Luck Plays No Part in Diplomacy

[Your Humble Narrator]

I can't believe how relevant that clip was, lol.

[Puscherbilbo]

Have you adressed the multiplayer component somewhere?
All your examples from computergames only apply to 1vs1 if i am not mistaken.
Imho this especially holds true for the first few moves but which are possibly the most important ones for the outcome.

He talks about this in the first point in the second article, the "free for all" problem. He includes a picture of super smash Bros with a caption that references this.

Nope. That is not what i meant at all. Pls take another look at my argument.

The use of poker as an example seems fairly illfitting here too. Without going into detail the effect described by OP plays only a fairly minuscule part in tournament poker and in very specific situations.
And while it may push variance in certain spots it will increase winrates in diplomacy and poker in the longterm since every other player has equal potential to profit from weaker players.

[RoganJosh]

In gunboat, I find that with Austria, the optimal opening is to expose Trieste to an Italian stab on the first turn

No, optimal strategy in Diplomacy will always be of the form: Submit order set A 75% of the time, and submit order set B 25% of the time. In each specific game, you should pick at random. This is what introduces an element of luck to the game. (I'm not making this shit up: it happens regularly in high level 1v1 games that people even use pseudo-random number generators.)

Yes, you can still lose a game even if you played optimally the whole game. But in practice this never happens. Btw, you were very creative with your interpretation of my previous post.

1. The majority of Diplomacy players don't even know what a Nash equilibrium is. Even among those familiar with the concept, I doubt that more than a handful understand how to apply the idea correctly
[...]
3. If your rivals are not applying these principles of game theory, you are foolish for treating them as if they are.

This I completely agree with! While in theory there is luck involved, in practice there is essentially no luck involved. And if the purpose here is to help people develop their gunboat skills (Swordsman - your blog is excellent!), then the title of the lecture should definitely be "Luck Plays No Part in Diplomacy."

[Carl Tuckerson]

Semantics. If the optimal strategy is not to cover Trieste 75% of the time then there still exist games where you do not cover Trieste and Italy executes the branch of his optimal strategy where he takes Trieste first turn and you quickly die in a Russian-Turkish-Italian fire despite making optimal moves each turn. To the average person using the word "luck" they will look at this situation and say that it is "unlucky" that they got crushed that way, because they followed an optimal strategy without making a mistake and still died. The situation is still distinct from games in which the player who finds and executes the optimal move every turn will not lose, and if "luck" fails to describe it then it's not very far off. (In fact, if we specifically agree that the optimal play is to do X on N% of Spring 1901s and Y on M% of Spring 1901s, and your decision in any given individual game is chosen at random, I'd think that following such a strategy and losing for it would fit any reasonable definition of luck.)

I agree that's a rare occurrence, as there are usually steps you can take to reduce the likelihood of this outcome happening--I don't agree it's completely zero to occur.

And if swordsman is deliberately framing a nonzero occurrence as a zero occurrence for the sake of writing a lecture to improve, that's fine, but he shouldn't be representing others' words as being in opposition to that mindset, when it's very clear they were speaking to the truth of the matter (that the occurrence is nonzero). I didn't think his framing of the comments claiming "luck" exists in Diplomacy was fair at all. This despite being in as clear accord as I can possibly be to his overall point.

[mhsmith0]

I do agree with the notion of treating the game as if it's non-luck as part of a self-improvement regimen

That said...

This commentator, however, unironically uses the abbreviation RNG and the metaphor “coin flip” to describe how the players choose their moves and strategies.

*ahem*
I believe that I was unironically using the RNG abbreviation to describe whether you happen to rand an opponent who will use a friendly or unfriendly strategy against you, particularly in spring 1901 (and if they choose opening moves off of some kind of rand metric, then there's obviously luck involved in which such move they choose in this particular game).

Using the most recent gunboat tournament as an example, a number of people drew Russia and were faced with Turkey moving into Armenia AND England sending both fleets north. Conversely, I'm sure there were plenty of Russians who saw England not just occupy the channel but explicitly move into Wales, and probably a few of those ALSO got a Turkey who sent their fleet into Con instead of BLA, which leads to an extremely strong opening position for Russia. I struggle to see how this is anything other than luck on the part of the Russian player.

I'd also note that there are a pretty reasonable number of situations where it really *DOES* boil down to a coin flip. Probably the most obvious one would be...

Say you're Turkey sitting at 17, you're blocked off from Mar/Iberia/MAO/STP, and the only way to 18 lies through Munich or Berlin. You've fully occupied Tyr/Boh/Sil/Pru (and let's say you snuck an army into Pie for good measure). Your opponent(s), meanwhile, have armies in Munich, Berlin, Kiel, and Ruhr. They also have a navy in Denmark about to go to Baltic Sea. You therefore have ONE chance to seize your 18th center before a full stalemate line will shut down your solo attempt.

You can either full attack Munich (Pru-Ber to cut; Sil-Mun supported by Tyr/Boh) or full attack Berlin (Sil-Ber supported by Pru; and then Tyr-Mun supported by Boh to cut support if nothing else)
Your opponents, not being morons, will enter the following orders: Ruhr-support Mun H; Mun support Ber H; Ber support Mun H. HOWEVER, they can only GUARANTEE a defense of EITHER Munich or Berlin, depending on which one Kiel support holds.
There's nothing particularly fundamental about attacking or defending one vs the other. They're equally vulnerable to assault, and for the sake of the argument let's presume it's the fall so holding it doesn't matter.
It's luck. Maybe you can squint and find a meta that people defend Munich 55% of the time and therefore you should attack Berlin; or that people attack Munich 55% of the time and therefore (as defender) you should defend Munich (these #s are blatantly made up btw). But essentially it's a coin flip, because no attack is inherently better than the other, ditto defense, and literally flipping a coin as to which you attack or which you defend is a perfectly valid approach here.

[Restitution]

Saying that Diplomacy has no luck is basically equivalent to saying that Rock, Paper, Scissors has no luck. Just depends on your definition of luck. It's not an insightful claim.

I like the majority of your articles, Swordman, but this 'aint one.

[Claesar]

You're all just wrong. If you want to cover Trieste, do it with Vienna.

[Squigs44]

I would like to point out that the original Avalon Hill statement (but not necessarily Swordsman's) was made about the original board game, and does not claim that in Gunboat luck plays no part. In full press I would agree with the rulebook when it says "At the start of the game, the players randomly decide which Great Power each will represent. This is the only element of chance in the game".

Also the whole choosing whether to go to Munich vs Berlin thing:
I think this is just a disagreement on the definition of luck. A Google search of luck states: "success or failure apparently brought by chance rather than through one's own actions". I think the fact that you have control over whether you go to Munich or Berlin could be used to argue that it is your own action that brings about your outcome. You could have chosen differently to get a different outcome. Now, you still are not guaranteed to win that encounter, but it isn't luck when you succeed or fail there, it's your decision that determines it. Again, it's fighting over definitions, which doesn't really help anyone

[Carl Tuckerson]

You're all just wrong. If you want to cover Trieste, do it with Vienna.

Right, presumably that's the 25% play to continue the earlier example. Or maybe it's a 10% play and a 15% play is some Hedgehog variant.
If the specific details are confusing the general point, then let me state the general point, which is that there is a % of gunboat games (however small) where all your neighbors attack you and stick with the attack until you are dead, and it largely doesn't matter if you made the best possible moves, you die. They're rare but they exist. (I don't believe they exist in a meaningful sense in press games; there, "assume no luck" is more than just a maxim for self-improvement, there's theoretically always a way to say the right words to get out of any given jam. But without that option in gunboat I do believe these games exist, however rare.)

I would like to point out that the original Avalon Hill statement (but not necessarily Swordsman's) was made about the original board game, and does not claim that in Gunboat luck plays no part. In full press I would agree with the rulebook when it says "At the start of the game, the players randomly decide which Great Power each will represent. This is the only element of chance in the game".

Also the whole choosing whether to go to Munich vs Berlin thing:
I think this is just a disagreement on the definition of luck. A Google search of luck states: "success or failure apparently brought by chance rather than through one's own actions". I think the fact that you have control over whether you go to Munich or Berlin could be used to argue that it is your own action that brings about your outcome. You could have chosen differently to get a different outcome. Now, you still are not guaranteed to win that encounter, but it isn't luck when you succeed or fail there, it's your decision that determines it. Again, it's fighting over definitions, which doesn't really help anyone

I also agree with this, but I can't help but point out that if that's the reason for the miscommunication, lifting quotes out of context and framing them in a certain light to make a rhetorical point (as I think swordsman did here, though not intending to do so) only adds to the difficulty and doesn't resolve it.

Hence why multiple people quoted in the post have essentially said "we agree with your broader point but think you're misrepresenting what we were saying."

[swordsman3003]

I am intrigued by the claim that failure of optimal move choices, made blind and without the benefit of hindsight (ie the gunboat openings), constitutes “luck.” This is the most thought-provoking critique of my analysis. I never directly discussed this phenomenon in my essays.

(1)

I do think my semantic argument responds. The choices of other players are not “random,” they are just unknown to you. If we admit that your uncertainty of other players choices constitutes luck, it leads to the absurd conclusion that Starcraft, Overwatch, League of Legends, Street Fighter, etc. are primarily about “luck.”

In Warcraft 3, getting a rare early item drop is “lucky.” Countering an opponent’s starting strategy with a blind guess is badass.

(2)

I also want to provoke you with further questioning: why is it luck when you opt to follow your favorite Austrian gunboat opening and it goes poorly for you? Why is it luck when you get attacked by your neighbors as Russia? What I’m driving at is: you are choosing examples where things go badly for you and you’re calling that luck. I stand by what I said in my Solo Win Tip essay — people have an irrational psychological bias to blame their problems on external phenomena, but nevertheless take personal credit for their successes. Unconsciously, you are drawn to the word “luck” when you feel bad about losing.

My point here doesn’t disprove what you’re saying as a matter of logic, but this psychological phenomenon is something that I considered and wrote about in my essays. Did you consider this when choosing examples? Did you notice yourself feeling drawn to examples of “unlucky” disasters?

(3)

Players do not universally agree on the “optimal” gunboat openings (I personally use only one opening for every power, and someone always rips my head off when I advise others to do the same)... so simply assuming away the possibility that you are playing sub-optimally is a huge red flag (for me). You have made a personal determination that a certain gunboat opening is “worth the risk” and sometimes that risk does not always pay off the way you hope. But your subjective belief that the Balkan gambit is the best gunboat opening for Austria does not create an objective fact that about the nature of Diplomacy.

I can’t help but feel that you are trying to discharge yourself of your responsibility for the results of your risk-reward assessment. You don’t think you should be blamed or criticized when the Balkan gambit doesn’t work out for you, so you call it “luck.” You have assumed away the possibility that you made a mistake...

...that is a grave mistake.

There was a entire generation of players who believed that the hedgehog, not the Balkan gambit, was the right opening for Austria — and there are still some around today! The metagame changes over time, and if Italian players abuse Austrians more often in the way you describe, the Balkan gambit will lose its luster. If you stubbornly insisted on using the Balkan Gambit in a metagame where Italians commonly make the opening that kills you immediately, that would make you a fool (and your rankings would take a hit). At some point, you would be making yourself “unlucky.”

Every time you get your ass handed to you by the Italian in the way you describe, you should consider: do I need to stop doing this opening? Maybe I wasn’t “unlucky” — maybe I have the wrong idea about how to play Austria.

In my view, so long as it is possible that you chose the wrong moves, “luck” is the wrong word to describe this challenge.

——————————

The rest of “counterpoints” I’ve read are, in my estimation, conclusory. snore

Engage with me! Why should I dignify a post in this thread with a reply, when my response would be to copy-paste what I already wrote? Is there anyone willing to write a take-down that respond to my essay, instead of restating views I already acknowledged and attacked in my essays?

Saying that Diplomacy has no luck is basically equivalent to saying that Rock, Paper, Scissors has no luck. Just depends on your definition of luck. It's not an insightful claim.

Respectfully, I disagree completely and I explained why I disagree in my posts. There is a difference between RPS and flipping coins, and that difference is called “luck.” If you fail to perceive or acknowledge a difference between the two things, then we have have a meaningful disagreement.

Further, I think that what someone views as “luck” offers immense insight into that person’s mind. The meanings of words are not just semantic issues. They are ontological questions. How a person views fairness, freedom, justice, love, and yes LUCK tells you a great deal about that person’s core beliefs and personality.

[Carl Tuckerson]

Don't have time for a full response for now, but I assure you I apply the concept both ways. I am working on an Austrian AAR right now where I felt forced to let Russia into Galicia in A01 in order to counter an Alpine opening from Italy that I thought was going to come back on me (and did), while Turkey opened F Ank -> Con -> AEG.
I believe I "got lucky" that Russia decided to attack Turkey and ally with me instead of killing me with Italy and Turkey. I would even say I "got lucky" that moving Vienna to Trieste instead of Galicia paid off, since if Italy hadn't attacked me, I'd have given Galicia away for free, and possibly died to an R/T for it while Italy attacked France.
I didn't randomly choose the move I made; I had an instinctive feeling that Italy was coming back on me, and I knew Russia had a reasonable incentive not to attack me, because he took and held the Black Sea. I can point to multiple key decision points (one example: building a fleet in Trieste in the winter) where I pushed the odds that Russia would attack Turkey instead of me more into my favor.
But I ultimately had no agency over Russia's decision, and Russia's decision was more important than my tactical decisions, because if Russia had attacked me I was stone cold dead. Because Russia's decision outweighed my own, I believe it's fair to say I got lucky to get out of the early game, even if I also made several very strong tactical plays that, had I not made them, would have also ensured I didn't get out of the early game.

Maybe that helps to explain the semantic difference over how we define luck?

I'll give a more thorough response later today if I can.

[Restitution]

Obviously there is a difference between RPS and flipping a coin. But most people would say that RPS involves a large degree of luck.

When people say "Diplomacy involves a large degree of luck", they are saying it in the same sense that they mean "RPS involves a large degree of luck".

Two players playing RPS optimally will each win 50% of the time. That's not to say that there's no skill involved (like a coin flip), just that even in optimal play, there are going to be random outcomes of the game.

You can't axiomatically adopt a definition of a word, and then claim that other people are "wrong" when they axiomatically adopt a different definition of a word. You are not meaningfully disagreeing with people in a way that is more meaningful than "I like my definition of a word more than yours".

Accusing other people of playing semantic games when they use a different meaning of a word is, itself, a semantic game. Language only works because people charitably interpret one another.

Like, great, dude. Well done.

[Restitution]

Like, are we allowed to say that there is some kind of element of randomness in RPS? What is a term that we are allowed to use under your language schema to indicate this fact? Or is it just normitively incorrect to point this out?

You're just erasing the linguistic capability to say things you don't like, rather than actually responding to a statement on its own terms.

[mhsmith0]

@swordsman: I think I'd probably try and clarify my headspace a bit more...

In PRESS diplomacy, you have plenty of control and influence over what the other players on the board are doing, and as a consequence, if you find yourself ganged up on early (England getting Sealioned with Russia in STP; France getting E/G/I'd; Austria getting R/I/T all up in his face; Turkey getting the ole A/I/R wipeout; etc), it's your own damn fault (and I say this as someone who finds it convenient to try and arrange such pile-ons in early game; sending Turkey home in 1903 as Italy was DAMN satisfying, etc)

In gunboat, you have no such control. The Austrian example is easy enough; if Italy attacks you, you're toast unless R/T are also fighting to the death (quite rare). In that context, Balkan Gambit makes plenty of sense. And say you rand Austria 10 times in gunboat in a reasonably short span of time. In three of those, Italy attacks you and you wipe out really fast. In the other seven, Italy attacks France or Turkey and you don't wipe out fast. Over the long haul, your results suggest that your opening was in fact the correct one (especially compared to alternatives that other Austrias experience over a lengthy span of time).

In the macro sense, you can say that over the span of 10 games, you overall succeeded as Austria due to the quality of your play, and because the opening you selected was a good one. But in the micro sense, was it somehow your fault that three times Italy attacked you and you bowed out fast? I think that it isn't.

And I think there are plenty of other relevant examples as well. As England, your game gets easier when F/G bounce in Burgundy. As Germany, your game gets easier when E/F bounce in the Channel (and a lot harder when both vacate it). As Russia, your game gets easier when Italy attacks Austria early, and when England opens south instead of north. etc. In press diplomacy, you have lots of control over these openings, but in gunboat you just don't. And I think it's reasonable enough to write off certain poor results as mainly luck, if your strategy and tactics were good enough but you happened to eat a bunch of unfavorable openings, and despite your best efforts you weren't able to recover.

PS maybe another really good example: if E/F vacate the channel, Germany *SHOULD* let Russia into Sweden most of the time (presume nothing in the south arguing against it as well for simplicity). If you play Russia and try for Sweden 10 times in that spot, and Germany bounces you twice, was that a bad decision on your end those two times, or was that just lolGermany and you just have to live with the potential for that outcome because it's the best decision on your end, but there's just no guarantees?

[RoganJosh]

I think that where several people here get it wrong is that, from a game theory point of view there, is no such thing as "the optimal move" or "best possible moves." There will be a number of moves which are reasonable. Your task, when analyzing your position, is not to figure out which possible move set is "optimal," your task is to figure out how strong the possible move sets are relative each other. Once you've done that, the optimal way to choose is actually to make a random choice weighted by strengths.

It is of course not only strategic position on the board which influences the strength you attribute to different move sets. The current meta game, knowledge of opponents, previous moves. It is all part of the evaluation. In many cases it is possible reduce to one or maybe two reasonable choices. "Guessing is skill."

But, sometimes, you will have to choose between a few options (and, if you are Austria, all options might be weak). Then, the optimal strategy is to make a random choice.

You can, of course, use some different method to choose than by random. In the long run, that will make you predictable. Actually, the fact that people don't make these choices by random, but follow personal preferences, is the reason to why we have a meta game at all.

[Puscherbilbo]

In chess there are two different ways to evaluate the quality of a move.
Either by a mathematical system basically assigning a pawn the value of 1 and evaluating each position accordingly.
Or by an estimation of your actual winrate vs yourself in a high number of games. (This one to my knowledge introduced by AlphaZero).

There is no such thing in diplomacy. So how to actually measure the valuedifference between those several options?

Unless you assume a certain option is always best (100%), as OP seems inclined to do, you will run into trouble using randomized decisions.

[RoganJosh]

Are you comparing how a human plays gunboat diplomacy with how a computer plays chess? I think it's more relevant to compare with how a human plays chess.

[Puscherbilbo]

no. it is about evaluationsystems of a certain position.
In chess humans are just incapable to reach that lvl a computer does.
But computerevaluations are generally accepted as baseline even though in human vs human games suboptimal solutions may render better results.

There is no such baseline here.

[RoganJosh]

AlphaZero's algorithm will work here too.

[swordsman3003]

Like, are we allowed to say that there is some kind of element of randomness in RPS?

No. Read on:

In the game Rock, Paper, Scissors, two opponents randomly toss out hand gestures, and each one wins, loses or draws with equal probability. It's supposed to be a game of pure luck, not skill — and indeed, if humans were able to be perfectly random, no one could gain an upper hand over anyone else.

There's one problem with that reasoning: Humans are terrible at being random.

Our pathetic attempts to appear uncalculating are, in fact, highly predictable. A couple of recent studies have provided insights into the patterns by which people tend to play Rock, Paper, Scissors (and why). Abide by them, and you'll be riding shotgun and eating the bigger half of the cookie for the rest of your life.

According to Graham Walker, veteran player and five time organizer of the World Rock, Paper Scissors Championships, there are two paths to victory in RPS: Eliminating one of your opponent's options — for example, influencing her not to play Paper — and forcing her to make a predictable move. In both cases, Walker wrote on the website of the World RPS Society, "the key is that it has to be done without them realizing that you are manipulating them."

What is a term that we are allowed to use under your language schema to indicate this fact? Or is it just normitively incorrect to point this out?

It is factually incorrect to describe RPS as "random." It's not a question of word choice. You are mistaken.

Two players playing RPS optimally will each win 50% of the time. That's not to say that there's no skill involved (like a coin flip), just that even in optimal play, there are going to be random outcomes of the game.

If the players played RPS optimally, the outcome would be random. But they don't, so it isn't.

You're just erasing the linguistic capability to say things you don't like, rather than actually responding to a statement on its own terms.

False. I am directly responding to your statements, and I am saying that your statements are not true.

Suppose I wrote an article that claims that a Pterosaur is not a Dinosaur. To prove a claim like this, the writer must 1) Define Dinosaur 2) Demonstrate that Pterosaur does not meet this definition.

And then you attempt to refute my article by saying: "well obviously Pterosaurs are dinosaurs; dinosaurs are huge extinct lizards are a Pterosaur IS one. Everyone calls them dinosaurs! I always had lots of dinosaur toys, and they included pterosaur toys."

But most people would say that RPS involves a large degree of luck.

When people say "Diplomacy involves a large degree of luck", they are saying it in the same sense that they mean "RPS involves a large degree of luck".

My only response is to tell you that you are wrong. You are mistaken as to what the word dinosaur means. The toy manufacturers are mistaken, your friends are mistaken, and your definition of dinosaur is incorrect.

Most people would call Pterosaurs "Dinosaurs," but they'd be wrong. They're wrong because they don't understand what a Pterosaur is, or what a dinosaur is, or both.

I am absolutely taking the position that the majority of people who disagree with me have a confused understanding of "luck" and/or don't know how Diplomacy really works. That's why I spent so much effort defining and defending my terms.

It might be easy to brush off my complaint as a case of paleo-pedantry, but word choice matters. "Dinosaur" is a word for a specific group of creatures united by shared characteristics and which had their own evolutionary history—it is not a catch-all term for anything reptilian and prehistoric. Calling a pterosaur a dinosaur is an error of the same order of magnitude as saying that our species is a marsupial[.]

You can't axiomatically adopt a definition of a word, and then claim that other people are "wrong" when they axiomatically adopt a different definition of a word. You are not meaningfully disagreeing with people in a way that is more meaningful than "I like my definition of a word more than yours".

You definitely can, and that is the main way philosophers, lawyers, and other intellectuals make an argument possible. Either you disagree about whether a word (that is, a concept) applies to a given situation, or you disagree about the meaning of the word (that is, the concept).

When I state "Luck Plays No Part in Diplomacy," you can refute my statement by attacking my definition of "luck" or my understanding of how Diplomacy works. Then, to defend my proposition, I have to defend my definition of "luck" and my understanding of how Diplomacy works.

So yeah, I'm saying your definition of "luck" (or now, "random") is crap. I thought really hard about it, researched the concept, kept notes, anticipated what you might say in response to my thinking, and wrote it all down in a series of essays. I think my definition is superior to yours, and I spelled out a case for why this is so. Like a normal argument.

Accusing other people of playing semantic games when they use a different meaning of a word is, itself, a semantic game. Language only works because people charitably interpret one another.

So yeah, maybe I text my brother a picture of Spongebob and Shrek shaking hands and he texts me back "lol so random" and people sometimes use the word "random" to mean anything unexpected. But that's just a misunderstanding or a misuse of the word random. In a casual conversation, it doesn't really matter. I wouldn't criticize my brother; I take his meaning.

But in the context of a philosophical or mathematical conversation, the word "random" has a fairly clear definition, and you can read more about it on Wikipedia.

RPS doesn't meet this technical definition, so your statements are false.

I don't think you and I have two different definitions of the word random. I think that the most likely scenario is that you are factually mistaken about how RPS actually works. I might also wager that you are using the word "RPS" as a metaphor for something you have not defined, and not a literal bout of RPS (with two people shouting and tossing hand gestures at each other) as I am taking you to be saying.

Of course, I'm sure now that I've criticized your writing in such detail, you will claim to have understood all along that RPS isn't "random" in the technical sense and you really meant something else. But when you do this, you'll be taking the position that "words mean, like, whatever I want them to mean man" so that you don't have to admit that you might be mistaken about something (either RPS or the meaning of the word random).

I am taking the position that words have precise meanings when used in the appropriate context. I don't consider myself to be playing a word game here. But if I am playing a word game, I believe I am winning.