Originally Posted by Boredpayne

https://www.reddit.com/r/GlobalOffen...anking/ckfhfir

But wasn't your point:
Well again, it's sort of ignoring the point that Glicko-2 is on a conceptual level just adjustments to the Elo model that account for statistical uncertainty in ranking. They both work on the assumption that there is a normal distribution of your skill, and your chance of beating an opponent is based on that distribution and their distribution. So what you're arguing (at least, I can't see any other argument being made) is almost every single major competitive game in the world uses Elo without any tweaks to handle ranking uncertainty. But if we look at other games that apply similar systems to their ladder:



Oh, okay. This sounds like Glicko-2, but I guess you could argue it's some other totally different algorithm for processing uncertainty via probability distributions. It's kind of moot.

Originally Posted by Boredpayne

[The game that spawned both should probably be enough on its own, but sure, 'most chess;' additionally, the xbl ranking system, Guild Wars 2, at least something resembling it from CSGO and Dota 2, and pretty convincing evidence that none of the above use JUST the original Elo model. Oh, and Rocket League uses some variation of it or TrueSkill as well, because why not. Links here: http://www.rocketleaguegame.com/news...ranked-season/ and here: https://www.reddit.com/r/RocketLeagu...ow_about_your/

I don't see this going anywhere, so I'm out. It should surprise no one that models implemented in 1960 are no longer industry standard.

I don't get the argument. In toribash with glicko-2 you can beat average players to achieve and hold a high rank. But you can do the exact same thing in placement matches in Toribash in an ELO system. Either allow players to climb quickly or don't.

In the first place, Toribash has little in the way of effective structured matchmaking, so comparative ranking systems are fundamentally pretty awful. Short of introducing this structure or punishing highly ranked players even for winning against lower ranked players, there's no meaningful solution for the particular issue you bemoan. You can play every single one of your placement matches against awful or tremendous players.

Originally Posted by Ele

This isn't a critical point to get bogged down on. Do you have any ideas for improving the current system? It's tiring seeing you shit on everybody else's suggestions without offering an alternative. Be constructive. You know I hate these sort of discussions...

I don't follow this competitive gaming thingo much, but what BP wrote made sense to me.


Seems like we should focus on improving matchmaking, strengthening the foundation of this whole thing, before we think about adding more extensions or changing anything. Matchmaking looks like it's at the core of the reason why the ranking system is shit, so lets make it better. In what ways can we do this?

Originally Posted by Ele

This isn't a critical point to get bogged down on. Do you have any ideas for improving the current system? It's tiring seeing you shit on everybody else's suggestions without offering an alternative. Be constructive. You know I hate these sort of discussions...

Seems like we should focus on improving matchmaking, strengthening the foundation of this whole thing, before we think about adding more extensions or changing anything. Matchmaking looks like it's at the core of the reason why the ranking system is shit, so lets make it better. In what ways can we do this?

Originally Posted by Ele

I don't follow this competitive gaming thingo much, but what BP wrote made sense to me.

Originally Posted by GnilRettemHC

No one has voiced any kind of concerns so I guess we can conclude there's no real problems with the current system? What exactly are we improving?

Striated seasonal ladders with placement matches would be an improvement, but not a necessary one.

What BP said is self serving nonsense.

He's grasping at straws trying to prove that because X game uses it (where X is Rocket League, GW2, and Chess - and tbh the first 2 I'm not sure if they actually even use Glicko-2, since BP keeps trying to bend over backwards to fit a square into a circle) we should absolutely use it. And why does he think these games have the best rankings? Well, because they use Glicko-2 of course! It's some nonsense circular logic.

Not to mention the whole "Chess changed to Glicko-2 therefore it must be better than ELO for Toribash" logic, which basically amounts to "someone is using Glicko-2 so we should as well". Chess and Toribash may both be turn based adversarial games, but that's about all they have in common. Is there any proof that Chess changed because Glicko-2 is better?

Is there any proof that these advantages are transferable to Toribash? I asked above if there was any advantage to Glicko-2, and as it turns out, there is NONE (at least none that anyone can tell me, why someone would hide such a crucial piece of information is beyond me!).

1) Glicko is proven to be more accurate on average (measured in terms of how often it is able to predict who will win a given match).

2) The Glicko rating system adds a reliability score to the rating number, which may provide a higher fairness regarding the rating score adjustment calculations.

3) The main issue with Elo was that ratings moved too rapidly. Against a player of equal strength, it was possible to gain or lose 100 points in less than a handful of games.

4) (In Elo) Most players (of vastly different strengths) being bulked together near 1200. This was incredibly problematic.

5) After changing to Glicko-2, players have been far more evenly spread-out within the rating ranges, making it far easier to find an opponent of appropriate strength. It is also far easier to maintain a rating.

Originally Posted by SmallBowl

Im not sure if you mean the Elo system or everything else surrounding it. Either way Im not a big fan - Ive written up a rework using the Elo system which I can no longer access :v

But I think itd be a big step forward if we were to ditch Elo and move to Glicko-2, its a very good system used widely, even in chess now (which Elo was created for).

Originally Posted by SmallBowl

Nah, the last system used in Toribash was still Elo, that hasnt changed - the only difference really is that only some rooms are ranked rather than all rooms

Originally Posted by Dare

My opinion from old ranking system to new one.

Old ranking system:
you can clearly define the difference in skill between ranks

The new ranking system right now:
randoms are at the top. Some with no skill at all. All they do is grind.


Tbh the new system ticks me off idk y.