Spring RTS Engine

Is there any chance the lobby can have correct ranking for wins-losses-draws? Having it based on time means nothing, in my case being one of the worst Spring players ever. Then matches could be balanced much more fairly.

If it were wins/losses you would get cheating, intentional disconnects, and other forms of fail by people who want high ranks.

Play ladder games.

The ladder system sucks, it is, by the admission of its supporters, made for tournament play, not for every-game-ranking.

It also weights things too heavily towards noobs, starting them somewhere near the middle of the board (actually, sometimes, kinda near the top).

IMO it stays time played... assuming it doesn't get removed all together. Ranking in a sensical way is impossible, the quality of the players won't match a numerical rank no matter how the ranking system is designed. The current system is as best as we can possibly get... asside from possibly expanding the ranking so there are more ranks above 100 hours played or whatever our current top rank is...

ladded system is cool

imo the time system is better than no ranking cos there is a difference between a rank 1 i havent played b4 and a star rank dusnt mean the star is pro but a star player wud give a better game than a rank1 who dust know how to play especially the ones who havent played ta b4

SwiftSpear wrote:The current system is as best as we can possibly get... asside from possibly expanding the ranking so there are more ranks above 100 hours played or whatever our current top rank is...

I can't see higher time-based ranks being terribly useful (although better than the rank-whoring that would result from automated WLD ranks *shudder*). It seems that after 100 hours you get about two types of people:

1. Non-fail people who get better the more they play.
2. Speedmetal / Greenfields players. It is a well-documented fact that Speedmetal and Greenfields kill brain cells, ergo the increased experience of these players with time is canceled out by their diminishing number of brain cells.

So, unless there is some way to make Speedmetal/Greenfields games not count toward total time...

A one strip is more likely to be a good player than a star, as a one-stripe is more likely to be a smurf than they are to be a new player.

Saktoth wrote:A one strip is more likely to be a good player than a star, as a one-stripe is more likely to be a smurf than they are to be a new player.

Nobody is contesting that.

The point here is that new users ranks fluctuate wildly and are highly unreliable untill they have a certain amount of experience.

As such any ranking system would continue to use the star rank system we have atm based on experience, and then start displaying the proper ranks when and only when a user is experienced enough to have reached star or 4th rank.

But because new users are immediatly shunted onto the ladder system, their scores fluctuate wildly as expected and it shows up immediatly and confiedence in the ladder is severely dented.

I remain strongly against any attempt to rank on W/L in the main community.

And any effective ranking system wouldnt rely solely on win loss ratios and allow people to opt out of the ladder lists so that they are just a score rather than nth best player.

What about both?
Use the time system, but also show their win lose ratio

well in an ideal ranking system score would start to drop over time after so long if the player stops playing. And the score would not be used untill a minimum amount of experience had been met. After the minimum experience, the score starts to settle closer to the true appropriate value.

Another optimization would be to slowly diminish the gains of fighting the same person over and over again, but do it so that the effect reverses if the 2 players dont fight eachother for a while to prevent deadlock while accounting for player X learning howto defeat player Y rather than player X improving their skills.

maybe use something similar to the glicko system? http://en.wikipedia.org/wiki/Glicko_rating_system

would need adjustments for team play etc.

Still, how do you count the wins and losses? What if a player disconnects, was that a ragequit, RL demanding attention or just a lost connection? What about a desync? What if .cheat gets invoked, can you track that from outside the game? What about mods other than BA (which consequently noone will play because it would hurt their rank) or map diversity? What about handicap? There's way too many variables to say every match was played fair and even if it looks fair at first there's no guarantee that it'll stay like that.

Seriously, tracking wins and losses on a platform as heterogenic as Spring is futile, there's dozens of mods and each player has different proficiencies at each, there's gamebreaking maps, there's team or 1v1, ... Even with perfectly accurate tracking there's no way you can get a single score for a player.

And stats lead to stat whoring and then it gets even dumber than now.

ragequits happen often enough as it is. I personally wish there was a way to find out how frequently a user ragequits(WITHOUT LOOKING AT DEMOS)

smoth wrote:ragequits happen often enough as it is. I personally wish there was a way to find out how frequently a user ragequits(WITHOUT LOOKING AT DEMOS)

You can't figure out which quits are ragequits.

You could see which are quits using the Quit function from Spring and quits because of lost connection (which are treated the same as crashes by the server).

However, I could be like "OMG I R GUN QUITE" and then Task Manager -> End Process -> spring.exe and it would show as a disconnect.