Why intentionally bomb a game?

[ampfaff]

I don't understand why people intentionally bomb a game just because people don't play exactly to their standards. It seems to be almost exclusively related to when someone clues a 5 early, but maybe that's just a coincidence. I'm just wondering if there is something specific about cluing the 5s early that sets people off into a toddler tantrum. Not all of us have a graduate level degree in Hanabi and just want to actually play the game through to the end. I can't imagine that some of these people have ever even played a full game to completion with how easy it is to set them off.

[AndreRhineDavis]

According to the conventions that we play with on BGA (for games with more than 2 players), an early save is a play clue. So if you clue r5 when it is not on chop (or when the player has known playable or discardable cards), you're basically promising everyone "you guys have the cards to make this r5 playable". So of course people will start looking for the appropriate red cards. These people are not "intentionally bombing", they're trusting your clue.

If it turns out that your clue is a lie, and that r5 is not in fact playable, then it makes sense that people would be annoyed and would try to explain to you why what you did was wrong. Sure, some players might do this more aggressively than others, and I don't condone aggression or rudeness, but there are many players who would try to explain this nicely and teach you nicely. In any case, you yourself admit you're not a "Hanabi graduate", so maybe if all the graduates and masters are telling you you're doing something wrong, you should listen to them and try to understand what you're doing wrong, rather than insisting that they're all just throwing temper tantrums.

Hanabi on BGA is a matter of learning the rich and complicated conventions that we play with here, which for the most part are unwritten uncodified rules with no central authority, and are instead simply the constantly evolving collective consensus of the majority of master players. The only way to learn these conventions is to play many games with various masters while being humble and being willing to learn and be corrected. Over time and with enough experience, I'm sure you will come to understand and internalise the conventions. But in general, if people are behaving differently to what you expect, especially if they're doing so consistently, you should probably assume that there is something that you're missing.

In short, don't early save.

[Romain672]

Since that talk about 5 stalls, I will add what I wrote about early 5 saves I really like on my old guide ( https://docs.google.com/document/d/1Vzg ... it?tab=t.0# ). Like many conventions, the main point is to understand why the rule exist, and when to not follow it.


X] 5-stall in early game:
A 5-stall is a 5 clue in the early game. It allows the player giving the clue to avoid discarding while not using a clue on a play clue if there are any available to them. If there are clues which they could give to play cards it can indicate to other players that their chop cards are important or that by giving a clue the player might cause another player to discard who they do not want to discard.

Firstly, the basics: if someone gives a 5 clue off chop in mid game, it’s a finesse. You are supposed to assume the 5 is playable and the team should play all the missing cards as a finesse.

Now, near the start of the game, if someone gave a 5 clue off chop (which means definitely not a save!) in the first few clues, you should check if the color clue on that 5 was available. Doing a long finesse with a 5 is generally better with color (but not always).
Then, you need to ask yourself, is a finesse likely? Check how many cards you need to play to fulfill the finesse (and assume your saved cards are on that color).
If it’s 1, there is little doubt, go for the finesse.
If it’s 3 or 4, it’s unlikely. I suggest you do not play into it, and ignore that clue. Most of the time, it would be just a stall clue because the person doesn’t want to discard.

Warning: those are for the number of cards YOU should play. If someone else has r2-r3-r4-x, and you received a 5 clue, you only have to play 1 blind card to fulfill the finesse. So you should trust it.
(After r1 play, it’s way easier for this player to assume r2+r3+r4.)

5-stall in the beginning can be dangerous with players whom you are not familiar as there is always the risk they will assume they have all of the missing cards even if they have to believe the stars aligned and their hand is r1+r2+r3+r4, but there are cases where it can be really useful.
Whether or not you use it, you should be aware that some players use it frequently.

Example:
Replay: https://boardgamearena.com/archive/repl ... s=84207926 (turn #4)

Romain got m2.
It’s now Romain’s turn.
Conventionally Romain should discard.
But here, if he discards, the team will be soon at 8 clues, since most of the time Romain got no others playable.
But Romain can give a 5 clue to Corys. The most annoying read would be for Corys to assume something like b2+b3+b5+b4, which is so unlikely.
That 5 clue has the disadvantage of losing one clue if the next draw of Corys is another 5, but has the advantage to save Romain’s chop, which is better on average than a m4 or a g4.



I will add that some player would just bomb on that case, but in my opinion they either want the strong to feeeling to be able to said to another master 'you made a mistake'. Or they just put enormous value on synchronising the team, and are fine to assume very unlikely lines or scenarios to have a 'perfect game' to see, without considering any exception.
Either way, I think it's an excellent example to make you start thinking to multiple interpretations for every clue, and make you improve.

======

And I have checked your last bombed game when you wrote that post, and you should just ignore some people. When there is multiple people which made mistake, or people are criticising other people and those last people can't accept a criticisim they could be bombing (if it's the other way around, they should discuss about it after the game).

[AndreRhineDavis]

I disagree Romain.
I don't know of any masters who play with these "5-stalls", that a 5 clue in the early game is not a play clue. This seems very arbitrary and against the principle that conventions should be as consistent as possible. If you clue a card not on chop, it is a play clue. Hanabi is not a game of probability calculations, you should never choose an option over another option based on "hmmm, that's not very likely is it?". Instead, clues should lead to deterministic behaviour, given the knowledge that a player has.

[aesche]

@OP: Only speaking about 2P-games, but if you play lots and always RT everyone that bombs or that feels entitled for condescending, unsolicited advices, you‘ll eventually have a pretty pleasant and undisturbed experience.

That said, saving 5s too early really is rather easy to avoid. For 3p+ players, there are many other examples of conventions, that lots of folks that follow them don‘t understand themselves.

[Romain672]

I disagree Romain.
I don't know of any masters who play with these "5-stalls", that a 5 clue in the early game is not a play clue. This seems very arbitrary and against the principle that conventions should be as consistent as possible. If you clue a card not on chop, it is a play clue. Hanabi is not a game of probability calculations, you should never choose an option over another option based on "hmmm, that's not very likely is it?". Instead, clues should lead to deterministic behaviour, given the knowledge that a player has.

I'm Bayesian, which is explained there: https://en.wikipedia.org/wiki/Bayesian_inference It consist to give a probability to any event which could happen (like chance for me to eat before 9 pm), and correct the probability depending of any new information I could gain (like passing more and more time writing that forum post)

When you create or think about conventions you should definitively think about probabilities.
When you are playing, I personnally think you should, but I think player's opinions will vary a lot on that.

The 5-stall in early game are kinda the exception, but a good rule of thumb is when you need to do 2 blind play, then it's unlikely you got both cards. So you should try to find alternatives way to think about the clue: was it a clue to resynchronise the team? Does that player which did something weird before still didn't get something? Is it a new player and it's obvious it was a mistake (still I know some players would want to bomb in purpose to show to that player that wasn't fine, while others would be careful of that clue and try to correct it if they can)?
If you find no other intepretation, then you should trust it's a finesse.

If it's 1 blind play, it's likely, so you just play normally: conventions were created for that.

I will do a section about maths, but if you are not into probability, you will dismiss that section all together.
Those numbers are really important. When you get to your second blind play you divide by 30 the chance for that situation to arise (10*3players). When you add a third one, it's by 300.
So let's said usually when you answer to a clue without much thinking, you are confident at like 97% (ie one bomb per game)(that englobe everything: chance for that player to do a bad clue, chance for you to have missed something, chance for something unlikely is happening, and so on...) that you should blind play one card. Now you receive a clue where you need to blind play 2 cards, that probability goes to 52% (this was calculated with odds, if you know maths well, feel free to correct my numbers if they are far off). If you need to blind play 3 cards, that probability goes to 10%.
Those numbers are confusing. What I mean by those is that you should check a lot more about that situation (what i said earlier) to make that probability goes higher. If per example by passing 2 minutes of thinking, you goes from 97% to 99.5% for one blind play (which is an enormous leap), then it will be 87% for 2 blind play, and it goes to 40% for three blind plays, and to 6.22% for four blind play.
87% can be fine, you will bomb one time out of six.
But 40% is annoying. Even if you would be 99.5% confident about an usual clue, then that would lead to a bomb 6 times out of 10. That's really bad odds. So if you try to win, you either need to be even more confident about the situation, or you do an action which have 60% chance to bomb.
[note that you should most of the time do something. If you ignore the clue, not only others must deal with that clue, but must deal too with you not answering. That can be a lot worst][and note that I never advised to ignore the clue, I always advise to find alternatives. And that's really important]

In my opinion, not taking any probability into account will make follow conventions strongly, not taking any exception into account, having a strong base, but will (not be fun first ) lead to trusting astronomical low odds for the sake of following conventions.
If you see a situation arise 20 times, 19 of them leading to a bomb, but persist to thinking that you should follow conventions, that can be fair, that lead to force everyone to goes into your way, but at the same time, the 19 other times there could have been a reason for not following them, and you lose all of that value there.

When I was playing on bga, I was doing them really rarely (I need strong reason, like on my previous post), but for me, that's a really good way to start to think more deeply.

And since many intermediate player just do 5-stall without reason in early game, just doing them too in masters game look a good way to get some extra value (when that lead to force other to assume the right interpretation, or something extremely unlikely).

[Travis Hall]

Example:
Replay: https://boardgamearena.com/archive/repl ... s=84207926 (turn #4)

That example is so old that it predates the implementation of Black Powder on BGA. (I can tell from the graphics.)

Even if there were a significant number of masters who accepted the use of a 5-stall back then, now it is uncommon for masters to play without Black Powder. Black Powder changes the game significantly in this area, as an early-game 5-stall will often look like a play clue for 5k to the clue receiver. We can usually expect the clue receiver to misfire their 5 if a 5-stall is clued before 5k is played.

And this reshapes conventions, too. Romaine, you mentioned the danger of doing this with players you don’t know well. Well, really, the danger comes when you don’t know whether your partners expect 5-stalls. Now that Black Powder is in heavy use, and 5-stalls are more dangerous, the probability of one or more of your partners not being familiar with 5-stalls, or simply not believing they are or should be in use themselves if they are familiar with the concept, greatly increases. And this snowballs, too; the fewer players agree to use 5-stalls, the greater the number of misfires caused by 5-stalls, which in turn leads to even fewer players agreeing to use 5-stalls (because they see the bad results).

Also, you don’t necessarily need all the cards required to make the clued 5 playable in order for a response to the clue to be successful. Some players (I would judge a greater number of masters than those using 5-stalls) like to clue a non-promise 5-bluff by cluing 5 to a non-black 5, with the next player responding with a blind play that will turn out to be 5k. (I’m not a fan of this technique either, but I’m aware of its existence).

In short, I think Romaine underestimates the risks of using 5-stalls, and overestimates the frequency of their use in high-level games. I don’t advise trying to use 5-stalls in games with masters, or even experts, especially when using Black Powder.

[Travis Hall]

Also, we should take into account the probability that somebody else will misfire if we fail to respond to an apparent play clue. If you and the clue-giver like 5-stalls, but the player after you does not, it may still be better to misfire than to let the next player misfire.

(Of course, we also need to account for whether the next player will play on after the misfire and misfire again anyway. Even among masters, the understanding of how to recognise and recover from errors is not great.)

[ampfaff]

I should have mentioned that I'm not even talking about myself anymore, because I've long since learned not to early save, unless I *have* to clue for some reason but there's no good clues to give, but that's a rare and entirely different situation. And I understand that an early save can sometimes cause someone to misfire, which is frustrating, but some people will then obviously bomb on purpose. It's frustrating because they can see players' levels before they agree to play. If you are unwilling to tolerate any mistakes from other players, you shouldn't agree to play with players that aren't higher level. I allow all levels of player when I set up my tables so sometimes play with beginners. So then it's very annoying when a higher level player joins the table and bombs us when a beginner plays like a beginner.

[Travis Hall]

unless I *have* to clue for some reason but there's no good clues to give,

You basically never have to. Even if you have to clue, it’s better to give an empty clue than a bad clue. (And empty clues can convey a lot of useful information.)

So then it's very annoying when a higher level player joins the table and bombs us when a beginner plays like a beginner.

So, you expect a beginner to play like a beginner, but you don’t expect a master to play like a master?

Master rank here on BGA isn’t really an indication of a player’s ability to adjust their play to suit their partners. 700 ELO is really only a fairly basic degree of proficiency with the game. Most players who reach it do so just by memorising and following a list of techniques. From what I’ve seen, learning to adjust during play in response to a mismatch of expectations is something players only start to pick up well after they have reached master rank.