Being able to abandon games

[Romain672]

Respectfully disagree. I think the best masters are able to see the entire board, predict what other players are going to clue/play (and when) and then react accordingly. If given a new clue for a convention they have not yet learned, they can analyze why it was given and what might be the expected response. Re: your example/conventions specifically, these are arbitrary rules made and memorized. Might be interesting to play 1-2 games like this, but not long-term, for me. They're not logic-based conventions you can build upon or learn from.

You are assuming that everything you play with came from logic, while it's not the case at all.
Even your basic 'discard oldest' is a convention. You can have some strong arguments for 'discard newest' if you build your conventions/playstyle around.
And many new players came with 'discard newest' with their group.
Another example, when someone turn 1 give a red clue on r3, and you saw no r1/r2 in finesse position, you are supposed (with bga) to assume you got r1 & r2, which is rare. So assuming another interpretation for the clue look fine (it gets lots of tempo, that's the big upside).

I disagree that this makes someone more skilled to make up conventions to memorize and play like a bot "if this, then that". It's the deductive reasoning that makes hanabi so challenging and fascinating to most 5p masters, and why many of us don't prefer 2p games where mostly only direct play/save clues are given.

I can first refer you to my last post of last page about objectives of players. But then for my objectives, I strongly believe this is totally wrong too.
By adding differents layer of complexity on conventions, you finish by having a really complex game, where if you know what a weird clue would mean and how everyone will react to it, and if you got another weird clue which is slightly more complicated but which (normally) does exactly the same thing, you can make that second possibility being different.
Another possibility is this first weird clue is impossible because (per example) someone have two copies of b3. So when you does that second weird clue, from the person who hold both b3, it's really weird since the first clue was possible. If he manage to deduce it (which is atrociously hard, but can be done more easily once you get promessed some plays/once you played b3), then you know you got a trash b3 in your hand, and can act accordingly with this information.
I can just assure you I know way more than nearly the totally of bga's master, and I still have difficulty to follow some of the moves of players better than me.

My mp are open for examples of what I'm talking here

[orchid]

You are assuming that everything you play with came from logic, while it's not the case at all.

You're extrapolating my statement to wreckage's conventions of [if I play a middle 1 it means this, vs if I play a chop 1 it means this], to speaking for how I feel about "everything" I play. Of course there are conventions to the site and the predominant bga convention does not mean it's the "right" convention. No one is putting forth that argument, only that other players can have a reasonable expectation to play with certain conventions on this site, and choose not to continue a game where another player doesn't care to follow it or discuss it. .live conventions are different than bga after a certain point but I can follow the logic behind it even if it is different from mine. But an arbitrarily assigned "if I hint an empty clue, it means play chop" doesn't really require as much skill to me. Can be useful, maybe, but less intuitive and if I wanted to play a memorization game, hanabi would not be it.

I don't believe I have ever played with you to need assurances on what you may or may not know, but I think part of the fun of hanabi is that we're all continually learning. I love discussing hanabi almost as much as I love playing it, and I'm happy to read your examples to learn more.

I 100% agree with you that more complex clues does not mean better players as I think the less ambiguity and the cleaner the clue, the more enjoyable the game for me personally. I also think it's much easier to clue (when you have "all" the information of that play) than to respond to a clue (where you're relying on deductive reasoning and trust), and it is why clue hogs annoy me too.

I recently disagreed with a player for continuing to hint layered on another player at the table who clearly didn't know this convention. As the game, for me, is also about reading your partners and their responses. And because I am part of the camp that *does* think elo has some indication into skill level (even if it is not always reliable. But it's what I have and so it's what I use, along with reviewing a player's last 10 games history), I adjust my play style for the partners I play with as well.

[Travis Hall]

I think a possible solution might be if we can implement table settings that restrict players with less than an x% of completed games join, the same way we restrict tables by reputation.

Reputation already is intended largely as a measure of how reliably players finish games. It can be affected by other factors too - mostly just moderation penalties for speaking badly in chat, but even that is relatively rare compared to unfinished games. The primary factors affecting reputation are game completion (-10 or -20 for each game quit or expelled from) and timeliness of play (+1 for each game completed if the player did not run out of time during the game).

Those who abandon often usually do so specifically to avoid these penalties. And a more sizeable part of the community let them.

If players were consistently refused abandonment for reasons other than outside-of-game emergencies, a recent incomplete rate of greater than 10% would keep reputation below 100%, and at best yo-yoing between 90% and 100%. (Quit at 100% drops reputation to 90%, then 9 completed games in a row takes reputation back to 99%. At 10% incomplete, the player simply can’t get back to 100%.) A recent incomplete rate of greater than 10% would also certainly drop their reputation below 90%, especially when the penalty becomes -20%.

I refuse abandonment not just to insist on ELO being adjusted to reflect play, but to insist on reputation being adjusted to reflect behaviour.

[Romain672]

You're extrapolating my statement to wreckage's conventions of [if I play a middle 1 it means this, vs if I play a chop 1 it means this], to speaking for how I feel about "everything" I play.

For that exemple, the thing is, when you don't play with this, for me, you just play order of 1s to tell if a finesse is going on or not. Per example, in multicolor variant, if Alice clue Cathy's 1s: 1-x-1-x, then if Bob clue red to Donald, we can got an ambiguity: maybe Cathy is holding somewhere r1 or m1, and maybe Donald have r2 or m2. If it's the case, then Donald could in some cases discard, which in rare cases could cost the game.
So, by not playing with that rule, you want instead to give information to Donald depending of the position of the 1 you play first.
I personnally believe that way is more instinctive, but for me it's totally a convention.
(note that even if you don't fully trust this convention as Donald, you can still do assumptions on what is your clued red depending of which 1 played Cathy, and so give better clues in average out of this assumption)

If instead you decide to assume nothing with the order of 1s, why do you not discard a random not clued card every turn? Of course I'm extrapoling here, but still curious why you think it's different.

Of course there are conventions to the site and the predominant bga convention does not mean it's the "right" convention. No one is putting forth that argument, only that other players can have a reasonable expectation to play with certain conventions on this site, and choose not to continue a game where another player doesn't care to follow it or discuss it. .live conventions are different than bga after a certain point but I can follow the logic behind it even if it is different from mine. But an arbitrarily assigned "if I hint an empty clue, it means play chop" doesn't really require as much skill to me. Can be useful, maybe, but less intuitive and if I wanted to play a memorization game, hanabi would not be it.

The thing is, once you got to more complicated conventions, you can find differents ways of logic (my discard oldest is already an example for it). So if you choose one path, everything will look logic and smoothly.
But yeah, I refer to my player who got 80% a play, and 20% not a play in my last page. If you play with strict conventions, this is impossible: you just have probability depending of what is present on your hand ; and not probability depending of what others think of what happenned before. And playing with that make the game feel really different.

don't believe I have ever played with you to need assurances on what you may or may not know, but I think part of the fun of hanabi is that we're all continually learning. I love discussing hanabi almost as much as I love playing it, and I'm happy to read your examples to learn more.

I personnally got the most fun by going in new paths and discovering things. Now I feel like I know most of the things which interrested me, and I believe what is missing will be similar in some way of what I already know. But bga's convention is like 10% of it.
I'm far worst at playing, since I overthink everything/got distracted easily.
[Hope I'm wrong ]

I 100% agree with you that more complex clues does not mean better players as I think the less ambiguity and the cleaner the clue, the more enjoyable the game for me personally. I also think it's much easier to clue (when you have "all" the information of that play) than to respond to a clue (where you're relying on deductive reasoning and trust), and it is why clue hogs annoy me too.

I will send you a video in pm, but having deeper conventions let you do the same thing but deeper, like I explained at my last post.

[Stroom]

Kind of off-topic already but... I'm one of those players who exclusively plays 2-player games. I do it because I hate the downtime in online games. I do not care much about 3+ player games.

I do not find 3+ player games either harder or easier. It's a completely different kind of game. You use a different kind of logic. I'd say that 3+ games are more robotic. More conventions that limit your possibilities. Also you have access to much more information (2x more in 4p games) so your own hand basically has less of an impact on the result of the game. It would basically be easier to create a bot to play a game for you when you teach it the basic conventions. Stricter rules, more dry memory game with more conventions.

2p is more fun to play for me. You have less room to store important cards so you have to manage both playable cards and savable cards. More thinking about statistics and probabilities. You manage your hand and the partner's hand. Your hand might lock up so you can't save it all. You have very limited room. Sometimes you actually try to save some cards that are not immediately usable but it would actually reduce the risk of getting a non-max score in the long run. Sometimes you intentionally let something go in order to get it back with flamboyant because doing that might mean that you have more clues for the end game. Lots of patterns can come up and you develop different small-scale tactics. And then you see that there are some long-term things you can do, too. You see the game more like a flow with long-lasting effects rather than playing turn-by-turn.

Sure, 3+ player games have that feeling too but the games are just so slow... I could play a 2p game in 5 minutes on average. 3+ player games often take 10. And as said before, larger risk of having a bad player who either does not play well or just ruins the game intentionally or just wants to quit when max isn't possible.

As elo is calculated by the average of all players and then the win/lose is based against some specific elo (Somewhere between 900 and 1000 I guess) then the more players you have, the better it would be for 1100+ elo players. Get a bunch of 700s to reduce the average but you still get the same as others for winning.

Either way... The only useful thing to stop the elo cheating would be to make everyone lose 10 points for abandoning. The games take 3-10 minutes so you can easily get it back by not quitting.

Winners never quit. Quitters never win.

[Blacktango]

I agree with the most part of what you said, Stroom

In 2 players mode, you just have to anticipate what your teammate will do, to decide what you do.
In 4 players, you have to guess what the player after you will do, with what they see, and knowing that they will wonder what the player after them will do, with what they see, knowing… and so on.
It's much more likely that we will incorrectly predict the move of the player after us. That's why we need to have more conventions to reduce this uncertainty.

This conventionality may look like a robot game
But as in all books of Asimov I have read, all the interest is in what happens when the robot reachs the limits of the rules!

[Wreckage]

At one point if you want to define what mean to master X, you would need first to know what is your objective:[/i]

This. Becoming a master player with known conventions is a skill. Being innovative with conventions through discussion is also a skill, but very different.

You are assuming that everything you play with came from logic, while it's not the case at all.

This is true. A convention mostly works because of agreement of what certain clues mean, through discussions between games.

"if I hint an empty clue, it means play chop" doesn't really require as much skill to me.

I mentioned challenge rules. In the simple challenges, like "you can only clue 1 color to your partner the entire game", I add the rule that no blank clues are allowed. About 1/2 the challenges blank clues were forbidden. But there is a skill to figuring out how green clue means this now, and green clue means something else later. It's interesting what people come up with.

I totally get your point though. You don't like "hat-guessing" type conventions. Although I think hat-guessing is somewhat impossible in 2-player games, I admit I was talking about similar mechanics...playing a card not knowing what it is. This is not totally different than you use in a bluff. But your point is valid. Playing with clues that generally share information is more fun. My real point was not to stop playing how you like, but to encourage getting a small team together to improve the mechanics through discussion and ideas between games. Playing with random new players every game doesn't allow any new idea to be implemented.

I believe a 5p master joining a 2p table can fare better than a 2p master joining a 5p table

I agree. The game complexity goes up with each added player.

Getting back on topic, I would love to see a penalty for abandoned games. Most of my abandons are legit reasons. But I have a few abandons because of mistakes. I think legit reasons are somewhat rare for most people. If an abandon penalty were ever implemented, I'd be for a total reset of everyone's ELO back to 0 as well. Making ELO mean something would be fun. I also think ELO gain should be the same at all player counts.

[Stroom]

I guess for some players even abandoning or not isn't the issue... I just encountered a player who boasted about his 1300+ elo claiming that he is more skilled. I then looked into his game history. For some reason he had a lot of sequential games with the same partner. And a little bit earlier another bunch of games with another player with a similar name... but that user has been deleted by now. Really weird.

Then I looked at one of the replays. The first move of the game was to blindly play a 1 card at the 3rd or 4th position. Some clues were given but when it seemed easier, random card plays without any meaningful clues given.

Some people are really desperate for that elo. Wow. Not sure if that can even be prevented other than actually analyzing each game.

[Blacktango]

I guess one solution would be to hide the ELO from the players.

For a few years, I played in a well known MobA. There were two ranking systems:
- a hidden ELO, which allowed you to meet opponents/teammates with the same level as yourself
- a public ranking for the competitive mode (like for the Arena on BGA)

Not sure that this would be pertinent for all the games on BGA, especially when you already know the physical game.

[Stroom]

That seems like a good idea indeed.