
|
Bluelightz
Indonesia2463 Posts
Am I Man enough to be considered a vet >.>? | ||
|
DropBear
Australia4365 Posts
| ||
|
rastaban
United States2294 Posts
On June 08 2012 02:34 DropBear wrote: so many invite only games T.T maybe with a couple wrapping up a few more might get started that are open to other players. | ||
|
blubbdavid
Switzerland2412 Posts
On June 08 2012 00:45 Bluelightz wrote: I think I'll ask the (not so big) question now, Am I Man enough to be considered a vet >.>? Vets are people who joined before the announcement of sc2 | ||
|
Meapak_Ziphh
United States6786 Posts
| ||
|
Radfield
Canada2720 Posts
SNMM 1,2,4,6,7,8 NMM 0-8 A total of 15 games. The most recent SNMM and NMM games I didn't bother to check, as those are a bit too recent to get good statistics from. I was aiming to see how many players only played 1 game, played 2 games, or played 3+ games. It's a bit tricky to calculate, as some players(say from NMM8) have only had a couple months to accumulate games, whereas players from early on have had over a year. 144 unique players were involved in the 15 games. Of those, 54 played only that game and have not played again. 20 players played 2 games(some of these have not had much time to accumulate games and have basically only played 2 Newbie Minis, others have had lots of time and are obviously not coming back). 70 players have played at least 3 games, and I'd guess at least half of those players have played many more than 3 games. However, I stopped counting at 3 games, as that was the most time consuming part. So, we're losing about 38% of players after 1 game. We lose an additional 14% after two games(this number is probably lower though). Just under half(48%) of the Newbie players stick around for at least 3 games, some of them for many more than 3 games(Palmar I'm looking at you). Overall: 1 Game 38% 2 Games 14% 3+ Games 48% So lets look at W/L records. For players who played in multiple Newbie games I tried to tally their first game obviously. 49 players won their first game, 95 players lost their first game. Of the winners: 1 game: 32% 2 games:12% 3+ games: 55% Losers: 1 Game: 40% 2 Games: 15% 3+ Games: 45% More winners stick around, though not by all that much. How bout depending on if people rolled mafia or town in their first games? 109 players were town in their first game, 35 mafia. Mafia 1 Game: 43% 2 Games: 11% 3 Games: 46% Town 1 Game: 36% 2 Games: 15% 3 Games: 50% Despite the fact that rolling town in your first game pretty much means you lost, there are actually more townies sticking around on average than mafia first timers. This got me thinking, what percentage of players who rolled town AND won in their first game stuck around? Unfortunately only 21 players have performed the feat, so the sample size is rather small. Town AND Won in first game: 1 Game: 29% 2 Games: 14% 3 Games: 57% Unfortunately, 1 of the 3 games that was won, also happens to be SNMM1. That game was played ages ago, so the players have had a lot of time to play games. Anyways, it's possible there could be a strong correlation between being both town AND winning, and how long a player sticks around. Unfortunately Newbie towns don't win enough for us to find out... TL;DR There doesn't seem to be a particularly strong correlation between Winning/Losing and games played, nor between Mafia/Town and games played. It's possible that there is a correlation between Winning AS town and games played, but our sample size is too small. At any rate, a 38% attrition rate for new players in Newbie games isn't terrible. Ideally though we can construct newbie setups that head towards a 50% win rate. Problem is, we don't want to do it by just adding power roles, as that doesn't train the kind of player we want. The game of mafia is won by analysis. Power roles play a part, but they lose games for town as often as they win them. Analysis is the real key, and that takes WORK. We want players on this forum who are willing to work, and I think we are getting them. | ||
|
Radfield
Canada2720 Posts
On June 08 2012 04:10 Meapak_Ziphh wrote: I'll just chime in and say that I don't think a lot of people tale advantage of the coaching opportunities. When people like Radfield offer to coach there's a wealth of knowledge there that goes untapped. There needs to be some way to encourage people take accept coaching. I agree with this. We need to stress coaching, certainly a lot more than just a line at the top of the OP. Perhaps our coaches should be allowed to have more initiative, actively PMing people to see if they would like advise or critique, and offering unsolicited advice to players. As long as coaches aren't PMing their reads directly, that could work well. | ||
|
Mattchew
United States5684 Posts
| ||
| ||
