French lobby - Page 2

French lobby

Please use this forum to set up matches and discuss played games.

Moderator: Moderators

User avatar
Cheesecan
Posts: 1571
Joined: 07 Feb 2005, 21:30

Re: French lobby

Post by Cheesecan »

bibim wrote:
Cheesecan wrote:But then a better matchmaking system made up of leagues would have been better. The core of the problem is that newbies should simply not play with pros. Look to the StarCraft 2 ladder system for a proven way of handling things.
Is it really realistic to think about such a league system with current spring games player bases?
I agree that a true matchmaking system should be better though.
We may have had say 10,000 players leave and go elsewhere since 2005. A league system is one of the things we need to reverse the negative trend in player base. It's absolutely necessary that newbies can have a safe haven so they can learn the game before being matched up against guys who long got past that, and don't even empathize with newbies problems because they forgot it nearly a decade ago.

We tried newbie hosts, but in the end people join them due to scarcity. If you look at Starcraft 2, the engine supports perhaps 8v8 but ladder only goes up to 4v4 ranked games. This is perhaps not only due to technical reasons but also likely because there's some quality assurance thinking behind it. E.g. 2v2 waiting time is just a fraction of the waiting time for 8v8, and there are plenty more opponents to choose from. Now that's for a huuge server like b.net, and they still decided larger games would just have too long waiting times. Then why are we having 8v8? The community is too tiny for it for sure.

Add ladder play for 1v1,2v2,3v3,4v4, and mimic the hidden MMR(matchmaking ranking) system of Starcraft 2. In essence, you have a hidden MMR then you are promoted/demoted in league after the system is very sure that you will have the nearest to 50% winning chance in your new league. In effect this means players will face off with at most one level above or below them, depending on their current win streak. This goes from bronze,silver,gold,platinum,diamond,master,grandmaster so there's a clear progress incentive to advance. Meanwhile players who are terrible can comfortably face others who are terrible. Without getting humiliatingly defeated. Research says that getting really good at something takes a certain amount of hours of practice. You can't "force" skill progress in every person by having them play with their betters in every other game, like we do today. It might even be counterproductive to let newbies play with much better players. They gain bad habits from relying on hiding in a "tech spot" or whatnot and don't learn2play.

Every player should be able to go and spectate or join unranked matches with better players once in a while, but that's a whole different thing than day to day play which should be league-based if you want to retain players.
bibim wrote:
Cheesecan wrote:Here's a thought experiment. Put five 500 ELO chess players together against one 2500 GM. Should the likelihood of any one side winning be 50%? Of course not, the GM will always win. So we then understand that ratings are not cumulative and therefore spring ratings must be flawed for >1 players. This is exactly what was done. This is why you have people shouting bloody murder and stacked teams whenever a good clan joins. This is why we have no real clans left also. Bad model.
Your comparison makes little sense to me. How current player balance system could be compared to several chess players sharing same chess pieces and playing against one GM? Are you referring to the coop mode?
This is disregarding that you have an offset of 2500 m/one commander for each player and greater spatial dislocation and proximity to mexes. Beyond a certain point into the game, a good player could probably micro and macro all across the map better than a bunch of noobs could. So then you have a team consisting of a set number of anchors who can actually survive and face a chance of making their team win, and a bunch of liabilities in the form of noobs that are just an encumbrance that have to be monitored, really.

Every player getting a commander can be both positive and negative, since newbie is likely to lose that commander to a superior enemy, simply resulting in him giving away metal to the enemy and thereby demoralizing his own team. It can also mean outright win in metagaming since you can combomb early game with little risk of losing if you have a superior number of players.

@Silentwings: I will attempt to understand this that you speak of. For sure, I am not qualified to comment on lobby matchmaking before I understand every balancing algorithm there is. How presumptuous of me to imply that matches are anything but perfectly balanced.
User avatar
bibim
Lobby Developer
Posts: 952
Joined: 06 Dec 2007, 11:12

Re: French lobby

Post by bibim »

Cheesecan, if you think a system as you describe here:
Cheesecan wrote:Add ladder play for 1v1,2v2,3v3,4v4, and mimic the hidden MMR(matchmaking ranking) system of Starcraft 2. In essence, you have a hidden MMR then you are promoted/demoted in league after the system is very sure that you will have the nearest to 50% winning chance in your new league. In effect this means players will face off with at most one level above or below them, depending on their current win streak. This goes from bronze,silver,gold,platinum,diamond,master,grandmaster so there's a clear progress incentive to advance.
would actually be used by BA (*) players (who, for some reason, would suddenly become a majority of competitive players instead of a majority of casual players just looking for relaxing gameplay and fun in 8v8), feel free to implement it. I would then do my best to interface it with SPADS if needed.

(*) I'm only speaking about BA here because ZK has its own structure for all that anyway, which I don't really know, and other games have even lower playerbase

Imo, at first a basic matchmaking system would be enough, without this ladder and leagues addition which would scare casual players more than it would attract them.
Cheesecan wrote:This is disregarding...
Exactly, and whatever you can say to minimize this difference, it makes little sense to me to try to criticize current balance system with a so badly chosen comparison. This is like saying current balance system makes 5 newbies share same commander at start and play against a pro.
User avatar
Silentwings
Posts: 3720
Joined: 25 Oct 2008, 00:23

Re: French lobby

Post by Silentwings »

cheesecan wrote: I am not qualified to comment on lobby matchmaking before I understand every balancing algorithm there is.
I suggest try to understand the sort of bayesian update steps in most balancing algos.
Seems a not unreasonble and not too difficult thing to suggest, at the same time as pointing out some problems with your example (and not being the only person to do so).

---
cheesecan wrote:How presumptuous of me to imply that matches are anything but perfectly balanced.
If you can produce a better algorithm feel free - it's easily tested since you can run it on a large sample of already played games and then see how good it as at predicting the results of games. If you manage to do better than what we already have I suspect you'll get more attention.
User avatar
very_bad_soldier
Posts: 1397
Joined: 20 Feb 2007, 01:10

Re: French lobby

Post by very_bad_soldier »

Cheesecan wrote: Here's a thought experiment. Put five 500 ELO chess players together against one 2500 GM. Should the likelihood of any one side winning be 50%? Of course not, the GM will always win. So we then understand that ratings are not cumulative and therefore spring ratings must be flawed for >1 players. This is exactly what was done. This is why you have people shouting bloody murder and stacked teams whenever a good clan joins. This is why we have no real clans left also. Bad model.
You are mixing different stuff up. TS fails to balance matches with uneven player count (like 3v2). Thats a known flaw. That is also what your example aims for.
In everyday games you rarely see those games anyway. Matches with even player count are fine though.

EDIT:
Well, the uneven player count problem isnt even a TS flaw as even a human admin is not able to balance those games correctly.
User avatar
Cheesecan
Posts: 1571
Joined: 07 Feb 2005, 21:30

Re: French lobby

Post by Cheesecan »

Holy grail soiled, defenders of the faith rally to the cause.

Sorry to set your straw houses on fire, please continue to debate bayesian space monkeys while the community continues to split off and dwindle. Remember, the important thing is that you are right.
dansan
Server Owner & Developer
Posts: 1203
Joined: 29 May 2010, 23:40

Re: French lobby

Post by dansan »

Bibm: can the TrainingCamp be configured to use TS instead of Rank to auto-spec anyone with TS>25?
User avatar
Silentwings
Posts: 3720
Joined: 25 Oct 2008, 00:23

Re: French lobby

Post by Silentwings »

@cheesecan: resorting to attacking people personally won't help your cause here.
User avatar
bibim
Lobby Developer
Posts: 952
Joined: 06 Dec 2007, 11:12

Re: French lobby

Post by bibim »

dansan wrote:Bibm: can the TrainingCamp be configured to use TS instead of Rank to auto-spec anyone with TS>25?
It could be. Though this server wasn't meant to be a "newbies only" place originally, just a place where newbies are welcome and shouldn't be yelled at.

I could try to create a newbies only server, though I had already tried something similar long ago (based on lobby ranks) and the server wasn't actually used.
MorgenDavid
Posts: 3
Joined: 01 Oct 2013, 05:05

Re: French lobby

Post by MorgenDavid »

a tips about..

==> based on the skill, more you are strong.. less you can take metal, energy .

example: Here where a beginner can take 4 metal on extractor... a great player can only take 1.

My community come from CSS who i have create a script called "antiroxx" (damage reduction based on skill). this was a success story (server full all the time).

but, for that ... the TrueSkill should be more accurate. take care about some parameters like who is the first come to units t2, number of extractors, etc..

I know it's a work quantity to make , but trust me, that's a great (best) solution.

i hope you to take time to think about this solution a little.

++
User avatar
FabriceFABS
Posts: 354
Joined: 28 Jul 2010, 16:20

Re: French lobby

Post by FabriceFABS »

bibim wrote:
dansan wrote:Bibm: can the TrainingCamp be configured to use TS instead of Rank to auto-spec anyone with TS>25?
It could be. Though this server wasn't meant to be a "newbies only" place originally, just a place where newbies are welcome and shouldn't be yelled at.

I could try to create a newbies only server, though I had already tried something similar long ago (based on lobby ranks) and the server wasn't actually used.
+1 This is a quite recent Spads feature, thx Bibim.
It's a nice idea, I think 50% of BA autohosts should be set for small games and also for newbies only, this is not a shame for them to dedicate their autohosts rank level, really.

As far as I'm concerned, I don't really like big games too.
The players mentality and the particular way of playing on such big games aren't suitable for beginners. I suffered a lot of aggression from some players to me, those players that never play small games, that players that can't stand to loose, and that players that needs to critic to feel better.

I'm still beginner maybe :-) and proud of it.
I don't deserve my actual rank, I really played too much 8v8 DSD for ~ 1 year. After meditation, most exciting games are around 4v4 : Everything is well balanced, solidarity and commitment of players are the most important.
This is where the newbies should play, never on big games.
I will surely do that on 2 of mine quickly and invite other admins autohost to do the same.
And really big games should be forced to accept @ least, ~ 25 TS ?
MorgenDavid wrote:a tips about..

==> based on the skill, more you are strong.. less you can take metal, energy .

example: Here where a beginner can take 4 metal on extractor... a great player can only take 1.
This will allow a beginner to increase his chances of winning against another a little less beginner or better than him, admit.
But there is something essential that I can not seem to grasp : For me, the victory is just under playing. If I play against a beginner that wants to watch game reaplay to learn a little, I really don't know how to explain the way of playing if a lot of restrictive rules are grafted to the player regards to his level ?
This will surely do a lot of intermediate levels and numbers of way to play and could be really really too much complex to learn.
It sounds good, but I'm not sure this is the best way to.
MorgenDavid wrote:My community come from CSS who i have create a script called "antiroxx" (damage reduction based on skill). this was a success story (server full all the time).

but, for that ... the TrueSkill should be more accurate. take care about some parameters like who is the first come to units t2, number of extractors, etc..

I know it's a work quantity to make , but trust me, that's a great (best) solution.

i hope you to take time to think about this solution a little.

++
Well...
First of all, T2 is really NOT the only way : Coming T2 quickly is not a guarantee of player skill level, the goal of the game is to kill commy opponent (for most of games, could be changed), not building towns of T3 for the same goal, even if it's amazing, etc.
So using for instance 10 peewees or 10 flash could let you win in ~5 mins.
Same for this as told just above.

The best way in my way for learning is to play against players of the same level or vs AI.
Changing rules (metal, E, shot power, ...) could be good, maybe, who knows... I admit that all these special rules depending on the rank player confuse me a little.
I prefer spec and talk to the newbie what is good or not during a game.
And remember, this is my think and I don't force anyone to think the same.

Word to the wise.
Post Reply

Return to “Ingame Community”