Why ELO doesn't suit to BGA

[latexrhombus]

Elo decay is, in my opinion, the worst idea by far. It is unfair, unjustified, and mathematically illogical. It is in contradiction with Elo system itself.

There is no such thing as Elo decay in FIDE rankings. I played the last ranked chess game 26 years ago, and I still have the same Elo (of course, I am considered inactive, and I do not appear on the ranking list).

There is also no decay on Yucata & Brettspielwelt.

You seem to be against every idea laid out here, which is not extremely surprising since you are one of the players at the tippy-top of ratings. Despite your claim that ELO decay is mathematically illogical (it's not), comparing the ELO system here to chess system is mostly far off-base. Many games on here have a certain degree of luck inherent in them, while games like chess have none. Because of the luck inherent to most of these games, players can get uncharacteristic upswings in their ELO not because they are the better player but because they had the better luck.

These players should not be able to sit at the top of rankings for 60 day periods. In my opinion, if someone does not have time to complete a game on a weekly basis, they do not belong on top of the rating system. Obviously, BGA is much more generous with the time restrictions and I imagine they will continue to be, but the idea is to promote PLAY on the website.

Players at the top obviously have a strong disincentive to play, unless there is some sort of decay to encourage them to do so. On the other end, players who want to reach higher rankings often have a difficult time doing so because the players who are ranked higher than them will not play them since it is better for them to sit for 60 day periods. This is not a fun situation for anyone, except those already on the top who do not play regularly.

[glitter and doom]

You seem to be against every idea laid out here...

This is simply not true. I supported the idea of seasonal points with smoothed ranges. I also supported the idea of increasing the requirements for active players. Nothing of that is favorable to me, however, I supported these ideas because they are logical and fair.

...which is not extremely surprising since you are one of the players at the tippy-top of ratings.

This ad hominem argument could easily turn against you. You are supporting the idea that is clearly in you favor (decreasing other players ratings while keeping your own unchanged).

Despite your claim that ELO decay is mathematically illogical (it's not)...

Could you please explain the mathematical reasoning for Elo decay in non-luck games?

Many games on here have a certain degree of luck inherent in them, while games like chess have none.

I agree. So, do you propose Elo decay for all games on BGA, or only games with a certain degree of luck? There are many games like chess on BGA. For example, chess.

[sourisdudesert]

Thank you all for your feedback.

Please remember that again, there is no good / bad / best system. There are just intentions for this service. What is going to happens is that we are going to change some things - after making sure that current leaders are rewarded for their positions - and see if things evolve in the right direction. If it does, we will keep this system. Otherwise we'll drop it. But this is almost impossible to tell without a proper experiment :/

[latexrhombus]

This ad hominem argument could easily turn against you. You are supporting the idea that is clearly in you favor (decreasing other players ratings while keeping your own unchanged).

Actually, it cannot be turned against me. I have no interest in being a top ranked player in any game on here, however, I am friends with people who play on this website (many I actually recommended it to) and several have mentioned that they do not like the ELO system because they feel that it is too hard to move up it since the highest ranked players do not seem to ever play. I would be interested to see what percentage of each game the top 20 plays because it does seem to be incredibly low.

Could you please explain the mathematical reasoning for Elo decay in non-luck games?

BGA is about participation in a great hobby. Even though this site has chess, I think it is safe to say chess players are not going to flock to BGA to play it any time soon. There is no reason for them to do so and games played here would not affect their real ratings in any way. There are plenty of tournaments for them to participate in.

The vast majority of games on this sit, even games like Hive, which are devoid of luck, do not have anywhere near the following (nor will they ever) that chess has. As a result, people NEED TO BE ENCOURAGED TO PLAY! The logical way to encourage someone with a high ELO to play is to take it from them if they do not. I am not advocating large amounts. In fact, your question confuses me, because luckless games should be easier for a better player to win (and therefore play) being that they are strictly a superior player. I would imagine being forced to play a game like Dragonheart is a much scarier experience for the top players (who I predict you will never see playing).

Also, I want to emphasize that an ELO DECAY SYSTEM DOES NOT AFFECT ANYONE WHO ACTUALLY PLAYS! I feel like you are acting like it just pulls top level players down. IT DOES NOT in any way, shape, or form provided that they play at some reasonable interval.

I agree. So, do you propose Elo decay for all games on BGA, or only games with a certain degree of luck?

As explained extensively above, I propose it for all games, since there is no reason for a better player to fear playing a luckless game. Their ELO should only rise. If they do not want to play a particular game on BGA, they do not need to be a top-ranked player in it. It is that simple.

Edit: I guess my point regarding ELO decay in a nutshell is these are not the official world rankings of any particular game (chess is a prime example). These are the ranking for that game on BGA. As a result, to maintain a high ranking in any particular game on BGA, you should have to play it there.

[Game Fondue]

I'm sure BGA will come up with a great system.

I do want to say though that when I proposed an ELO decay, I did not necessarily intend for it to only affect the top ranked players. Also, I was thinking pretty small amounts compared to what has since been suggested. I was thinking closer to 1 ELO point per 15 days of inactivity. This 1 point already costs the player at the top more than it would cost a lower player since gaining additional points when your ranking is harder are more difficult.

People work hard for their ratings, so I do not think they should be taken away from them, but as other posts have suggested, I think the real value of BGA is getting people together to play. Just now, people in the "play now" lobby are commenting that it is hard to find people to play with in certain games. A system like a decay system is good for this since it does not automatically harm anyone or force them to do anything, but it does give them a nudge to host or join a table.

At the end of the day, the more active members, the better it should be for all of us.

I also want to add the main problem I see with a "seasons"-based system is that it is somewhat unfair to players who join BGA in the future. They will never be able to catch up in prestige simply because they missed out on earlier opportunities.

[cosmosis]

Also, I want to emphasize that an ELO DECAY SYSTEM DOES NOT AFFECT ANYONE WHO ACTUALLY PLAYS! I feel like you are acting like it just pulls top level players down. IT DOES NOT in any way, shape, or form provided that they play at some reasonable interval.

I'm a slightly different kind of player than Glitter and Doom. My goal for a while now has been to get to a Good rating on each game on this site. I'm currently >1600 ELO on 91 of the site's games. ELO decay would be horrible for me. There's no way I'm playing that many games at reasonable time intervals (plus the many more games that will be added in the future). Furthermore, I feel like I'm the type of player these changes are trying to promote, since I enjoy playing a wide variety of the games on this site (with people of whatever ELO).

The logical way to encourage someone with a high ELO to play is to take it from them if they do not.

I think that positive reinforcement is much better than negative reinforcement.

My proposed solution would just be to shift the huge trophies that are currently there for top ELO (best, top 5, top 10, etc.) and have them linked instead to seasonal tournaments (somewhat like sourisdudesert suggests, but actually using the tournament system already built here). It seems to me that it's the rewards from these top trophies that are incentivizing people to fake their ELO.

In this system, ELO would not be attached to gaining points anymore (besides getting Good, Expert, and Master achievements, which are very small in comparison). The top-scoring trophies would be part of the tournament system and would encourage people to play in the tournaments (and play games outside of tournaments to train, or just for fun). And the tournaments seem really good for stopping cheating since you're randomly paired with opponents.

Also, tournaments are great for promoting more obscure games. I already feel like tournaments are the main place I go to play them since it's a nice set time for the few players of, say, United Square, to get together.

tl;dr: Just don't have trophies correspond to ELO anymore. But keep ELO around because it's a handy reference value that shouldn't just drop occasionally.

[cosmosis]

I really hope that ELO won't be dropped seasonally, mainly because:

• It gives me a sense of progress. I like seeing that my ELO is slowly becoming higher. What sourisdudesert seems to be suggesting is just a continuous build-and-reset cycle that I think would become tedious. At some point I would just feel like Sisyphus, pushing the rock uphill only to have it roll down each season and having to do it again.
• I like knowing how strong my opponents are. With the new system I don't know if I'm playing an ELO 1800 or an ELO 2200 who has just been dropped down to 1800.
• The new system doesn't stop cheating! If it really is just a race to top ELO each season, all those cheaters can use their same techniques, they just have to do it on a shorter timescale, which might even lead to more cheating.

[cosmosis]

Here is how reassigning trophies to a tournament system and leaving ELO alone would align with sourisdudesert's objectives:

• We want to introduce regular "Seasons", to give you permanent achievements when you reach the top, in replacement of the current "volatile" trophy that you lose as soon as you lose your ranking.

• Yep. Every season, top players in the tournaments receive their trophies. No more top ELO trophies.

• We want to give "Seasons trophies" to players who really deserve it, which means you won't be able to win a trophy just by beating a beginner every month.

• Yep, you need to perform well throughout the season.

• We want a system than is less discouraging for weak players. At now they starts at 1500, and then decrease to 1200-1300 after few games which is quite discouraging.

• It would be awesome if the tournaments for a season were bracketed, so that people of similar ELO values faced off against each other. There could be a small, medium and large trophy (or something similar) for the beginner, good, and expert+ brackets for a game. This would hopefully make things much nicer for beginners without having to make some sort of ELO floor (and all the possible consequences of that), since they have a chance of winning games in their bracket.
• Bonus: This doesn't actually change BGA's ranking system or permanently affect anyone's ELO, so you don't have to fear fallout from players dissatisfied with a sudden rank drop. It also uses the tournament system, which is already partially built, and which inherently makes cheating much harder.

I'd be interested in feedback on these thoughts. Maybe they just sound good because it's late where I am.

If I'm understanding it correctly, I'm definitely against the new system sourisdudesert is proposing.

[latexrhombus]

I'm a slightly different kind of player than Glitter and Doom. My goal for a while now has been to get to a Good rating on each game on this site. I'm currently >1600 ELO on 91 of the site's games. ELO decay would be horrible for me. There's no way I'm playing that many games at reasonable time intervals (plus the many more games that will be added in the future). Furthermore, I feel like I'm the type of player these changes are trying to promote, since I enjoy playing a wide variety of the games on this site (with people of whatever ELO).

I am actually like you in this respect. But the points I was making is a decay should only affect the top players who just hold ratings. Not people who reach "good player" in games. Also, I do not think anyone is proposing a decay that continues after you are "unranked" in a game. So overall net loss even for those top guys would be low. Just enough to encourage participation IF AND ONLY IF they want to stay at the top. I think that is a fair price of all the prestige they gain by staying at the top.

My proposed solution would just be to shift the huge trophies that are currently there for top ELO (best, top 5, top 10, etc.) and have them linked instead to seasonal tournaments (somewhat like sourisdudesert suggests, but actually using the tournament system already built here). It seems to me that it's the rewards from these top trophies that are incentivizing people to fake their ELO.

[/quote]

I strongly support the seasonal tournaments idea as well. I think that is really the only idea so far that has made sense for running tournaments. I do think there was a good point previously though that a downside to that is if the prestige gains are as massive as the existing trophies that new players to the site will be at a big disadvantage.

[latexrhombus]

I really hope that ELO won't be dropped seasonally, mainly because:

• It gives me a sense of progress. I like seeing that my ELO is slowly becoming higher. What sourisdudesert seems to be suggesting is just a continuous build-and-reset cycle that I think would become tedious. At some point I would just feel like Sisyphus, pushing the rock uphill only to have it roll down each season and having to do it again.
• I like knowing how strong my opponents are. With the new system I don't know if I'm playing an ELO 1800 or an ELO 2200 who has just been dropped down to 1800.
• The new system doesn't stop cheating! If it really is just a race to top ELO each season, all those cheaters can use their same techniques, they just have to do it on a shorter timescale, which might even lead to more cheating.

These are strong points and that is why I have supported the tournament idea in previous posts. I was number 1 in Dominion when it was on this site for a while and since I actually played, I finished number 2 when it was taken off. I know how frustrating it can be to lose all your progress with ELO (to an even higher degree than resets).

I still think there has been overreaction to an ELO decay on the top few players in each game. Originally I proposed a 1% decay per week of inactivity and I do not think that is unreasonable, but maybe 1% of the difference between their current ELO and 1600 makes more sense (i.e. if a top player is a 2000 in a game, that would be a difference of 400 (2000-1600) so they would lose 4 (1% of 400) ELO points if they do not play in the designated period).

The one problem with the tournaments is they may not encourage basic site participation, since players would just be playing against others of similar ELO "on the side". It is really important that people are hosting and joining tables, especially for the less popular games to keep BGA fun and I think the potential (it doesn't automatically happen) of taking a few ELO points from those who care about ELO (the top ones) is enough to get it done.