hanabi-live
hanabi-live copied to clipboard
Hanabi-Live
remove efficiency and add spare clues
Plea to keep efficiency. For end-game efficiency is useless and spare clues could be nice, but as an overall metric for how the game is going and what kind of clues you could afford, efficiency gives more info than spare clues.
Furthermore, I think spare clues would be nice (instead of number finesses required) in the dropdown. But it’s often misleading to rely on for tricky end-games. First of all, tricky end-games are usually more about who can draw what. Secondly, you still need to count if you can actually use all the spare clues. I would say that count is just as much effort as counting what to do in the first place.
Recent game example. Spare clues would be 4 in this scenario. We need only 3 clues to stall. But still we are 1 clue short if Markus plays instead of discard.
as an overall metric for how the game is going and what kind of clues you could afford, efficiency gives more info than spare clues
clarify this statement because it seems wrong. do you know what spare clues are
Yes I do (remaining possible clues - cards not gotten). But if you know spare clues is 2. I would say that is less info to decide if you can afford a 1-for-1 (mid-game) than when you know (future required) efficiency is 0.85. (And for end-game it’s the other way around.)
I mean, that are quite literally clues to spare, but please enlighten me.
Or are you just being pedantic and saying I should also subtract clues to stall exactly 1 round (player count - 1)? If so: stalling exactly 1 round is pretty random, I don't see why you want to know that at the start of every game. Anyway, subtracting a constant doesn't change anything about what I wrote earlier. (For the example spare clues would be 1 in that case and still you are a clue short if you play.)
I think the spare clues=4 metric is helpful in that endgame. Spare clues assumes you save all the 4's for the final round, so you if you play both 4's you'll only have 2 clues, which is not enough, therefore you must not play both 4's.
Would it be satisfactory to show remaining possible clues and cards not gotten in the efficiency section and let players do the subtraction or division?
i dont think so, it is an important design goal that players should do no mental arithmetic
to clarify my statement - you could keep future eff and stick it in the tooltip
I definitely agree it's quite a useful metric, like I said, but imho not better than (future required) efficiency. During a game you usually want to know if you should discard or give a (inefficient) clue, for which I think efficiency provides better info.
A lot of games are lost on end-game, for which spare clues would be more useful than efficiency. But I tried to demonstrate it's not such an important metric that it should replace efficiency. If you can figure out if the assumptions hold, you basically already solved the problem I would say. (I think knowing the assumptions is usually less important for situations where you use efficiency.)
And I like it to be a single metric, also for remaining possible clues the same assumptions apply of course.
To make sure I understand correctly, this is currently implemented as "number finesses required"? If so, I suggest we can close this.
@vEnhance no. A big difference is that efficiency is shown always, while "number finesses required" is shown only when you hover mouse on some obscure number somewhere.
Oh I meant whether they were the same number, not the UI presentation.
I will say I aesthetically prefer having future eff shown and spare clues in tooltip rather than vice-versa; the former number feels more intuitive to me. I understand this is a point of debate though.
Spare clues = remaining possible clues - cards not gotten Number finesses required = cards not gotten - remaining possible clues
So no, but spare clues is more useful imo. Especially since number finesses required has no direct relation with finesses (also more convention agnostic). So I would replace that, but I wouldn't swap places with efficiency (see arguments above).
I aesthetically prefer having future eff shown
a single positive integer shown seems like the most intuitive solution and is the one that i currently prefer
Spare clues and future required efficiency are based on the same numbers. The first one is the absolute value, the second one relative. The absolute value only really matters at the end of the game, but says nothing about the game state. If you want to make a decision based on the game state (can I afford to spend an extra clue) you want to know the relative value, since the same absolute value can map to very good or bad relative game states. So if you only know the absolute value, that contains less relevant info for most of the game.
So sounds like we want to have both spare clues and future required efficiency, and the question is just which one should be visible always and which one is hidden within the tool-tip. I agree with jack's previous post, meaning I vote for future efficiency to be the visible one. Would be curious if anyone else wants to vote.
Also, let x := remaining possible clues - cards not gotten. The website could print either max(x,0) or just x itself for "spare clues". Former is more informative, latter is more intuitive, maybe. I vote former here.
I don't know, a negative number doesn't seem equivalent to 0 to me, so maybe don't show a number at all in that case. But a negative number seems perfectly fine to me; consider that for harder variants it will be negative for most of the time.
The first one is the absolute value, the second one relative. The absolute value only really matters at the end of the game, but says nothing about the game state.
i don't understand this formulation of "absolute" vs "relative". if spare clues are at 2, and then you give 1-for-1s for the rest of the game, then the game will finish with spare clues at 2. if spare clues are at 2, and then you give a 0-for-1, then spare clues will drop to 1.
thus, it seems dynamical and relative in nature, communicating the same information regardless of whether you are in the middle or at the end of the game.
Tried to write in more general terms the examples I gave before.
But let's say spare clues is 2 at the start of the game and future required efficiency is 0.94. Once there are 8 cards left to get and spare clues is still/again 2, future required efficiency will be 0.8.
so you are saying that information is lost because having spare clues at the end of the game is slightly more important than having spare clues in the middle of the game
i think the counterargument is that the dynamical nature could make it harder to plan for the future. the current paradigm is that players make binary strategy decisions based on current future eff and an arbitrary threshold. (i use somewhere in the range of 1.0 to 1.2, flopping my decision if it exceeds the threshold).
this is problematic because i cannot simply use the existing shown number as my threshold. what i actually have to do is play out the line and examine future eff at the end of the line. in the world of spare clues, i might still have to do this, but the big difference is that subsequent 0 for 1s in the line are just simple subtractions to the spare clues, which is easy enough to understand and compute mentally. namely, i dont have to actually rely on the interface + doing a double hypothetical for every single turn in which i want to make a binary decision.
overall, i accept that some information loss is happening in other words, players would have to mentally take into account that "spare clues are worth more in the end game". the consequence being that the same decision based on a spare clue threshold would be "incorrect" in the mid game and "correct" in the end game. but that might be acceptable if we can get the other benefits described in my previous paragraph.
The absolute value of clues available only matters when you want to spend them all. The relative value incorporates the game state and is more useful as a stand-alone measure of clue usage.
Making a nice profit is more important for your business than the exact amount of money in the bank, unless your bills are due.
Recent game example. Spare clues would be 4 in this scenario. We need only 3 clues to stall. But still we are 1 clue short if Markus plays instead of discard.
i find this a surprising example to use to argue for future eff, since it displays 0.00, which essentially communicates "you've won" and provides no value beyond that. when contrasted to spare clues, at least you are getting some useful information there, even if it can be potentially misleading in the way you described above
i find this a surprising example to use to argue for future eff, since it displays 0.00, which essentially communicates "you've won" and provides no value beyond that. when contrasted to spare clues, at least you are getting some useful information there, even if it can be potentially misleading in the way you described above
The intention was to show that I agree spare clues is valuable, but certainly not as valuable as efficiency. Spare clues matters in the end-game and there efficiency is of course useless. But from my experience with tricky end-games, play/draw order is often more important than clue usage.
By the way, I hope it was useful, but I don't feel like I added much to the discussion anymore. So I rather just try something if you're not convinced at this point, if there is no more input.
I don't suppose we have the UI space to have both? We could make the clue log a bit smaller and have an extra line...
If we did choose spare clues over efficiency, where would we put the variant-required-efficiency?
Just run into that subject. What you want to know is until which player you can delay until the end game start. Here is an old picture I shared on discord in depth: https://cdn.discordapp.com/attachments/776176933120507964/993775313869885492/startendgame.jpg You take the first player (which is displayed in the top right), you advance by one player for each number on that picture, and you get one player X. Then you move back one spot for each bomb.
Then, to know which player can start the end game, every time you want for the next player to start the end game, you need to keep one non-5 played during the end game.
This can be done automatiquely since it's globally known information.
Per example, in the picture shared in the first post, Markus started the game. It's 4players/5stuits, associated in my picture to number 2. So jeff is the player X. (you could decide to put a special icon on jeff, while there is no bomb) There is one bomb, so you move back one spot, you arrive on Helana. So you could decide to move the special icon to Helana, while there is exactly one bomb. Then, we want jeff to start the end game, that mean [since jeff is 1 spot after the holder of the icon] we want 1 non-5 to be played during the last turn: b4 is not great, and r4 is possible. So Markus should keep r4.
[edit: note that if the icon is done, you only need the bold part to think about as a player]
One of the big upside of it, is that if you see during mid game that r2+r3+r4+r5 are in order, you lack of effeciency, and you have r2, instead of calculing what mean that "spare clue: 7", you can just see if that would be perfect or not for the last turn of the game with that icon. (already happened twice to me)
And I believe having that icon would be better. But it's babysitting some players, one more step into playing instead of them. It would be a nice tool to have if it's done automatiquely and I would give a yes on having it.
