A Game Theory of Ethics

When people hear game theory it might conjure a range of ideas. Firstly- the most boring: ‘playing the game’ – any wannabe sociopath or elbows-out-and-down social ladder scrambler might take it as an invitation for competition, a battle of wills, wits, and social maneuvering. They often don’t consider that this ladder is a conditioned value system they’ve internalised rather than created and nurtured with reason, and it usually corresponds to money, reputation, and influence (or power). The latter of which is not entirely without merit, and money certainly makes it easier to enact one’s will on the world- but there are other means by which to do so. So why is influence of innate value? Maybe it isn’t, and maybe to relinquish all wants for control is the dissolving of the ego that the spiritual gurus teach we should aspire to. But gurus still speak, still teach, and in particular times of strife have been known to exercise their will for the betterment of society, which are all forms of influence, and so I will make an argument in the favour of it. But first of all, I will explain what I mean by game theory:

Game theory to my mind can be compartmentalised into two formal fields. We have Transactional Analysis as described by Eric Berne – the observation of the social micro-games that people play. Though touted as erring on the side of the ‘pseudo-scientific’ I think there is merit to his ideas- and if nothing else, it gets one thinking and observing the social minutiae that govern our interactions – particularly, the repetitive games that trap us, that we become familiar with, and that often don’t serve us. Simple examples of such approaches to life might be represented by the victim complex where someone wins if they lose- their payoff is that they get to complain, or the addict that surrounds themselves with enablers, heroes and antagonists to justify returning to their vice.

This is a phenomenological and wholly psycho-social game theory, and whilst it is useful to understand game theory through this lens, I will repeat the opinion of Berne in that this knowledge and awareness should be used to avoid playing games wherever possible- they’re often a waste of time and energy. ‘Stupid games, stupid prizes’ I think is the phrase. However, sometimes by the nature of certain games initiated by others and within which we are ensnared- choosing not to play is to concede or lose, and sometimes the losses can be costly.

The second type of game theory I hold in mind is that commonly associated with John Nash. A system of logical operations based on probabilistic outcomes in a game system of given rules. Decision logic for optimal return. It can be applied in the prisoner dilemma, or to optimise poker strategy. An interesting side point in the realms of this application of game theory is the idea of a Nash equilibrium, which is a play-style or approach that, if all players adopt- the system will balance over an infinite time. That is to say, if all poker players on a table approach with this optimal strategy and simply value-bet (bet chips in proportion to their probability of winning vs what they potentially stand to win) then the game will in theory stagnate. If you play just the stats, you have no greater chance than anyone else who does the same- and this is where the true nuanced beauty of poker begins to shine through: accounting for Nash equilibrium on the assumption that all other players will play close to the statistically optimal play-style, its becomes an art to know when to deviate from optimal strategy in order to give false impressions, posture a different presentation of assumed chance of winning (remember, bets are place in proportion to your chance of winning the hand), and so one might raise to 70% of the pot size to indicate they think they have a 70% chance of winning given the cards on the table, the possible hands, and the assumed unknowns of the opponent. A good bluff is indicative of a specific possible hand, given by a bet in specific proportion to the pot size). It becomes game theory Jazz – deviating from the standard rules of ‘good play’, dancing around the line of optimal strategy- deviating slightly and resolving; the better the player, the more daring they are to deviate. This is why experienced players often struggle against newer ones- the language of proportional betting isn’t understood – the learned musician riffs out a complex sequence of notes only to have the new guy loudly jam out some chords on a guitar and wail the lyrics to wonderwall – “I’m all in.” – it doesn’t do service to the abilities of the Jazz musician, and their best bet is to just join in. In these situations the experienced player should just maintain value-betting and play the pot odds – sticking to ‘optimal strategy’ against someone who doesn’t know play it will give a solid statistical advantage. Anyway, I deviate from my point now – what does any of this have to do with ethics?

Kant’s Categorical Imperative is a good foundation of moral behaviour, and maybe all that is needed – I don’t know it in all of it’s granularity, but I’m aware of the basic overarching concepts. To my knowledge, one of the principle motifs is ‘always tell the truth’ – to tell falsehoods is to obscure another person from successfully navigating the truths of reality and experience, the endeavour of which should be a shared aim, and to deny a person of truth is to do them a disservice. The second (but more fundamental) is a deontological approach- that is to say that intentions and outcomes do not matter in the realms of ethics- it is actions in themselves in the moment that hold the quality of being either ethical on non-ethical. For instance, the aforementioned: ‘actions involving speech should always be in truth’. I am inclined to disagree somewhat (if only for its practicality), but first I will highlight the apparent shortcoming of the always-truth imperative, and then respond to it.

The classic ‘always tell the truth’ counterexample is given by the hypothetical circumstance in which an axe murderer knocks on the door and asks if your mother is upstairs (presumably in order that they can kill her). The presented fallibility of the always-truth imperative is that the person being always truthful must affirm this to the murderer- leading to the harm of a loved one. However, I will use the magical language framework of a childhood book to demonstrate that this need not be the case (and this is often how real life maneuvering plays out). In the world of the fantasy book Eragon, the race of elves speak a language with which they cannot lie, as it is used to speak things into existence in their truth and essence – the basis of their magic; a will of truth upon the world. However, it is remarked at some point that they still have ways of obscuring the truth or perceived facts of the matter, by telling half-truths, or parallel truths that deliberately give wrong impressions. I think there is a quote that fits nicely: “Everything you say should be true, not everything that is true needs to be said”. And so I would suggest that Kant’s doorman simply reply “I’m not telling you Mr axe murderer, you’re a wrongun, now fuck off” – which isn’t untrue, if not a little direct. This could be approached with more delicacy and tact, which I’d recommend. Personally, I would take the opportunity to differentiate honesty from truth here, the distinction being that to my mind- honesty is forthcoming with all of the relevant truth that the other would want to know. Truth can omit useful details.

I’d also like to briefly state that always-truth might come at the cost of much of the symbolic richness that permeates poetry, metaphor, simile, and the semantic ambiguities that make language an art form. Though perhaps instead they illuminate the subjective and multifaceted nature of truth. On another post, I’ll go into more detail on semantic ambiguity and ‘ulterior’ or double-pronged ego interactions (so-called in Berne’s framework) – colloquially known as doublespeak; very interesting stuff.

I agree with most of what Kant says, but I would like to explore a philosophical framework of ethics that includes intention and outcome too.

I believe that honest intention plays a factor in the morality of an action (and perhaps what is moral and what is ethical are two different questions, but let’s leave that for another time). I do not think that intention alone defines goodness of an action, but does play a factor in whether it is done in good faith; we are not perfect beings with perfect knowledge – sometimes we do the wrong things for the right reasons – or the outcome ends up as an unintended disaster. Now, it’s difficult for another to know what a person’s intentions were, and often people may not even be fully honest with themselves in their full intentions and reasoning behind certain behaviours – but the theoretical essence of the intention exists somewhere as a combination of conflicting wants and whims- even if it is unknown and unknowable. If the resultant intention is not in the wish to inflict unnecessary suffering on another, or in fact to alleviate some suffering- then I think that it is a fair and good intention. (There is a case to be made for growth in necessary suffering, but this is a difficult line to draw, and a conversation for another time).

Outcome then, or ‘Consequentialism’ in philosophical terms: Do the outcomes of our actions matter in a system of ethics? I think so. I think that we are in our nature- predictive beings; given a situation or set of variables in a specific state or set of states, we predict future states and act accordingly in an environment-agent synergy. We often respond in a way that optimises some output of ‘value’- whether that be survival, money, well-being, or update of our prediction model in learning, (or combination of the above). We test the environment by making hypothetical predictions and then acting upon it, getting feedback on how certain actions cause certain responses from reality in the form of sense-data. We can use this to update our prediction capabilities, testing more behaviour patterns of the system. Carl Friston’s ‘Free Energy Principle’ or ‘Active Inference’ model of cognition captures this behaviour nicely:

The principle states that cognition (to be differentiated from consciousness), seeks to minimise its ‘free energy’. Free energy is proportional to the number of unknown potential states of a system, ‘phase-space of states’, or its ‘information entropy’. We should differentiate between known and unknown projected future states. Given an environment, a cognitive agent will seek to act upon the environment in such a way to limit the number of unknown future states. It can do this by experimenting and learning about the patterns of behaviour that govern the system, by which it increases the range of it’s internal ‘cone of projected potential states’. Proverbially: ‘fuck around and find out’. It can also minimise unknown states by ordering or acting upon the system, reducing the total range of possible states, and directing it to known sets of behaviour (even if it hasn’t mapped all of them out – though this can lead to entrenchment and repetition of negative behaviours simply by the virtue that they have known and familiar outcomes). An analogous example to demonstrate might be that of a messy room: it is ‘high entropy’ and difficult to navigate – when in preparation of a request to retrieve any one specific item from the room, we might approach a plan by exploring and remembering the layout of the room- mapping where each item is (increasing our knowledge), or- order the items so that we can find what we need more easily (reduce entropy). Ideally, after initial experimentation and exploration, this process happens as a synergy of the two approaches- adapting to new variables and influencing those we know how to. A constant process that tends to optimal limits in different environments.

If you’re familiar with many-worlds interpretation, it can be thought of like that. Our decisions span out in front of us like a tree, but unlike the items, or our intentions- the outcomes are not absolute. And assuming a non-deterministic universe (more on that another time), made possible through quantum phenomena – and assuming that we as conscious beings can interface with and influence the superpositional probability distribution of future states through free will (as biological quantum computers)… Well, that’s a tall order- and a big responsibility, and we should consider the mind of a poker player when approaching a probabilistic world. In the endeavour of maximising beneficial outcomes- learn to play the odds. More information is revealed as we play.

In essence, I would like to coin the phrase ‘entropy shepherds’ which I believe is what we are as conscious beings; guiding the branches of the world-tree of possibilities, pruning and nurturing. Yggdrasil’s keeper and gardener. We can’t know the outcome of each of our actions but we can get pretty good at weighing up the potentials, which should be our endeavour- and… To push the prospective branches of possibility in the right direction… So, which way is that?

So. The question of value systems. Which is the right one? I don’t know. A combination of a few in synergistic balance, perhaps. But first, I have some criticisms of the prevailing value systems. Namely the likes of self benefit, money, and fame… Money is measurable which is its biggest merit- and though it often enables an increase in freedom, influence, and well-being, it can also come at the expense of them in many settings (ultra rich people are often miserable and depraved – people need meaning). Money has become the object of desire and value instead of a mediator of value which is what it was supposed to be. Self benefit, again in healthy doses it is of course sensible- if nothing else, an element of self interest is required of altruists in order to help others; ‘you cannot fill from an empty bucket’. But self interest in its extremes comes at the expense of the individual, and the health of the system itself (which in turn damages the individual). Let me elaborate:

Profit might be maximised by large multinationals, they might do this by automating more jobs, by reducing costs, by increasing production, by increasing prices… But if it drives an economic crash (less workers, less consumers to buy the products), or destroys the environment through waste- are they really working in their own interests? In the value system of money, perhaps- but only in the short term. What is ultimately required is to actively go against the self-interested drive for growth and increased profit margins, and slow down to ensure a stable system. But, if one multinational slows down and doesn’t automate jobs, and another does… Well then they put themselves at an immediate disadvantage in that money-driven value system. It’s a game theory Mexican standoff. A sunk-cost-fallacy entrenchment of a short term Nash Equilibrium (if that isn’t too much jargon for one phrase). Each party acts in their own short term self interest for maximal returns, they are entrenched in this approach as it is only optimal in the short term- known as a ‘local minima’ in dimensional analysis (hence entrenchment). Each party knows that there is a better minima in another approach- but it requires short term sacrifice of value in the familiar system that is already mapped, they already have an edge in compared to most, and is a comfortable ‘known’. ‘The strategy has worked until now, money and capital-driven incentives have been functional for the development of society so far- and we’re in pretty deep at this point, so why change it?’; hence the sunk-cost-fallacy. This happens on an institutional level, and in the conditioned psyche of the individuals that ‘succeed’ in this value system- they got to where they are by playing /that/ game, so they’re pretty good at it- it’s a familiar system. Though people of this mindset succeed by having ignored a lot of other dimensions of value that don’t directly correlate to progress in the singular-dimensional value system in which they are trying to excel. It’s a tunnel-vision addiction to the idea of value.

The best strategy for each party to implement in order to maximise the acquisition of wealth comes at the expense of the system itself. The ‘game’ here is life and the world, and health, stability, and longevity should be primary factors to maximise. If one player at the poker table holds all the chips, the game stops. Or god forbid- the table falls apart out of pure negligence (‘I’ll buy a better one in the future, just one more hand I promise’, rings true of ‘more growth and progress will fix it’). Not all damage is reversible. And the only good solution is a question of trusting the other parties not to press the button and take the short term winnings at the expense of the game. Bloody idiots, the lot of us – god help us.

I would suggest then, that value systems worth optimising are the following: Freedom, well-being (reduction of unnecessary suffering), and the ability to learn at will. An interest in the well-being of others is good too- sure, busybodies can be annoying and there’s a line there somewhere, but trying to help too much is not quite as dire as someone mugging you. Let’s err on the side of looking out for one-another – not to appease our own moral self-image, or any expectation of return, but because it is the right thing to do.

There’s something to be said somewhere about leading with empathy, the naivety that can run parallel, and the ability for one bad actor to take advantage. Empathy is great, but is also easily hijacked if we aren’t careful; a conversation for another time I think.

Finally, if you’ve made it this far- I’d just like to state that this is a very rough draft, and the aim is to get ideas stated in some solid form so that I’ve got something to improve on later (much of what I post has existed in fragmented notes here or there for years). The writing will be revised and I’ll add more details in the future, as well as diagrams, etc. I apologise for the slightly haphazard ramble of words, but I often struggle to write without a sense of urgency, and it’ll read a lot better eventually.

For future reference for myself- I’d like to add Sartre’s comments on Camus’ l’etranger and the denial of game in the name of truth, and how it reflects Jesus’s resolution at the end, as well as that of Socrates (“apology for what- you guys should be fuggin paying me”). Also some elaboration on semantic ambiguity, and ego state interactions in transactional analysis (you’ve got written notes on notepad for this).

Life is not a zero sum game- reality expresses emergent properties.