Hi folks, it’s my turn this week.
I’m planning to share some ideas in progress about the concept of Moloch (developed by Scott Alexander, and often seen as central to understanding the metacrisis) and how Evolutionary Game Theory can be used to clarify both what Moloch is, and how we should respond to it.
This ties into some previous discussions here on the forum about game theory and values.
And we now have a Luma calendar for these meetings that should make things a bit more accessible:
Looking forward to it.
Question on the time. Per Luma, the session is scheduled for: 9:00 AM - 10:00 AM PDT. But up till now, these calls have always been 8:00 AM -9:00 AM PDT. Is this an intentional shift? Or an artifact of something else?
Probably due to the fact European daylight savings time ended on the weekend…
Makes sense. Our daylight time ends in the upcoming weekend. So the hour changes for me this time only.
This video would be a good prerequisite for studying Moloch.
Change the last verse of this song to “beware of Moloch” and see how it works as theme music for the metacrisis.
Rufus rough notes / comments
PD / Moloch is credible as a description of many situations that have happened or could happen.
BTW on the piece in the wiki: Meditations on Moloch
Then the question is how do we get out of it …
[Interesting sociological question]
Seems many people who don’t have a classic academic background or know about political economy etc are engaging with these kind of ideas via moloch. so why was it successful? (Especially in Engineer / Rationalist / EA community).
If we follow Turchin’s logic to the end, it seems like global “ultrasociety” (beyond Moloch) would be a likely development. The macro-trend has been toward collaboration in larger and larger social formations. We just need to stay on trend.
A problem with that, of course, is the trend toward collaboration in larger social units flowed from imperial expansion and global conflict. Nukes, missiles, drones, etc. make such a fight-to-death, winner take the spoils model of establishing a global society rather high cost. So can we get there without the fight to the death first?
Here is where I find it necessary to exit Turchin’s model and delve more into psychology, culture, and consciousness. If we have to learn everything the hard way, WWIII seems like just a matter of time …. If by contrast we are able to visualize future dangers and adjust prior to disaster striking, then perhaps we can establish global governance without prior Armageddon. My own PD take on this is there absolutely no downside to working to avoid Armageddon. How does anyone “win” or “maximize” in a WWIII scenario?
Exactly - in game theoretic term avoiding existential risk is actually a simple cooperation game where there are rational incentives to cooperate. But because of lack of understanding of the risks, cultural factors, etc. agents’ actual preferences can be modelled as a PD (where the payoffs represent their perceived utilities rather than their actual utilities).
Yea this was great. This bit I think brings out the way in which thinking about the PD is still very relevant in today’s world - and to Second Renaissance.
Here’s one that I listened to today. Very interesting to see how Martin Nowak combines deep understanding of evolutionary game theory with a specifically Catholic religious perspective, especially given that we’ve been talking in this space about the possibility of rejuvenating existing religious frameworks.
Love the animation. Through that I also came across this on Nowak’s spatial models:
A few thoughts about the bit at the end on the “game theory of love.”
To me, spiritual practice in general renders identity fluid and expansive, so the notion of a “game” with distinct “players” and “payoffs” seems incoherent in that context. To me, in so far as spiritual practice has a goal or payoff, it lies in tension resolution. Metaphors of call and response work better for the phenomenonal experience of that process IMO rather than competing with anyone external. There is not really an “other” to overcome. The experienced tension is profoundly intimate. Moreover, the “prize” for moments clarity and tranquility arising from such tension resolution is quite likely to be even greater tension or sense of lack as one drills into further layers and depths. The “win” is in the “game” itself. Turning away from ”the game” will, however, absolutely manifest in a sense of loss. The most triumphant sort of victory in a game such as this is indistinguishable from surrender.
I agree, to become a winner you have to make a loser.
But still, to the analogy, many folks, myself included, speak of games from a non-trivial perspective of thinking strategy, rather than reducing life to something trivial as an actual game. Finite and Infinite Games by James Carse is such a deeply compassionate and beautiful book that deals with this notion in a manner I think you would enjoy. His idea is that highest gameplay, is one that enables us to keep playing, whereas winning ends the game, making it a finite rather than infinite game. I have a PDF if you would like?
Carse’s position reminds me of Johan Huizinga in Homo Ludens. For Huizinga, play is the origin of culture in general. Huizinga encompass all manner of ritual - dance, music, costume, etc. as “play”. Ancient lacrosse, for example, was more of a religious ritual than a competitive sport.
Anyway, I don’t really want to go down semantic rabbit holes about “games” and “play” and “strategy” and so on. PD-style game theory implies atomic actors separated from one another, in which the “winner” is indifferent to the suffering of the “loser”. Huizinga’s more anthropological notion of play is more like participation in the mystery of existence, practiced in societies for which the notion of an atomic individual scarcely existed. In primal human cultures, PD would be nonsensical. My sense of the future of culture is something like a reconnection with with the primal, in which PD will again be nonsensical.
I guess that rather than a game theory of love, I would prefer a love theory of games.
There are two ways in which in which PD-style theory allows taking into account the interests of others, including love etc.
The payoffs can represent total utilities, and the utility function of an individual can include the utility of others. If I am a pure loving altruist, my utility function might even be composed only of the wellbeing of other people, and ignore my own wellbeing.
The payoffs can be more limited, e.g. financial payoffs. When PD is interpreted in this way, it is no longer true to say it is irrational to choose cooperate strategy. Since your total utility can still be loving in the way outlined above, it can be rational for you to choose a strategy that gives you lower financial payoffs, given your knowledge of financial payoffs for others. This is called social preference theory.
So the individuals in game theory do not have to be ‘atomic’.
That’s an interesting perspective. My idea is to model “self” as infinite series of concentric rings encompassing ever larger social formations. Your idea seems tantamount to using a utility function that would in effect be an integral summation over the discreet utilities attributable to each of those self-domains.
Let me apply this to a use case that occupies my attention quite a bit lately: retirement planning, including estate planning. Estate planning, in particular, makes nonsense out of the notion of “victory” for the atomic self. https://www.youtube.com/watch?v=Z_edI_guXyc Likewise, the matter of personal “utility” looks quite different to the 60 year old, compared to the same person at 70, 80, 90, etc. In laying out an optimal sequence of deferred income, withdrawals for spending, insuring against disability, providing for offspring, and targeted giving to worthy causes, what exactly is one optimizing? One of my masters thesis advisors (about my age) recently passed from cancer. A hybrid in-person/online memorial was held for her. Many had stories to tell. Her influence in IT security education circles was prolific, touching at least a half dozen universities spanning from Hawaii to Wales. She mentored hundreds of students (including me) setting quite a few them on lucrative and socially impactful career paths. That’s what a wide-angled self looks like: ripples to the end of time. How does one optimize for that sort of self?
I’m also a big fan of the concentric rings idea. In general I suppose the utility function will be any function of those different domains U=f(D1,D2,D3..Dn) where D1,D2 etc are utility functions based entirely on each separate domain of concern. But a simple function would be a weighted sum: U=aD1+bD2+cD3… where a,b,c are any number greater than 0.
Estate planning as part of retirement planning is a really interesting use case here: I guess it forces one to quantify those concentric rings in financial terms, and therefore maybe shows you what a,b and c are in your case, as well as D1, D2 and D3.
Your final point raises the very valid concernt that when you extend beyond a certain time horizon, prediction and quantified utility functions become little more than make believe. (This is sometimes called ‘cluelessness’ in discussions of utilitarianism.) So here, as often, attention to the nearer term becomes unavoidable.
I want to describe how we really worked with a retirement planner a few years ago. This will sound very prosaic, but I believe similar principles could be extended to any domain one likes, up to and including addressing the metacrisis.
First of all, one takes inventory of assets (cash, stocks, bonds, etc). Then add in projected income sources (pensions, etc.). Then project expenses in various categories and in sum. The question is, will assets and income cover expenses into any foreseeable future (for retirement planning, through the age of 95 or thereabouts)? The problem is, anything based on stocks and bonds might vary with market fluctuations. Likewise, although pensions are secure on paper, events might get in the way. Also, inflation can turn current-currency financial flows into fractional purchasing power. The financial planners game all this out is by putting current numbers into a computer program that runs “Monte Carlo” simulations and returns a probably of reaching retirement targets. A Monte Carlo simulation just runs the scenario dozens of time with a random mix of good or bad market results in any given year. That’s about the best one can do, because who exactly can predict market and social conditions 20 years out? At most, one can hedge against the most likely risk categories, buffering against letting bad luck in a given year snowball into catastrophic consequences over the scope of an entire life.
To play this game on a more social level, one might try it with energy consumption, global average temperature, human demographics, etc. Then, depending on what time horizon one cares about and the scope of the society one cares about, one could place bets accordingly on which “investments” in social action now are likely to have the most impact over the modeled time period. There are literal computer simulations for this purpose, but I think even if this is just done intuitively, it can help mold thinking in useful directions.
Many retirees are tight-fisted with money, because they saved up a lot by being that way. Switching over to a personal consumption lifestyle goes against the grain. Such retirees tend to leave a legacy to children or other beneficiaries. I’m intrigued by the idea of painting a social purpose bequest picture that would appeal so such prudent investor types. What sorts of bequests would be prudent and canny in the sense of being demonstrably likely to yield a favorable return of whatever social purpose the donor favors?
