HMS Dolores (2016)

I’m pretty sure I’ve spoken about the Prisoner’s Dilemma before, even if only briefly and obliquely. It’s about as close as any piece of relatively arcane game theory can be to ‘common knowledge’ – a lens through which we as a society can view the inherent duplicity of the people around us when co-ordination signals go askew. It’s been used on game shows, as major plot hooks in blockbuster movies (you might remember the Joker employing a variation of this in the Dark Knight), as a frame for understanding nuclear geopolitics, and even in giveaways at the UK Games Expo. It’s often reinterpreted, reskinned and rebalanced but the core is unmistakeable. It’s a kind of mathematical shibboleth that permits discussion to move beyond the particulars of a scenario and into more rarefied forms of specialised sociological analysis. ‘We understand the context of this and the motivations of the participants, so let’s talk about what their behaviour means’.

HMS Dolores is basically ‘Prisoner’s Dilemma: The game’

A tiny box of betrayals

If you’ve managed to get this far through your life without some pompous blowhard like myself huffing the Prisoner’s Dilemma into an unrelated conversation, then well done you! Unfortunately that’s going to change here and now. Don’t feel bad about it – you had a good run. For you though, here’s a quick overview of the basics of the scenario:

Imagine two prisoners, held in separate cells with no means of communication. The cops are willing to cut a deal because they don’t have the evidence for the solid conviction of both. Prisoners A and B are presented with the same offer at the same time. The deal is – ‘rat on your colleague and you’ll be spared’. If they both stay quiet, they’ll each serve two years in jail. If they both rat on the other, they’ll both serve five years in prison. If one unilaterally grasses up the other, the one that was silent will spend ten years in prison and the snitch will go free.

It’s not so much ‘snitches get stitches’ but ‘snitches get riches’. The power of the scenario lies in the fact that there’s no conversation or collaboration. Each prisoner makes this decision without knowing the decision the other will make. The decision isn’t made based on what’s right or sensible, but rather that which is optimal based on what the other person is likely to choose. If prisoner A thinks prisoner B will rat, it makes no sense at all to stay quiet – after all, you turn a ten-year sentence into a five-year sentence and punish the rat in the process. On the other hand if you believe that they will stay quiet, then… well, you could get away free and clear while they rot in prison. Except they know that too. While all these wheels are turning in one head, they’re also turning in the other. I mean, they know that it’s in both of their interests to stay quiet but can they trust the other to do that if they know the other is disincentivised towards solidarity?

Flotsam is a lot less robust than I recall

It’s a lovely scenario that is justly famous for how interesting it gets. However, it’s also a very shallow scenario because it is heavily weighted towards short-term rewards. This is a ‘one and done’ scenario – there’s no comeback for what you do, no reputational damage that you incur. You can focus on this one specific choice and never worry about what it means for the future. It’s important to make the right decision once rather than make the right decision consistently.

A more interesting variation of this is the iterated prisoner’s dilemma where the scenario is played out hundreds and hundreds of times (usually with AI routines) to assess optimality of strategies. For example, what happens if you always stay silent versus always rat? What happens if you randomly choose one or the other? What happens if you trust your opponent until their first betrayal and then switch to a permanent state of betrayal yourself? What if you copy the last move of your opponent, or wait until an opponent develops a notable pattern before copying? There are wonderful academic papers on all of these strategies because it’s not just about how successful they are in theory, but how successful they become in the ecosystem of strategies. Always co-operate is a terrible strategy when paired up against always betray, but a very good strategy when paired against another that copies the actions of its opponent. In case you’re wondering, ‘tit for tat’ as its often known is the most generally robust strategy that yields optimal results in non-orchestrated scenarios. It’s the golden rule, adjusted for a more vengeful social context – ‘do unto others as they have done unto you’.

Okay, this is now a seven-hundred word article that has talked about board-games for approximately twenty of them. I don’t know if that marks this out as an interesting review worthy of your attention or a mounting train-crash that should be abandoned immediately. Let’s try to get the damn thing to stay on the rails – let me explain why I’ve been talking about this. I swear it’s all pertinent.

Within the game of HMS Dolores we’re playing the part of Cornish Wreckers in the 18^th century. We are taking advantage of the vagaries of the rough coastlines around Cornwall to plunder those unfortunate vessels that run aground on our native shores. There are lots of false legends around the practice of wrecking, some of it active governmental disinformation. These myths and stories are illuminating – they show both the economic importance and the cultural prohibitions against the activity. Some of these stories say that ships were deliberately lured into the false harbour of fake lighthouses, although that almost certainly wouldn’t actually work. Others say survivors of shipwrecks were killed to ease the task of plunder, but there is little trustworthy evidence that was really the case. What is certainly true though is that the laws of maritime salvage were sufficiently loose to permit a thriving industry around the practice of wrecking. In HMS Dolores we’re engaged in a task of dividing up some loot that has ‘washed up’ on our shores, and we do this through a series of prisoner’s dilemma style negotiations. It’s important to note that this isn’t an exact copy of the dilemma as we outlined it above, but an obvious variation that draws very heavily from a common well.

This is worth thirty points. Don’t question it, it just is.

There’s a twist though – it’s explained in the framing of the game as a ‘strange ritual’ but is more properly considered a ‘complete load of utter bollocks used to add a gameplay contrivance’. We’re looking to build up sets of goods, but we’re only ever going to be scored on the sets that have the greatest and least value. Our job isn’t to obtain the most loot, but rather to carefully cherry-pick the loot most likely to increase the value of our accumulated flotsam. If we have multiple sets of the same high or low value, they all count towards our booty. HMS Dolores isn’t a particularly thematic game, but while this system does ensure the game sets the context for interesting decisions it also serves to forcefully undermine what scant thematic elements are present. It also saves the game from being a one-note symphony while simultaneously ensuring it can never rise to genuine greatness or even really ‘goodness’.

In game theory terms this would be called the payout matrix

Negotiations are conducted in dyads – two players, looking at four goods in an offering. Two sets of goods are in front of each player, and they have to discuss what they’d like to do with the deal in front of them. When the discussion is over each player reveals their decision by extending a hand that reveals a symbol of intent. These might be ‘peace’, which means ‘I’ll take the two in front of me and you take the two in front of you’; ‘war’, which means ‘I’ll have everything, thanks’; or ‘first pick’, which means ‘I’d like to choose one of the things from the four and you can have some subset of what I’ve left’. The actual outcome of the negotiation and reveal depends on the exact way the actions have manifested. If I pick peace and you pick war, you get everything and I get nothing. If we both pick war, neither of us get anything, and so on.

Occasionally a special power will wash up in the offer. This is a spendable card gives you a chance to change some of the mechanics of the game at a time of your choosing. If you have the ‘broken lantern’ card, you can spend it to have the next deal arranged over face-down cards, the contents of which only you know. The ‘new wave’ card lets you discard the current offer and draw a new one. My favourite of these is the ‘foiled’ card which you can play once two players have extended their hands. You then say ‘No, we’re having a do-over’ and the negotiation has to be reconducted with the additional knowledge of how the results went last time. If someone was just betrayed and suddenly the negotiation has to be conducted again you just know the next outcome is going to be comedy gold.

A range of wrecker superpowers

The game continues until the special ‘dawn’ card is drawn from the deck and then scoring is handled and a winner is declared.

HMS Dolores is a nice idea, and it’s mostly well implemented.

Unfortunately, despite this it doesn’t actually work very well. This isn’t an issue of the theme- which is shallow. It’s not an issue of the the mechanics – which are entirely functional. It’s basically a misalignment between the game design itself and what makes the prisoner’s dilemma interesting. HMS Dolores is an iterated prisoner’s dilemma that manages to be less meaningfully engaging than it would have been if there was only a single decision to be made throughout the course of the game. I genuinely think that if HMS Dolores had all hinged on a single decision made after a random partition of goods it would have been twice as much fun.

Let’s return to the iterated prisoner’s dilemma again and see some of the reasons why it’s such an interesting piece of design.

1. The central conundrum is impactful. It’s important to get each individual answer right because there is significant (hypothetical) cost to getting it wrong. As a consequence, failure is a heavy penalty.
2. The decisions are consequential. People don’t forget what you did to them in the prisoner’s dilemma because of how impactful the results are.
3. The dilemma is reliable. It’s always the same value proposition that is equally valued by both participants. That creates a shared context of competition over scarce resources (in this case, time away from prison)
4. The past is influential. By that I don’t mean that what someone did in this round is what they’ll do in the last round. Rather, what they did in the last round is heavily influential on what you’ll do in this round.
5. Betrayal is projected. If you betrayed someone in the last round you will naturally expect that will come back to haunt you in a future round.
6. Dishonesty is perpetually reinforced. Once you have betrayed someone, you are incentivised to continually betray them because you worry about the retaliation.
7. Trust is all but impossible to rebuild without someone showing mercy. Remember, one of the key elements of the prisoner’s dilemma is that people can’t communicate. You can only rebuild trust with your actions, and those open you up to disadvantage.
8. Strategies have a chance to cohere because you iterate over them dozens, hundreds, or even thousands of times.

All of this comes together to make for really powerful, important and heartfelt contemplation when a decision is put in front of you. I mean, take a look at this:

Hard to watch, isn’t it? This is a scene from a television game-show called Golden Balls. The whole show is hard to watch because it’s terrible. This particular episode though is a whole new level of uncomfortable. Neither of the participants feel good about what happened here. Neither of them will ever feel good about what happened. Neither of them will ever forget it. The iterated prisoner’s dilemma is that, over and over and over again. What you don’t see in the clip above is all the lying and deceit and misleading that led up to the final revelation. All you see is the impact of a single decision taken in a moment of great uncertainty. The Prisoner’s Dilemma is individual human nature shorn of its pretenses. It shows not just the darkness in an individual’s soul, but how dark they imagine the souls of other people to be.

Incidentally, I’ve always thought my approach to this if I was ever on Golden Balls would be to say ‘Look, I’m going to steal. If you steal, we’ll both lose out. If you let me steal, I’ll give you half after the show has ended. But make no mistake about it, I’m going to steal. Your choices are ‘nothing’ or ‘maybe half of the offer if you think I’m a man of my word. The absolute best chance you have is to trust me because even if I lie the worst that can happen is that you’ll end up in the same place’

I’d then spend a quarter of my winnings having the other person killed before I had to pay out.

Anyway, these are the things that are necessary to give the decision in the iterated prisoner’s dilemma weight and gravitas. Unfortunately, none of these elements are well emphasized in the design of HMS Dolores.

1. The conundrum isn’t impactful. Often you don’t want anything that’s in front of you because of the way the scoring works. Sometimes betrayal would be more costly to you than it would be to an opponent because it can add an unwanted item that now makes your previously carefully curated lowest-value set worthless in the scoring.
2. The decisions aren’t consequential because you’re dithering over a few points, not years in prison or life-changing amounts of money.
3. The dilemma isn’t reliable because the worth of an offering depends on the sets you have collected, and as such you might see great value in an offer that an opponent sees as worthless.
4. The past isn’t particularly influential because the dilemma isn’t reliable – you can’t predicate your future actions on a shifting proposition of temporary worth. The sting of betrayal is a lot less when the cost to the betrayed was negligible.
5. Betrayal isn’t particularly projected because while you regularly make deals with the same people they’re making deals with others too. Your relationship in the dyad of your interaction is largely lost in the greater sea of how everyone interacts with everyone else.
6. Dishonesty isn’t perpetually reinforced because the value of the scoring context means that you can easily explain why someone should trust you in a way that can fully express in optimal mathematical logic.
7. Trust can easily be rebuilt because this is a game of communication and negotiation, and as such you get to talk people into a course of action. You’re not merely a product of your previous decisions, you’re a product of your previous decisions as expressed through a haze of self-justification.
8. You might take part in half a dozen negotiations during the game, meaning that even if none of the above were true you still wouldn’t have a chance to cohere a strategy.

A game like this really needs betrayal to be coruscating around the table like a sonic boom of treachery. The problem is that in HMS Dolores there often isn’t any reason to betray and that means in most cases people are getting what they want out of every deal. Sure, some asshole (usually me) will cheat another player to great hilarity. The problem is that I do that because it’s in my nature to make a game as interesting as I can. I don’t really do it because it’s in my interests as a game player. HMS Dolores is very odd in that it critically needs people that will inject distrust into the proceedings, and players are only incentivised to do so for weirdly unselfish motives. As such, it all comes across as a befuddled mess where the central mechanic is completely at war with the incentives of its players.

That’s not to say it doesn’t have moments of fun, because it does. However, in my experience it needs someone prepared to act a bit like a gamesmaster – someone who sees their role to be maximising the fun for everyone. If you don’t have that you rely too much on players in a negotiation evaluating the random things put in front of them with approximately equal avarice. I feel like HMS Dolores is a game that would become richer over many dozens of sessions with the exact same group of people – I just don’t think that you’ll find anyone willing to put that effort into a game that doesn’t make any individual session particularly interesting. Some games thrive in an energetic meta-game of social context – consider our discussion of One Night Ultimate Werewolf as an example of that. Individual games of ONUW are fun in and of themselves though – they don’t rely on a complex political history, that’s just something from which they benefit. HMS Dolores feels like you’re just kicking the tyres of the endeavour until the proper game falls out. As a result it never really feels like you’re seeing it at its best.  I think the prisoner’s dilemma is something crying out for a good design to bring it out to the fore. I just don’t think HMS Dolores comes close to accomplishing that.