OR  Zero-K Name:    Password:   

Unable to find matches under current MM elo weight

8 posts, 384 views
Post comment
Filter:    Player:  

8 months ago
(edited 8 months ago)

Now people start to realize that how absurd the current elo weight for mm matching is. This is the reason why we have 45 min q when there is 5 available and can't even launch a single 2v2.

There is nothing wrong with teaming up. RTS is about strategy. It means trust. Trust and predict what your enemy will do, what your ally will do. There is just nothing more comforting than getting "sending penns omw" from your ally when your spider is under firewalker. Fighting with a stranger you have not ever heard or met is not making the game more "strategic".

Teaming up is getting suppressed; the elo matching range will even become shorter according to this post(http://zero-k.info/Forum/Thread/25687), and that's evil. The original author of this post doesn't even frequently play mm anymore and the inability of seeking someone you can trust with and predicting the game got officially encouraged without even a vote at the expenses of other people's (those who are coming prepared and who are working hard to strategically play.) time.

That opens the new point: Why can a single dev with a gigantic permission change a setting without a vote in an opensource community? Dev is having too much rights and abusing rights brings no good to the community. This is what currently happening in other opensource communities too, because core developers are abusing their rights, we have OpenBSD, we have dragonflyBSD and freeBSD, we have warzone2100 and we have linux distribution vs android. Vote is good, and please keep voting even on the very basic dev level. It shows what majority thinks.

Therefore, in conclusion, I do not think teaming up is bad, and its elo matching range shouldn't be deducted.Vote may not be able to solve everything, and for the balancing issue it may also suggest the opposite as this post suggested, however, I believe a vote gives a better result than someone asking someone else who in charge of summer pw and all of the sudden setting changed after all.

+1 / -5

8 months ago
Although there are probably not written objectives, based on what I have seen people talking about it, the current matchmaker is to randomly match people to have equally skilled teams. I claim that the idea of "party" is the thing that breaks the matchmaker, because it is not part of its objective.

You can make custom games and find not only your team mate but also your opponents. I occasionally see passworded games, with usually 4 (or rarely more) people from the same country, which my guess is exactly the case above (people playing in parties against friends parties).

I fail to see how "efficiency" suddenly became about "trust". Good players do actions when (they think) it is good for the team, not because they "trust" their teammate.

All your examples, are not about voting, are people that wanted to prove they can do better and they did that by forking and doing something else. As a matter of fact Zero-K is exactly that, it split of another project in the moment some developers thought they could do better.

For me:
  • matchmaker should not be used for parties (because don't think it can work in fact, due to all reasons)
  • parties should look for other parties (maybe a dedicated queue? maybe a chat channel? custom hosts?)
  • voting to be when developer feel, but they should follow their strategy
  • people should make a new game if they think they can do significantly better and have the time for it
  • (and bonus) objective of matchmaker should not be "equal skills" but "maximum satisfaction" (no clue how to model that though :-p).

+1 / -0

8 months ago
In every competitive game there is a vocal minority like this. They try to claim that some unfair advantage is good somehow when really I think they just want to win.

If you look at the other thread (why didn't you post there?) all except one post are agreeing with the OP. Also -- this is people who post on the forum.

90% of people who experience terrible games like that are just gonna uninstall the game. No post, no visibility, nothing, just gone. Maybe you see it as "Lobsters uninstall lol" or something but its really bad in the long run.

To me its very clear that the community doesn't like this premade teamstacking rubbish -- I think only you would complain if it was banned entirely. Its fundamentally not competitive, frankly.

+1 / -0

8 months ago
People don't like losing because they'll lose elo for it. Why not add an unranked queue which has a much bigger skill gap difference? I know I won't mind losing if it is unranked.
+5 / -0

8 months ago
While I have seen the idea ("don't like losing because they'll lose elo for it."), not sure it is something we need to agree/encourage/fix. What if I say "people like building storage so that they build fast"? Solution to this is not to give infinite resources, but rather to teach (somehow) that you want to build fast you need to expand.

Back to rating systems, the WHR (it is not ELO) is not a "resource". You can't "loose" it. You can be very good (then you have high WHR) - and you need to constantly prove it by playing, or you are not that good (then you have lower WHR).

Now, people are irrational, and this is a game, so maybe we need to tell everybody something that he is the "best" at, even if it is not completely true/relevant. Example: in the last X games you played, you are rank Y in how many turrets you did, how many mexes you made, how much energy you produced, etc. And choose X small enough to be high rank enough to keep people happy.
+0 / -0

8 months ago
The team matchmaker was originally designed simply to make games with a bounded difference between the worst and best player. When there were no parties this works reasonably well, as it is likely that a decently balanced game can be made out of the selected players. Parties broke this system because they can constrain the balancer to put the two best players against the two worst players. A win chance sanity check was implemented but it destroyed the database through inefficiency. So now we have the range reduction as a stopgap.

You cannot say, with a straight face, that there is nothing wrong with teaming up. There are obvious cases in which this is a bad thing. Imagine if there are two people of high skill and two of lower skill that want to play a 2v2. The game will be worse, overall, if the higher skill people play as a team against the lower skill players. This example actually occurs in the MM and, when sufficiently common, will destroy peoples trust in the system being able to generate them good games.

On open source, it is a misconception that open source means that everyone has an equal say. Clearly there are some people doing the infrastructure maintenance and distribution, and this confers some level of control and responsibility. Open source means that it is easier for you to write a better matchmaking system and/or collect data to show how it should be improved. It does not mean that everyone gets control without doing anything.
+5 / -0

8 months ago
(edited 8 months ago)

I would just like to point out what a success this system seems to be. Thread/petition is made asking for some improvement/change. Community discusses and sides for or against the idea. Admins implements favored change. Opposing parties post counter-petitions and community discusses the effects of said change.

Edit: Also, in regards to trust: just like people have the freedom to play with or against others, others have the freedom to play with or against "them". Matchmakers are not designed to just throw arriving lobsters into the fire as fast as humanly possible. They are designed to help match ranks. For everything else, there's mastercard there's custom games.

In memoriam to all those brave souls who didn't make it out of the great Steam release of 2018 alive, rest in peace:

+3 / -0

8 months ago
Kudos for mentioning Warzone 2100.
+0 / -0