Unless I'm victim of confirmation bias, it seems balancing of 2v1 teams always result in highest-ELO player being alone. Same kind of result for 2v3 and more... best player getting 1 teammate less and 2 coms, allowing rushes. I think this is generally unfair... cases like 1400 + 1650 vs 1850 (getting 2 coms) feels wrong. Its not better in 4v3 (just an example), where highest ELO player ends up in smallest team and gets a bigger share (1/3) than best player of larger team (1/4) and twice as much starting resources/com power. And this is even more obvious: * Springie1 3 players balanced respecting clans to 2 teams (ratings 2=1841:1=1308) Where 1841 is 1 player and 1308 is 2 players average. I'm sure it would be more fair to put the 1200-ish player with the 1800+ so the 1800+ has to share resources. Alternatively, not giving 2 coms to the "best" player when the ELO difference is so large wouldn't be unfair. Or getting 2 coms but only 1 free factory... so brawler/bd rush becomes a less obvious/successful choice for the lone player. What do you think?
+0 / -0
|
I think if Elo work by averaging then if you got Elo 2000 and some newbie got Elo 20 then sum of both of you then divide by 2 will gets Elo 1010 (which is below newbie level!)... (but I don't know how this work! I'm just assuming k...). Also there's change in ZK's communism that give moar resource to proactive player (eg: securing mex, build E). So being team-ed with newbies shouldn't effect much (right?), therefore the average Elo (eg: 1010) will not be important anymore
+0 / -0
|
This is shortcoming of the current balancing algorithm. Try to alter it.
+0 / -0
|
That would be a "make stronger player weaker" option, sheep. I think the current "make weaker players stronger" option is better. It just doesn't work with an insane 500 elo difference. Games should need "!forcestart" when elo differs by more than 400 or so.
+0 / -0
|
Agree with Skasi. I honestly don't think that the second comm thing is appropriate for 3v3 and less and is only barely appropriate for 4v4. Anything above that and it works fine, any rush you do can be dealt with fairly easily if the enemy coordinates properly and the elo differences are not likely to be as large.
+0 / -0
|
A !resourcebalance or !handicap option would be nice, to redistribute the initial resources of each team based on elo. Something like this: Team 1's average elo:1400 Team 2's average elo:1600 difference = 200 So reduce team 2's initial e & m by 200, and increase team 1's by 200. So if a high elo player gets 2 comms and facplops in a 2v1, it's less of a problem, as it'll be harder to rush. The result: fewer steamrolls and more close games. The trouble with this is it would bugger up the elo system a bit, as less skilled players would win more often. But that's a different, and lesser, issue.
+0 / -0
|
Handy caps are not fair. They reduce one team's options.
+0 / -0
|
Handicaps aren't unfair, they just make for a different game. In the same way that a team game and a 1v1 are different games, or a big map and a small map are different games. Handicap games are inherently asymmetric, but that's not a flaw. Many very good games are asymmetric. Handicaps allow for games that are more interesting and fun for both sides, even when the two players or teams have very different skill levels. Handicap systems are a very, very good thing to have available as an option. The canonical example of an outstanding handicap system is Go, where you can easily play an evenly-matched game between two very unevenly-matched opponents. Devising and testing a workable handicap system for Zero-K would give the game a valuable asset.
+0 / -0
|
There is the danger of a player learning strategies that cannot work in a non-handicapped game.
+0 / -0
|
Personally, my ideal solution in a 3-player or a 5-player game would be to fall-back to FFA... but I like FFA.
+0 / -0
|
GoogleFrog: One can learn strategies in FFA that won't work in 1v1. Or learn in 1v1 that won't work in teams. Etc. I don't think it's a problem. Again, consider Go. There are definite differences in play between even games and handicap games. But it's still Go, and the games are still good.
+0 / -0
|
Handicaps destroy elo as a rating system. If you have a 400 elo diff. You should expect the higher elo to win. Handicaps can only work in unrated games. As a hypothetical question, who will win this game. 2100 + 1150 vs 1750 + 1650 This will be a 2vs1 game.
+0 / -0
|
"Or learn in 1v1 that won't work in teams" Uh, care to name/describe a few? Everything used in 1v1 is also used in teams.
+0 / -0
|
Yes, with 400 ELO difference and twice as much resources as best player of the duo team, what else can you expect but increase even more the ELO difference...? And with current zero-sum ratings, it means higher ELO will quickly increase thanks to 2v1. "Destroying ELO" is a radical statement, still, maybe some thought about how to improve it wouldn't hurt. Some players are an handicap, although not an "artificial" one like go is using... because they waste part of their share, and because they may need tutoring/indications. Its easy to have all of the team forces mobilized on 1 attack when you are the whole team. Instead of 2v1, playing FFA where the 2 lower-ELO players make an alliance would be more fair. With "fair" being that outcome is not obvious beforehand. Because the job of a game is to be fun/open, even for competitive games. I don't think its fun to play a game when outcome is known... and the outcome of the hypothetic 2v2 game jasper asks about isn't as obvious as 2v1 (what this thread is/was about), although I'd bet on the avg team. I would play that game. But I wouldn't 2v1 jasper if I'm given a 1200 teammate...
+0 / -0
|
Sheep 1750 + 1650 I would likely lose. 1850 + 1650 I would lose 100%. even 1650 + 1600 2vs1 I think I would have a high likelihood of losing. Handicaps creates a race to the bottom. Artificially handicapping one team makes everyone equally bad. Saying it would destroying the ELO system is not a radical statement. Elo is based on likelihood of winning. A handicap reduces that likelihood hence destroying the ELO system.
+0 / -0
|
I don't understand the "race to the bottom" thing. Neither do I understand how handicaps would make people "bad". There is no absolute value of "good" or "bad", its about relation between skills of both teams. When talking about handicap... it could be (what you are thinking about) a way to penalize the best player. Or it could be a bonus given to the worst player. Handicaps would level chances of winning, and thus, create an incentive to play games that look/are strongly biased, because you may still have a chance to win even if opponent is much better rated. Obviously, when using handicaps/bonus... reward should take into account the handicap itself, or it would be even more borken... But with zero-sum ELO, this seems impossible to do... So probably, "destroying" ELO as it is would be required to implement handicaps. So what? Note that I'm not convinced about artificial handicaps, but I'm not more convinced about ELO as a way to rank people given how easy it is to favor games whose outcome seems obvious...
+0 / -0
|
If a handicap system can be developed which accurately balances games where the two sides have mismatched Elo ratings, then the Elo calculation method can be enhanced to accurately compensate for handicapped games. So, no, a handicapping system would not destroy Elo. It would instead bolster the purpose behind having Elo in the first place, which is to create balanced games. Games that can't be balanced by moving players around - such as 1v1 games, or games where players want to use pre-established teams (i.e. clans) - can still played as balanced by using handicaps, if the players all agree to do so. Edit: Okay, I understand now what Sheep was saying. To clarify - I agree that if a handicap system was developed, the Elo calulation method currently in use would have to be changed as well. Doing the one without the other would cause problems, clearly.
+0 / -0
|
How are you doing to create handicaps without using ELO or some other rating system? In chess, handicaps are normally informal and are used in unrated games. I think a better solution is to have an elo floor where you need at least 1600 elo to play in some rooms.
+0 / -0
|
I don't think handicaps work here. There are some pairs of players out there I could beat with 2x metal mult for their side. And there are some players would lose if they were playing speedmetal against me normal. It isn't quite the same as chess or Go, where you can say "here's a thing, there it evens it up" because ZK isn't easily quantified. Two players have twice the attention of a single player. Beyond a certain base level where your opponents don't know what they're doing, it gets almost impossible to win as the single player vs 2. 's why one of my favourite battles was ChimChimTheMonkey winning 1v3. They might not be as good, but their ability to look at multiple places really stacked the deck.
+0 / -0
|
I dont know how we could devise a fair handicap system, which is why I'm not advocating for it :) Some rating system would ofc be required, its the zero-sum quality of it that may have to be changed. Or maybe not, if the handicap system is so perfect. All in all, I think what we have is ok in most cases, but maybe we can still improve it. The 1v2 farming is what bothers me...
+0 / -0
|
