This is well stated. It reminds me of an argument I used to have with social justice friends like 15 years ago, where I would be like “this trend (eg cancel culture) seems bad if we win/ get powerful” and they would say something like “that will never happen, we will always be tiny and comparatively powerless.” But then it turned out that not handling those problems was in fact bad when the movement did get powerful!
In astrodynamics this is a formal operation. You exploit the fact that the earth is way more massive than a satellite, and in LEO way stronger than the moon or sun etc. Elaborate approximation techniques exist to avoid solving 3-body or 4-body problems using the big-world approximation to use your metaphor. Astrodynamics also supplies the nicest metaphor for when you’re too big to use such approximations — you say “you exert your own gravity field.” Relatedly for problems where big world effects make you an ineffective npc, the Amazon heuristic of “get big fast” applies. One domain where this really matters is marketing. Below a certain scale of reach you’re basically a non-entity, no matter how clever your messaging. So you have to get past that threshold asap.
I actually like the astrodynamics comparison for two reasons:
You don't flip a switch from "big world, I pretend I have no gravity" to solving the full 3-body problem- Studying orbits was where the field of perturbation theory originated](https://en.wikipedia.org/wiki/Perturbation_theory#Beginnings_in_the_study_of_planetary_motion. The big world approximation is very good (like 98% accurate often), and it shows how you can iteratively add more disturbances and make your solution more accurate, depending on how much you care. We also have a very good handle on how large the perturbations are (the oblateness of earth is 100x larger than drag, which is 10x larger than the moon's effect, which is ...), though it's interesting that it's how much they accumulate, rather than the raw sizes, that matters for how wrong you are to neglect them.
But also, the orbit example brings up the idea of resonance, where it doesn't always go smoothly from "big world/2 body" -> perturbations -> small world. There can be many small tiny effects, all ignorable on their own, which, when aligned, suddenly overwhelm you and change the problem completely. The marketing example might actually be closer to this than a clean small/big world distinction, since if you make something that "resonates" with people (lol), suddenly you go viral and your effect is no longer small.
Re: marketing this is only true for mass market products. There are sets of products out there where there are only a handful of 'buyers' and size is less relevant vs personalisation.
This is really interesting. The first thing that came to mind for how to train the two in one sitting is poker. Every player who is serious about getting better starts by understanding the mathematics and general principles, what materializes as playing a game theoretically balanced strategy. What are pot odds on each street, what is your equity vs. a range of hands, how do you play a balanced, non-exploitable strategy in most situations based on your perception at the table. Where this gets interesting is when you have history and are playing against players who understand game theory optimal (GTO) strategies, and how to exploit them.
This is precisely where game theory exploitative strategies emerge, and where you have to think about things like "player A saw me do X in a previous situation, probably thinks I arrive to this situation with this range of hands, therefore I can bluff Y% and still have decent enough equity on the turn and river to make a value bet or let it go, if I make a hand against my representative range, then I can even bluff-shove (all in), so this is a good place to continue betting"
I'm sure sufficiently advanced players of most other games build this up, but you can't really advance beyond a certain level if you can't inhabit the other players' psychology and out-level them, i.e. this sort of adversarial, exploitative strategy seems like a natural endpoint in small-enough games. tl;dr play more infinite games :)
it's big in that you're far from the end of the game if you think about it like an iterated game. the impact of the "long run" outweighs the impact of your next move.
this is fairly nitpicky and doesn't reduce the value of the distinction overall BUT: the chess (and extend it to all two player games) example doesn't really work for this distinction as, on the one hand, your opening moves are locked into responding to your opponents moves from move 1, rather than being free to just generally do whatever in solitaire until some grace period ends, and on the other hand "just gathering resources that are valuable across many possible game states" is true throughout the entire game including the endgame, as in the branching factor of calculations explodes so fast that you are more often "calculating the general advantage of having a passed pawn or bishops vs knights" in the endgame, which is valuable across many possible game states, rather than calculating every specific response all the way to the end. the distinction is more like "sharp vs quiet" positions but a) that does not map all that well onto "ignoring opponents plans vs responding to opponents plans", nor b) does it map well onto "early in the game vs late in the game", as you can have very sharp openings trade into quiet/positional middlegames, for instance
Does big world equate to or allow for positive sum games, or mentalities, while smaller worlds, more cramped and constrained, are forced towards zero or even negative sum approaches? And this directionality could be either/both an empirical feature of the environment, like resource limitations, or attitudinal, like high-trust vs low trust cultures.
Thank you!! This has been refreshing and helpful :)
— —
Context:
I’m building a company and believe we’ll be really really big. But right now we are not, world much much bigger than us at this point. Big world intuition helps in uninstalling some of the behaviors which are unhelpful to us at this stage!
One thing this made me think of is that American political institutions were designed for a big-world with many small players on the assumption that it was mostly about preventing someone from becoming too big a player by force since that was the failure mode the Founders had seen. But our institutions are strained by the existence of many more ways to become big, and for efficiency reasons we've chosen not to impose stronger restrictions since roughly industrialization and the railroad boom (minus some occasional reversals, but on whole I think the trend holds).
nitpick: Use of heuristics versus other sources of truth is orthogonal to consequentialism vs other ethics. Heuristics chosen to maximize utility are still consequentialist ethics.
I think you are technically correct but maybe not practically so. One could argue (as I do) for a kind of 'virtue utilitarianism' that involves being virtuous I'm certain helping others ways but *not* doing galaxy brained multi step analysis into the deep future.
My beliefs here are, one could argue, just utilitarianism in the sense that I think it'll work out better this way, but the methodology and reasoning are distinct enough from classical utilitarianism that I don't think it's useful to conflate them.
It's impossible to do galaxy brained multi step analysis into the deep future that is qualitatively different from heuristics, because you need either to terminate in heuristics at some point (exactly in the same way that chess programs using A* search do), or you need to do an infinite amount of computation. So you can't separate the two.
In any case, we should divide ethical systems at the basic level into their basis for ethics, such as "consequentialism" as a class (and "utilitarian" as a subclass of that), rather than divide them along computational classes, because the greater difficulty and more-important question in ethics is not how to do ethical computations efficiently, but to decide what is good and what is bad. I guarantee that a person doing heuristic utilitarianism will have more significant disagreements with someone using heuristics based on a canon of theological texts, than with someone doing long-planning-horizon utilitarianism.
This is well stated. It reminds me of an argument I used to have with social justice friends like 15 years ago, where I would be like “this trend (eg cancel culture) seems bad if we win/ get powerful” and they would say something like “that will never happen, we will always be tiny and comparatively powerless.” But then it turned out that not handling those problems was in fact bad when the movement did get powerful!
In astrodynamics this is a formal operation. You exploit the fact that the earth is way more massive than a satellite, and in LEO way stronger than the moon or sun etc. Elaborate approximation techniques exist to avoid solving 3-body or 4-body problems using the big-world approximation to use your metaphor. Astrodynamics also supplies the nicest metaphor for when you’re too big to use such approximations — you say “you exert your own gravity field.” Relatedly for problems where big world effects make you an ineffective npc, the Amazon heuristic of “get big fast” applies. One domain where this really matters is marketing. Below a certain scale of reach you’re basically a non-entity, no matter how clever your messaging. So you have to get past that threshold asap.
I actually like the astrodynamics comparison for two reasons:
You don't flip a switch from "big world, I pretend I have no gravity" to solving the full 3-body problem- Studying orbits was where the field of perturbation theory originated](https://en.wikipedia.org/wiki/Perturbation_theory#Beginnings_in_the_study_of_planetary_motion. The big world approximation is very good (like 98% accurate often), and it shows how you can iteratively add more disturbances and make your solution more accurate, depending on how much you care. We also have a very good handle on how large the perturbations are (the oblateness of earth is 100x larger than drag, which is 10x larger than the moon's effect, which is ...), though it's interesting that it's how much they accumulate, rather than the raw sizes, that matters for how wrong you are to neglect them.
But also, the orbit example brings up the idea of resonance, where it doesn't always go smoothly from "big world/2 body" -> perturbations -> small world. There can be many small tiny effects, all ignorable on their own, which, when aligned, suddenly overwhelm you and change the problem completely. The marketing example might actually be closer to this than a clean small/big world distinction, since if you make something that "resonates" with people (lol), suddenly you go viral and your effect is no longer small.
Re: marketing this is only true for mass market products. There are sets of products out there where there are only a handful of 'buyers' and size is less relevant vs personalisation.
These are typically sales-driven markets
This is really interesting. The first thing that came to mind for how to train the two in one sitting is poker. Every player who is serious about getting better starts by understanding the mathematics and general principles, what materializes as playing a game theoretically balanced strategy. What are pot odds on each street, what is your equity vs. a range of hands, how do you play a balanced, non-exploitable strategy in most situations based on your perception at the table. Where this gets interesting is when you have history and are playing against players who understand game theory optimal (GTO) strategies, and how to exploit them.
This is precisely where game theory exploitative strategies emerge, and where you have to think about things like "player A saw me do X in a previous situation, probably thinks I arrive to this situation with this range of hands, therefore I can bluff Y% and still have decent enough equity on the turn and river to make a value bet or let it go, if I make a hand against my representative range, then I can even bluff-shove (all in), so this is a good place to continue betting"
I'm sure sufficiently advanced players of most other games build this up, but you can't really advance beyond a certain level if you can't inhabit the other players' psychology and out-level them, i.e. this sort of adversarial, exploitative strategy seems like a natural endpoint in small-enough games. tl;dr play more infinite games :)
I'm confused why this one isn't small world:
"when you expect that you’re going to run into the same people over and over again for a long time to come"
It seems like saying I'm spending the full enviroment's social budget at all times.
it's big in that you're far from the end of the game if you think about it like an iterated game. the impact of the "long run" outweighs the impact of your next move.
I see, thank you! I missed that angle on it.
this is fairly nitpicky and doesn't reduce the value of the distinction overall BUT: the chess (and extend it to all two player games) example doesn't really work for this distinction as, on the one hand, your opening moves are locked into responding to your opponents moves from move 1, rather than being free to just generally do whatever in solitaire until some grace period ends, and on the other hand "just gathering resources that are valuable across many possible game states" is true throughout the entire game including the endgame, as in the branching factor of calculations explodes so fast that you are more often "calculating the general advantage of having a passed pawn or bishops vs knights" in the endgame, which is valuable across many possible game states, rather than calculating every specific response all the way to the end. the distinction is more like "sharp vs quiet" positions but a) that does not map all that well onto "ignoring opponents plans vs responding to opponents plans", nor b) does it map well onto "early in the game vs late in the game", as you can have very sharp openings trade into quiet/positional middlegames, for instance
you clearly know more about chess than me! (I usually lose to my 8yo).
What's "sharp vs quiet"?
Does big world equate to or allow for positive sum games, or mentalities, while smaller worlds, more cramped and constrained, are forced towards zero or even negative sum approaches? And this directionality could be either/both an empirical feature of the environment, like resource limitations, or attitudinal, like high-trust vs low trust cultures.
Thank you!! This has been refreshing and helpful :)
— —
Context:
I’m building a company and believe we’ll be really really big. But right now we are not, world much much bigger than us at this point. Big world intuition helps in uninstalling some of the behaviors which are unhelpful to us at this stage!
One thing this made me think of is that American political institutions were designed for a big-world with many small players on the assumption that it was mostly about preventing someone from becoming too big a player by force since that was the failure mode the Founders had seen. But our institutions are strained by the existence of many more ways to become big, and for efficiency reasons we've chosen not to impose stronger restrictions since roughly industrialization and the railroad boom (minus some occasional reversals, but on whole I think the trend holds).
nitpick: Use of heuristics versus other sources of truth is orthogonal to consequentialism vs other ethics. Heuristics chosen to maximize utility are still consequentialist ethics.
I think you are technically correct but maybe not practically so. One could argue (as I do) for a kind of 'virtue utilitarianism' that involves being virtuous I'm certain helping others ways but *not* doing galaxy brained multi step analysis into the deep future.
My beliefs here are, one could argue, just utilitarianism in the sense that I think it'll work out better this way, but the methodology and reasoning are distinct enough from classical utilitarianism that I don't think it's useful to conflate them.
It's impossible to do galaxy brained multi step analysis into the deep future that is qualitatively different from heuristics, because you need either to terminate in heuristics at some point (exactly in the same way that chess programs using A* search do), or you need to do an infinite amount of computation. So you can't separate the two.
In any case, we should divide ethical systems at the basic level into their basis for ethics, such as "consequentialism" as a class (and "utilitarian" as a subclass of that), rather than divide them along computational classes, because the greater difficulty and more-important question in ethics is not how to do ethical computations efficiently, but to decide what is good and what is bad. I guarantee that a person doing heuristic utilitarianism will have more significant disagreements with someone using heuristics based on a canon of theological texts, than with someone doing long-planning-horizon utilitarianism.
Aligns well with my preference for abundance thinking. 🤔