Suggestions to improve point allocation

[Vincent696]

A few things I've noticed on Hanabi more than on other games:

a) people tend to abandon games collectively much more than on other games, especially on 2P games
b) this creates asymmetry between honest players who accept to lose points and other players who don't accept to make a perfect score. Consequently, their points keep raising. That doesn't mean they're any good. I've seen experts or masters not knowing about elementary finesses or GDs ...
c) On top of this, the point allocation favors harder decks, harder being known as "avalanche" and "black powder". It does not take the distribution of the cards in the deck into account. On top of this, I believe that plain vanilla games with only 5 colors, no BP are actually decently hard because the deck is so thin that you can't afford sub-optimality in your choice

Some proposals I would love to see :
i) when people are in Hanabi games but their collective abandon rate is much higher than some threshold (either Hanabi's or cross game), they still lose point.
ii) Would it be possible to access BGA Hanabi's DB with stats about victory of players, their ranking and deck distribution etc? Would love to run some stats about it to assess a deck "toughness"

Cheers
Vince

[Blacktango]

Hi,

I completely agree with your statement.


About the 1), this can't be managed from Hanabi, because the abandonment is a generic feature provided by BGA. This means that the Hanabi's developer can't give negative ELO points if a team chooses to abandon, and may not even know if a team has abandoned to perform some action.

About the 2), from what I saw in BGA's game management, you usually only save the current state of the game in the database.
So at the end of the game, the deck is basically empty and you can't know anymore what it was at the beginning.
This is why it is a hard work feature to allow some undo action. Because you have to keep track in the database of what the state was before the action.

[Malo77]

Hi
Interesting points.

It is true scoring is a bit unforgiving.
And I find 2p games more prone to randomness as you have less options do clue/deal with it.

It would be very intersesting to have an api/access to data and the log.
As you can replay games, the data must be kept somewhere I suppose ?

[Stroom]

5 colors without multicolor is actually very easy. You only lose points if you have discarded a 3/4 and the other one happens to be the last one in the deck.

What I would propose is that all collective abandons should count as a loss for every player. This way there will be much less quitters and the collective player skill will improve as players are forced to learn how to deal with tough situations.

[Wreckage]

As far as I know, there is no added elo for using black powder. There should be, because it certainly is a higher difficulty.

I would also say that flamboyants make it easier to get better scores, so elo should be adjusted for that as well.

There shouldn't be much argument about elo gained from normal 5-color games, because the elo gain is so low, that nobody is artificially raising their elo on that variant, or any variant other then avalanche.

[smallman]

I completely agree with Vincent and wreckage. ELO for hanabi is currently so meaningless, you basically need to see how many games they have played to have an idea of them being any good.

[ChiefPointThief]

As far as I know, there is no added elo for using black powder. There should be, because it certainly is a higher difficulty.

I would also say that flamboyants make it easier to get better scores, so elo should be adjusted for that as well.

There shouldn't be much argument about elo gained from normal 5-color games, because the elo gain is so low, that nobody is artificially raising their elo on that variant, or any variant other then avalanche.

I Agree 1000%. Some sensible players here. 2p multicolor allocates almost the most elo (only second to 5p). When you add flams it becomes much easier. You get to resurrect 2 cards that you threw away as well as multiple clues. If you throw away 2 of the same 2s or 3s your score is shot but with flams you can still get a perfect score. Basically any two decent players cannot fail in this format.

ELO for hanabi is currently so meaningless, you basically need to see how many games they have played to have an idea of them being any good.

Pretty much. A player doesn't have to ever play on the hardest settings (multicolor, black powder, no flams) and can still climb to the top of elo. Play enough games with easier settings and your score will eventually get amongst the highest.

To the point about deck distribution flams helps with that as well. When you start w/ 3 5s and then you may have a card that needs to be saved it can be difficult. But with flams you can get rid of a 5 and still get a perfect score. Now you have more room in your hand to play that 3 that needed to be saved and suddenly what would be a very difficult challenge w/o flams is rather simple.

They should find a way to rank 55cards as well. Another challenging setting. I suggest allowing the variant in which you are allowed to play out your hand once the deck is exhausted. That way even if one of the cards needed to get a perfect score is at the bottom of the deck you can still win.

[Stroom]

Elo based on different game settings... Why compare apples to oranges? It would be quite hard to find out what a balanced way to adjust elo based on game settings is. You can just deal with it and understand that people who play with game setting A will have a naturally higher elo than those who play with game setting B or C...

Based on my 25k 2p games, I would not say that black powder is "harder" per se. Just more randomness is involved but the game plays pretty much the same way. I prefer not using it because of the randomness aspect. You can more often end up in non-savable situations. Also, although flamboyant does make it easier to get a score of 30, it is with a tradeoff of having less clues and the randomness of which flamboyant effect you can use. If you play at a higher level, you can take that into account to play a more lenient way when it comes to conventions and invent better ways to play the game. Often a lot of players fall into the convention-trap where there is always one way to play and no creativity is allowed.

[Jellby]

To the point about deck distribution flams helps with that as well. When you start w/ 3 5s and then you may have a card that needs to be saved it can be difficult. But with flams you can get rid of a 5 and still get a perfect score. Now you have more room in your hand to play that 3 that needed to be saved and suddenly what would be a very difficult challenge w/o flams is rather simple.

Flams help, but they are not the free out-of-jail card that you are implying. On the one hand, they may be useless, I've lost count how many times we get the "play from discard" flam when there is nothing that can be played, or the "put back in deck" at the end of the game when there is no time to play whatever card is rescued. On the other hand, they're random, you never know if playing a 5 will give the token you desperately need or not,

[Stroom]

Flams help, but they are not the free out-of-jail card that you are implying. On the one hand, they may be useless, I've lost count how many times we get the "play from discard" flam when there is nothing that can be played, or the "put back in deck" at the end of the game when there is no time to play whatever card is rescued. On the other hand, they're random, you never know if playing a 5 will give the token you desperately need or not,

This is why you should not play 5s ASAP, rather you wait until the last moment when it is really necessary (you think your chop might need to be save or the deck is running out). And your partner should also understand that with flamboyants, playing a marked 5 is not guaranteed