Being able to abandon games

[Wreckage]

A penalty for abandoning co-op games seems possible. As a close example, there is no penalty for quitting a game on training mode. Somehow, this was possible to implement.

I think we can somewhat agree, that once a player has about 500 ELO or higher, ELO has very little meaning. But, this is mostly because of the current ability to abandon without consequence.

Wouldn't it be more fun if ELO did mean something?

[bayareascott]

A penalty for abandoning co-op games seems possible. As a close example, there is no penalty for quitting a game on training mode. Somehow, this was possible to implement.

I think we can somewhat agree, that once a player has about 500 ELO or higher, ELO has very little meaning. But, this is mostly because of the current ability to abandon without consequence.

Wouldn't it be more fun if ELO did mean something?

So you think there is no difference between say a 500 ELO and a 900 ELO? Is there no difference between ELO earned from 2p or 5p? Or from Avalanche vs. not? There is a huge difference in nuanced understanding of the game.

It's unfortunate that ELO is able to be manipulated, but there is some representation. The problem is that because ELO is gained without regard for the type of game it comes from, people have some general expectations of ability from ELO. Then you find a Master (700+) level player who lacks some basic understandings because they mostly play 2p or playing Basic or Tricky and don't learn the nuances of MC and BP. It's fine if people want to learn and ask for assistance, but that doesn't happen a lot.

When I start a game with someone with 500+ ELO, I have some expectations of at least basic understanding of finesse/reverse/bluff concepts. And I sometimes find people who still lack even that level of understanding. If someone says they would like something explained to them, great! But more often, you find players who think they know it all already when they clearly don't. We end up in a collaborative game where people are basically playing different games. That's not fun for anyone, and I'd rather end a game like that.

[jgpaladin]

I'll add some additional perspective.

Clearly there's a problem with the ELO system with Hanabi, because there are WAY too many people with inflated ELOs. No one is happy about it.

How did it get this way? It's some combination of...
- Players who cheat by abandoning every game where they won't get a 30.
- Players who fluff their ELO via playing tons of 2p games
- Players who fluff their ELO via playing easier versions of Hanabi (basic, tricky, and/or no black powder.)

It's not clear to me which of the above is the biggest reason, but I can tell you that I busted my butt starting from zero, learning the game, playing more and more difficult versions, and in the past year or two playing 5p almost exclusively. And now I'm pretty highly rated with ELO, I start masters-only games, and I keep running into players that don't know some pretty basic concepts. And oh btw, some of the better players have been discussing this and tend to agree that the difference between a fairly-rated 700 and 1000 ELO player is about as big as the difference between a 400 and 700 player, even though ALL of the former are ELO masters. So the assertion that once you get to 500 they're all the same is comical.

I RT a ton of players (seriously, I wish I could see this list because it's insane) because I don't think they belong at the masters level in Hanabi and it's not fun playing with them. (This is a little unfortunate because I am sure they are players I would be perfectly happy playing games other than Hanabi with...)

I cannot make my games any more restrictive than I already do unless I made them invitation-only.

So I do get stuck in games far more frequently than I'd like. I don't like to abandon, and I don't do it because of "mistakes." But when it's clear that we're playing a convention-based team game and there are players at the table who are playing a different game than I am, yeah I'll consider it. FWIW I just checked and similar to Travis I used to abandon more, so I'm at around 20% overall but this calendar year I'm under 9% in just under 160 games. Still too many for my tastes and I can't go back and justify every one, but I'm far from a serial abandoner, and I've taken plenty of lumps when games didn't go my way. I have never once abandoned a game because the deck was too difficult or we were approaching the end and weren't going to get a 30.

Would getting rid of Abandons potentially help my situation some as everyone's ELO took more of a hit (myself as well as all the folks who don't belong?) Perhaps eventually the wheat would be separated from the chaff.

I'd really like for them to track separate ELO for 2p/3p and 4p/5p games. I have yet to find a competent master who plays mostly 2p, yet they get to join my games and unless I RT them all INDIVIDUALLY I'm out of luck. That's a lot of pain to have to absorb as I meet them and dole out the RT's one at a time.

[ExaltedAngel]

I think we can somewhat agree, that once a player has about 500 ELO or higher, ELO has very little meaning. But, this is mostly because of the current ability to abandon without consequence.

It has very little meaning in general, I found some 800 Elo players being much worse than some others 300 Elo players that were instead effectively using the finesse and understanding bluff. There is no thing like a "safe threshold", Elo is absolutely irrelevant as of the current state and it sucks.

[Travis Hall]

Nah, once I’ve filtered heavily for players who heavily boost their ELO through abandonment (and it helps that my refusal to abandon results in them red thumbing me too) ELO provides a reasonable measure of skill. It’s not precise, and it’s not perfect, but in most cases the difference in skill between 500 and 800 is huge.

There are exceptions, of course. I usually don’t red thumb serial abandoners if they accept a refusal without being obnoxious about it and also try to improve their skill. Just recently, though, I red thumbed someone who had gotten to 800, but it became obvious they hadn’t actually learned anything since making expert - like, not only didn’t respond to finesse, didn’t respond to direct play clues either. So I looked at their history, and they had abandoned 7 out of their last 10 games (and 1 was a training game). Going back 50 games, the abandonment rate wasn’t as ridiculous as that, but still stupidly high, and 47 out of 50 games were 2-player. Why a player like this thinks they can join a game with three masters and keep up, I have no idea. So yeah, it’s not an assurance.

But most masters who can join my games are significantly better than experts.

There’s even threshold effects. If somebody I’ve never played with before joins with an ELO of 503, I know they will probably misfire three times, we all lose 10 points, and they can’t join again because they’ve dropped below expert. I’m prepared for that. Losing points with these people is part of what it takes to improve the community. More likely the third time they make expert, they’ve got enough understanding not to mess it up entirely and maybe start creeping up instead of yo-yoing around the threshold.

And there’s a reason I start masters-only 5-player games and expert+ 4-player games. Experts are far less reliable in 5-player, which is complex far more often. Experts can usually handle 4-player, though, especially the ones who have made 600 or so.

[Jellby]

Have you compared failure (3 bombs) rates, instead of (or in addition to) abandonment rates? One may think that a low bomb rate is good, but too low a rate may indicate cheating (through abandonment, chatting, etc.).

[Travis Hall]

It has very little meaning in general, I found some 800 Elo players being much worse than some others 300 Elo players that were instead effectively using the finesse and understanding bluff. There is no thing like a "safe threshold", Elo is absolutely irrelevant as of the current state and it sucks.

ExaltedAngel, you’ve never made expert yourself, unless you’re running multiple accounts. Have you considered that an 800 ELO player who is willing to join in games with good and strong players is likely to be somebody with an inflated ELO? The experience of my heavily-curated games may well be quite different to what you can get into.

[ExaltedAngel]

Have you considered that an 800 ELO player who is willing to join in games with good and strong players is likely to be somebody with an inflated ELO?

I didn't, but I don't really think that is necessarily the case, many good players join tables without being so meticolous in the selection like you are.

Then of course "your" 800 Elo mates are better than the average 800 Elo players, but it's just a logical consequence of you having 5k games and going through a continue selection by red thumbing bad players, checking for inflated ratings, etc... This doesn't counter my statement, it just adds to it that with enough selection you might be able to have the rating matter more than it does by picking players at random. But being able to roughly tell the skill level of a player you know nothing about is the first and most important reason why ratings exist and this is totally impossibile in the current state of the system.

[Blacktango]

I agree, it is not easy to play with many players because their knowledge of the game is very different.

When you alternate between a table of master players and those who are “inexperienced experts”, you need to be very adaptable in the way you give clues.

I would say especially around expert, where the level is very heterogeneous. Sometimes you can do some advanced moves, and sometimes your teammates don’t know how to give a bluff.

[Travis Hall]

Then of course "your" 800 Elo mates are better than the average 800 Elo players, but it's just a logical consequence of you having 5k games and going through a continue selection by red thumbing bad players, checking for inflated ratings, etc...

But I don’t red thumb bad players. I specifically red thumb three classes of players: those who attempt to coerce abandonment, those who are obnoxious in their refusal to learn, and those who mistreat their partners.

Two of those categories will correlate with better play, but I’m certainly not simply restricting games to a select group of players I know are good. And that last category generally catches highly skilled players.

And especially in my 4-player expert+ games, I get a fair number of players who have just recently breached the 500 rating. But not only them; I get players I’ve not encountered before (or at least don’t remember) across the range of expert and the lower end of master. Not so many from the 900-mark up, simply because if they’ve taken the time to get that good, I’ve probably already encountered them.

And most of the time, I can be confident that a new player rated 750+ will follow ordinary finesses, bluffs, and reverses, usually won’t play on a save, and maybe will play on a number clue non-save touching chop. They probably know trash bluff, too. These things I can’t count on below 700, and below 550 I expect them to mess up.