🌿ELO System Updates

[ThreadPacifist]

I'm not a fan of these changes at all. Elo decay makes no sense when you already fall off the leaderboard for inactivity and it punishes players who play a large variety of games. Arena already exists for resetting of rating, so this change feels so unnecessary and negatively impacts many players.

I am also not a fan of the elo decay feature. Last season, I won the Photosynthesis arena. There were only 656 games played that season between 163 players. I preferred to play real time which meant that it was hard to find games and the people that I found tended to be the same small group of people. My most frequent opponent was 15 games vs them with 15 wins for me. Why should I be punished by getting 0 elo for each of my wins because the player base was small?

[Silene]

... but every single user here has ONLY said that these changes should be removed. No-one has proposed any ways to fix these issues.

This is just a blatant lie. Many alternative solutions have been suggested.

Glicko. Reducing K factor, not just removing the gained points.

... and these were among them.
Reducing K factor has been brought up starting page 2 and has repeatedly been suggested as an alternative.
Glicko is a great suggestion from you and people who are criticizing the implemented changes have confirmed that would be a great alternative covering the needed stable hidden rating (which is used for further game-result-evaluations) while showing a lower displayed rating after phases of inactivity.
And there were several more other possible fixes suggested.

However the way things are done now are just worse than before. You are defending the changes that were done with the quality of an alternative imaginary way that is not implemented. If that imaginary way was implemented, we wouldn’t fight it this way. But it’s not.

[Silene]

Many players here complain that elo changes make scores lower... as if elo is supposed to increase indefinitely.

The "as if" makes no sense here. It is true that we complain that the changes make scores lower. But it does not mean that we want elo to increase either. What we want is elo (the distribution within the community) to be as stable as possible. And the previous system did that better than the new adjustments.

Yes, the new way ultimately will get to another stable state. Because when everyone's elo went down, the lost points from power-leveling and decay will become lower and lower over time while the points gained in the system by players bumping into 100 will get more and more (because so many people will be closer to 100).
But we do not want a stable state that has the best players (of highly skill-dependent games) around 300 elo, sustaining their higher-than-average-score by bumping new players into 100 while avoiding repeated good opponents as if they weren't fun to play with. While the implemented way down will take some time to get there, the bad feelings about meeting players again is already in place and painful. It's great and should be supported to play with new people. But it's also great and should be supported, to play with people again.

[Patrick of the Isles]

So to sum up where we stand: BGA admins made a significant change to an important site feature in an attempt to address user concerns. We now have a 20+ page thread of many frequent site users who play all different types of games saying these changes unfortunately made the site less usable and less enjoyable. Other than one account who has posted about 50 times on this thread to share that they like the changes, the reception has been universally negative. Since a fair amount of time has passed we are beginning to ask if we are just shouting into the void, or if the site admins are taking this feedback under advisement?

[postmans]

Please revert the changes for Power leveling and ELO decay.

Instead of power leveling just ban cheaters, or turn-off ELO for ELO boosters (they can keep playing but no longer will have an ELO).

Instead of ELO decay just hide the players a bit quicker from rankings, e.g. after 30 days of inactivity.
ELO decay will cause players which are only occasionally active to show way lower ELO than there actual skill.
Even for those who don't care about their own ELO this will have negative effects:
- In some games good players set minimum ELO requirements, you can't join your fellow top players.
- In other games people set a max ELO, they want somewhat of an equal fight I assume, if a hidden-pro joins that table it's no fun for anybody.
- Personally I've experienced that opponents aren't always a fan of being beaten by much lower ELO players. For some games I've either played them offline, got tips from top players or in another way learned the game to a decent (300-600 ELO) level before starting to play. On several occasions where me as a <200 ELO player beat a 400+ ELO player this has resulted in the opponent not appreciating the game. Sometimes even red thumbing me or insulting me. The new ELO decay introduction will give this experience to many players who come back after (longer) breaks of playing a game.

Sadly the actual interesting problem of the ELO system wasn't addressed: How to fairly/quickly estimate the correct level for new players.

[Wreckage]

The goals:
1. To continue to have a system like most of us have enjoyed.
2. To not have players who are basically inactive parked at the top.
3. To discourage cheating using multiple accounts.

The answers:
1. Revert to the previous system for now, being open to other ELO fixes that aren’t the topic of this thread.
2. I suggest a single 1 point ELO decay every 3 days for ONLY the top 20 ELO leaders of each game (not arena leaders). The vast majority of players would not be effected. The top rated players would be effected equally. Perhaps the number of top players effected could be adjusted by devs, based on a game’s limited pool of players.
3. I think my previous suggestions help with this too. It doesn’t solve cheating, but it will make it more work.

Confident competitive players should welcome these changes, while those less confident leaders will want to keep their ELO without competing for it anymore.

The current concept of playing a single non-ranked opponent every 60 days (they can be later) to keep the top position for years and years is not good.

[franko]

So to sum up where we stand: BGA admins changed the site without any advance consultation. We now have a 20+ page thread of many frequent site users from all different games saying these changes made the site less usable and less enjoyable. Other than one account who has posted about 50 times on this thread to share that they like the changes, the reception has been universally negative. Are we just shouting into the void? Do the site admins have any interest in the user experience?

I'm not sure. I just played another ranked game and realized that the streaks aren't indefinite as I originally thought (although having clarity on what defines a streak would still be good / I still think it needs tweaked from the current state).

From 12/14/2023 https://boardgamearena.com/table?table=450231855 -- I earned 3.04 ELO (25% consecutive win penalty), they lost 4.06 ELO. Our game history https://boardgamearena.com/gamestats?pl ... me_id=1225 has me winning 5 consecutive games, although the streak was computed as 3. So that seems to imply it was computed as 11/28/2023, 15:45 to 11/29/2023 at 15:26, and 12/14/2023 at 17:07. That's between 2 and 3 weeks of time or 164 games across 125 unique opponents (based on my games history) and still counted as a streak.

From 12/19/2023 https://boardgamearena.com/table?table=451946662 -- I earned 7.77 ELO, they lost 7.77. Our game history https://boardgamearena.com/gamestats?pl ... me_id=1225 has me winning 3 consecutive games, although there wasn't a streak penalty included. My last win there was 11/17/2023 at 15:38, so about a month, or 285 games.

It's not clear to me what resets a streak, but they aren't indefinite which is at least good. I still think the current implementation isn't sufficient for games where the player-base literally prevents active players from gaining as much ELO from simply playing games in a non-abusive manner. I'm defining active as someone who plays at least 200 tables of a game in a month, which sure might sound like a lot, but why should they be penalized for being active? I'm all for ensuring ELO represents strength of active players, but the definition being used for streaks seems to be off for at least some games on this site (whose player base simply isn't large enough to avoid "streaks").

I'd propose introducing another way to reset a streak -- e.g. playing against N distinct opponents since your last game would reset a streak against an opponent, where N can be based off of a % of the total active players for a game so that it can reflect the difficulty in finding unique opponents when a game has a small player base while then adjusting if that player base were to grow. That way, I'm still confused and disincentivized to ever hit the "propose a rematch" button (since that's literally forcing a potential streak), but if I play dozens of games before rematching an opponent I'd perhaps not get hit with a penalty.

Additionally, I'd agree that the ELO Decay as proposed is useless. I think there's two easy fixes that would make everyone happy based on complaints I've observed in this thread:
* Do not apply the decay for people who no longer show up on leaderboards. I believe that happens if you stop playing for 2 or 3 consecutive months, and while your ELO there is obviously not reflective of your skill "against current players" it doesn't show up. This lets people who have moved on from a game maintain their "historical ELO" for whenever they come back.
* Apply the decay more meaningfully for people that do show up on leaderboards. Make it so that they have to play N games where N is based off of a percentage of games played during the last arena season across all players for the game (and make sure this is visible for the game so that players know how many games they have left to play to maintain their ELO). Having to play 1 game per arena season (currently proposed) does nothing to force activity. If this was something like 100 games, that would actually cause players parked on spots to expose themselves to a significant enough number of games to have their ELO not be stagnant.

Making the ELO Decay threshold be based on # of games played during an arena season would, I think, also disincentivize players that make multiple accounts to hold either Arena or ELO spots, since they'd then need to play M*N games (M = accounts, N = the threshold based on last season # of games). I suppose some people would still be willing to do that, but that also seems potentially easier for BGA to identify and crack down on (although since I'm pretty confident that 3 players in the top 20 arena spots for Kingdom Builder are the same person, and that's happened for at least the past year, I'm not sure they're actively trying to stop that).

[Meeplelowda]

Instead of power leveling just ban cheaters, or turn-off ELO for ELO boosters (they can keep playing but no longer will have an ELO).

This would require a manual assessment of whether someone is ELO boosting. It appears to me that what they want is a system that is auto-moderated so that it doesn't consume administrative resources. But the system they chose has negative consequences. I won't say "unintended" consequences, because I can't know whether they 1) just didn't anticipate the effect or 2) anticipated it and decided to go ahead anyway.

[sprockitz]

Instead of power leveling just ban cheaters, or turn-off ELO for ELO boosters (they can keep playing but no longer will have an ELO).

This would require a manual assessment of whether someone is ELO boosting. It appears to me that what they want is a system that is auto-moderated so that it doesn't consume administrative resources. But the system they chose has negative consequences. I won't say "unintended" consequences, because I can't know whether they 1) just didn't anticipate the effect or 2) anticipated it and decided to go ahead anyway.

But this makes it so much worse for legitimate players than for people Elo boosting bc Elo boosters won’t ever lose to their multi accounts whereas real players sometimes will even if they have a 400 Elo difference. So it brings down non cheaters more than cheaters and thus actually benefits the cheaters

[ChiefPointThief]

The goals:
1. To continue to have a system like most of us have enjoyed.
2. To not have players who are basically inactive parked at the top.
3. To discourage cheating using multiple accounts.

The answers:
2. I suggest a single 1 point ELO decay every 3 days for ONLY the top 20 ELO leaders of each game (not arena leaders). The vast majority of players would not be effected. The top rated players would be effected equally. Perhaps the number of top players effected could be adjusted by devs, based on a game’s limited pool of players.

This is even worse than what we have now (at least for me) I am currently top 20 in 10 games and have been and could get into top 20 again in another 10 or so. You are saying I would have to play all of these games every 3 days. Even my favorite game (Sobek 2players you guys should try it ) I've gone a month or so w/o playing it several times.