Ratings for multiplayer games

[clavz]

Hi everybody,

Even if I never post here, I have been playing 0ad for quite a long while: I think I started playing with a17/a19. I was mainly playing with a real-life friend against AI for short periods and then quitting because we felt it was boring. A couple of time we tried to play in lobby, but, since our skills were quite low, we ended up badly losing. This was the main reason that kept us away from joining the lobby more often. During 2020 we started playing a bit more constantly and ability improved (many thanks to @causative for the a21 guide: it is a big help for people starting to play). Some real-life friends joined us and since then we started being online quite often.

 

Getting smashed (without understanding why it is happening) could prevent many from joining the lobby. Now I know the nicknames of many good player and I feel like many in the community know mine, so we normally play more or less balanced matches, but for a long while we couldn’t guess if the matches were balanced before starting.

This limits the growth of the community (a smaller number of active players in the lobby means that less people will be interested in the game and, I believe, its development).

In the lobby there are often private games of people playing against AI: many players start playing with some buddy and won’t play much of 1v1s.

Sometimes I got caught in troll-games where skilled players enjoy smashing noobs (sometimes I have been part of it even in the trolling team). I understand some could get fun from it, but I don’t think hazing helps the community.

 

Making the lobby a bit more noob-friendly would help the community to grow, and the solution is easy: introduce ratings for multiplayer games.

I know that ELO rating system does not adapt well to multiplayer games, but IMHO it’s more important to have some rating that is not precise than not to have it at all.

Having some will help to understand how to improve its precision.

 

I think it could even be temporary (and reset after some period), before a better implementation is found, hoping that the community will take it as a big tournament, and some will try to hack it or find some loophole how to increase it.

 

In the forum I found that some attempts have been made. I do not have programming/statistical skills for implementing it. To me, even summing up the ratings of the players of a team and the ones of the other, calculate the ELO as 1v1 and then spread the difference between players would work.

 

Edited April 3, 2021 by clavz