🌿ELO System Updates

Board Game Arena Official announcements
User avatar
ChiefPointThief
Posts: 745
Joined: 14 August 2020, 22:27

Re: 🌿ELO System Updates

Post by ChiefPointThief »

lordalx wrote: 13 December 2023, 10:41 → Team play
It is obviously unfair for games that play with teammates to have different ELO wins/losses within the same team. From now on, when playing games such as Belote, Decrypto, Tichu etc, you will not gain/lose ELO from the other players within your team.
Not all games that can be played as a team are currently under consideration, as some of them are not exclusively designed for team play. These require individual consideration and additional dedicated work.
Bang is always played as team game but the change wasn't applied to it. It is bad enough being partnered w/ ppl who barely have a concept of how the game is played. So much so that I was just threatened with a report for winning us the game :lol: but then I also have to lose pts to my teammate. Keeping the system the way it is just keeps players elo relatively the same which defeats the purpose of the system. You mentioned 3 games that the change does apply for but left us in the dark about the other 38 team games on this platform. There are several things about this update that weren't made clear and we are just having to find out these things on our own.
User avatar
orgle
Posts: 27
Joined: 26 August 2012, 04:14

Re: 🌿ELO System Updates

Post by orgle »

h_illes wrote: 17 December 2023, 18:09
Ceaseless wrote: 17 December 2023, 17:47 https://boardgamearena.com/forum/viewto ... 21#p162021 To give credit where credit is due, the new system for player repetition has caught a cheater. That's a point in the update's favor to be fair.
Decreasing the K factor would have achieved the same positive result, without the negatives.
Oh. He wasn't being sarcastic? I'm not familiar with the game -- how is it cheating to win against the same player 6 consecutive times?
User avatar
h_illes
Posts: 381
Joined: 14 March 2022, 20:08

Re: 🌿ELO System Updates

Post by h_illes »

orgle wrote: 17 December 2023, 19:52
h_illes wrote: 17 December 2023, 18:09
Ceaseless wrote: 17 December 2023, 17:47 https://boardgamearena.com/forum/viewto ... 21#p162021 To give credit where credit is due, the new system for player repetition has caught a cheater. That's a point in the update's favor to be fair.
Decreasing the K factor would have achieved the same positive result, without the negatives.
Oh. He wasn't being sarcastic? I'm not familiar with the game -- how is it cheating to win against the same player 6 consecutive times?
Uhm... he may have been sarcastic. In that game, the game is solved, and cheaters using the solver can be identified. I am not sure if that has any consequences on this site or not though, nor do I know about the specific case.
dinnyesbacsi
Posts: 2
Joined: 07 May 2017, 23:29

Re: 🌿ELO System Updates

Post by dinnyesbacsi »

I am glad to see that there are some thoughts against power levelling, however the rules seems a bit too simplyfied.

I have 500+ ELO in Nova Luna, there is a very small player pool in the game even in the peak hours. Since I am in the top 10, there is a handful of players having around 50-50 chance of winning. Means that I mostly play against players having 100-300 less Elo points , or with other words the expected length of winning streaks around 3-5 games, with a better than average streak (some luck), it might be easily 6-8 games in a row without a loss, however with most opponents it would take several months to play and without any intention to cheat.

long story short I would recomend add some other factors too... for example in case both players had X number of games (opponents) since their last game, do not decrease. Or just start decreasing after 5 wins or so.
User avatar
Meeplelowda
Posts: 3832
Joined: 14 March 2020, 10:31

Re: 🌿ELO System Updates

Post by Meeplelowda »

h_illes wrote: 17 December 2023, 20:25 Uhm... he may have been sarcastic. In that game, the game is solved, and cheaters using the solver can be identified. I am not sure if that has any consequences on this site or not though, nor do I know about the specific case.
It wasn't sarcasm. And the accused player wasn't using a solver. That's all I can say publicly.

But I agree with the comment that the result could probably be achieved with a K factor adjustment instead.
Play Now lobby 4ever! https://boardgamearena.com/lobby
User avatar
Wreckage
Posts: 396
Joined: 18 January 2017, 02:10

Re: 🌿ELO System Updates

Post by Wreckage »

Lowering the K factor each time you meet a previous opponent is the exact same thing as gaining less ELO each time you play them. I'm not sure how that is seen as a different thing.
User avatar
nik592
Posts: 1389
Joined: 16 October 2022, 13:54

Re: 🌿ELO System Updates

Post by nik592 »

orgle wrote: 16 December 2023, 03:09 Perhaps the competitive players were too vocal about their disappointment during their games, but the casual players failed first. They failed at warning players at their tables about their desired play style & known conventions in the Presentation field before the game started. It's easy enough to ward off competitive Hanabi players: tell them in advance that you're "casual," that you don't play with conventions (or list the handful you do use), or turn on Friendly Mode -- they'll never bother you again.
While I don't necessarily agree with the change to hide ELO for Hanabi (and/or other coop games), I think you're saying here that casual players should be avoiding high ranking players tables and/or announcing to the table that they are casual players. This seems like a strange idea to me - casual players are often unlikely to even know that there is a non-casual way to play Hanabi. Surely the experienced player should be announcing what they expect?

I owned Hanabi IRL before joining BGA and I'd played it... maybe 20-30 times. We'd never heard of conventions, we just played casually I guess you'd say. But sometime after I joined BGA, I decided to try out Hanabi on here. I think in my early games, I had the experience of a person asking to abandon as soon as we made an error - with no explanation in chat. I found that frustrating but didn't think much of it. Maybe the player just had to leave, though it did seem tied to the error we'd made - and as someone who had only ever played casually, and was used to accepting errors as a part of play, I didn't understand a mentality where 1 error = abandon game.

Another game I had included an (experienced) person with myself and another inexperienced person. The experienced player brought up (what I later learned was) a convention, but didn't really explain how to use it, just made some digs that we should be playing better than we were (to be fair, there was also a mild language barrier, which may have made explanations harder). And then after a while of this, also abandoned. After that experience, I went to read in the forums and discovered conventions, and ultimately decided I wouldn't be playing Hanabi on BGA anymore.

The thing that got me was that you can set your tables to only include players of a certain level OR you can choose not to join ones that include/are open to inexperienced players. If you want to play with a specific convention, that can be noted in the settings of the table (which it wasn't for any of the tables I joined - I don't join tables with options I don't understand), and as you say, you can mention others in the chat. Typically, a casual/uneducated player is likely to set their table with default settings and without restrictions and surely it is up to the experienced player to decide they don't want to join that table and/or setup their own tables the way they want to play.

And while I can play in Friendly mode (and actually might, now that I'm thinking about it), I typically don't consider playing in Friendly mode for any game (probably because in my mind it's still associated with being "Training mode" and once I understand the rules of a game, I generally don't play that way).
orgle wrote: 16 December 2023, 03:09 The proper solution to appease casual players who say they don't care about ELO is to tell them to only play Friendly games. Or announce themselves as "Casual" in the Presentation field. We all need to warn our table what sort of game we're expecting. High Ranked Masters used to do that simply through their host's ELO or the "Layered" tag.
I'm not totally clear on what you mean here either - what is the Presentation field? Is this the "Convention" field in the settings? If it is, there is no option for Casual, though I expect Standard convention is much the same. You say they use their host's ELO - so if the table host has an ELO of 750, this indicates that... players of lower score shouldn't join? How would an inexperienced player know this? And I'm also not clear what the "Layered" tag is - all of these are probably just reflections of my lack of knowledge within the Hanabi community, but that's pretty much my point - how am I to know to look for these things, or how to interpret them to avoid joining games with experienced players who are going to get angry at me when I inadvertently spoil their perfect games? Surely the easier way is to set the table to only include players of Master level and higher, nominate a convention, etc. If a player joins a table where a convention is nominated in the table settings and doesn't play with it (or doesn't know how to), I agree that players have a right to be annoyed by that, tell the player (politely) that they shouldn't have joined, and/or abandon the game. But otherwise, I don't understand how experienced players are ending up at tables with casual players - I understand the reverse, that can be just inexperience on the part of casual players. But experienced players know better, so why aren't the settings being used as designed? I am genuinely curious about this, wondering if there is some issue with the table settings that I'm not aware of.
User avatar
tcarlaw
Posts: 30
Joined: 04 February 2021, 08:04

Re: 🌿ELO System Updates

Post by tcarlaw »

lordalx wrote: 13 December 2023, 10:41

→ ELO Decay
In the past, some of the strongest players refrained from playing as they could lose ELO, attempting to prevent losing their hard obtained current rank. This lead to situations where players kept their rank without playing anymore. We clearly want to encourage you to play to keep your rank. This will also help other players to be able to compete against the best.
Doing so, a regular ELO decay is introduced to slowly drop the ELO from top ranked players that do not play anymore. The decay will be -10% at the end of the Arena Season for all Strong players and above that did not play any ranked game, but this could be adjusted later (both scope and frequency). The decay will not move player from a rank lower than Strong.

→ Red Thumbs for competitive play
One last thing that is going to be changed in a few weeks will be the impact of the Red Thumbs.
You can currently give a Red Thumb to any player with which you had a bad game experience. This prevents you from competing against them in ranked games or the Arena, but not in Tournaments.
This system can, and unfortunately is, used to prevent some players from competing against the best players. Considering that the Red Thumb system is being abused this way, on purpose or not, we will update the way Red Thumbs are impacting the matchmaking for the next season.
ELO decay seems counter-intuitive. Arena already resets each season which discourages high ELO players from participating in games with a large chance component as they must re-qualify against much lower ranked players. The major problem I have seen is once players achieve a high ranking in Arena they are penalized for continuing to play. Some sort of weekly decay factor for players in Arena who have not played a recent match would be more effective.

Red Thumbs are a problem but as your own system states for certain types of kingmaking or team play Red Thumbs were the recommended solution. I questioned this approach but overall I have found using Red Thumbs to avoid unpleasant players has greatly improved my experience. Considering the number of players and games this seems like a solution that scales well. I would be happy if I could create groups of friends and have statistics compiled the same way they are for Arena and be responsible for moderating my group taking the onus away from BGA.
User avatar
tcarlaw
Posts: 30
Joined: 04 February 2021, 08:04

Re: 🌿ELO System Updates

Post by tcarlaw »

zizzeus wrote: 17 December 2023, 00:43 Adding my agreement to the majority of the thread that is not happy with the ELO decay and power-levelling features brought in, for the reasons already mentioned many times.

Power-levelling will absolutely kill small (or emerging) arena games like Homesteaders, where there are only 4 regular players but could grow if spread.

As for ELO decay: Personally, I like to work a game until I hit a threshold, at which point I consider it 'learned' and move on, typically with my group of 10 RL friends who also play here. We then tackle a new game and repeat. I really don't want to play a game as a chore just to maintain a ranking, and I don't have enough time to play all of my historical games, work current games, try new games, and also work, see my family, sleep, etc...

Yes, I could not care about ELO more, you might say, but I do, and it's clear others do too, so why not just accept that some people care about it? ELO is also a decent enough commitment device to get one playing to win, and I use it to scout new opponents and see what skill they might have, and how much time I need to spend analyzing a game with them. I also look at other, similar games, to get a sense of their strategic ability. This lets me cut corners when making moves, which returns time back to real life. ELO decay would take that away from me, costing me more personal time.

And, as mentioned, it doesn't even solve the problems brought up, which to most players satisfaction are already solved by the 60-day cutoff on the leaderboard! So, you're stabbing at a ghost here. I really hope you revert these two changes, and though I don't play co-op games, it sounds like that change wasn't great either.

While you're at it, please just show the ELO in the arena. I can see it by clicking through to the person's profile if I don't recognize their name, but you're just adding extra time to my experience which I resent.
I think these are really good points. My frustration with ELO in games with high chance and where the other players skill levels have a larger effect than your own personal skill is that ELO is too variable. ELO was developed for chess where there is complete information and virtually no chance. The mechanism where wins against much higher players is rewarded so generously is the opposite condition necessary for games that involve chance and incomplete information. For Arena I think some sort of decay factor for inactive players would encourage people to play the entire season and not just park themselves at the top of rankings until they are displaced, or a system that allowed a certain amount of inflation so active players could expect to pass people at the top based on playing regularly.
User avatar
h_illes
Posts: 381
Joined: 14 March 2022, 20:08

Re: 🌿ELO System Updates

Post by h_illes »

Wreckage wrote: 18 December 2023, 01:38 Lowering the K factor each time you meet a previous opponent is the exact same thing as gaining less ELO each time you play them. I'm not sure how that is seen as a different thing.
Right now, if the player with the win streak wins, they get less than full points added, but their opponents get full points deducted. It's a negative-sum game instead of a zero-sum game, so the player base as a whole will lose ELO long-term. With the K factor decreased, it remains zero-sum.
Post Reply

Return to “Official announcements [en]”