in the same way that Minecraft teaches you to exercise agency and Factorio teaches you to optimize, are there any games that teach you to stare into the abyss? the ideal game would (a) reward you on a tight feedback loop for constantly admitting that you were wrong, (b) give you the option to not admit that you were wrong but make that decision acutely hurt. pastcasting is good for (a) but not good for (b) because you are sort of forced to confront being wrong all the time, which maybe teaches you that it doesn’t feel as bad as you might expect, but it doesn’t teach you to intentionally seek out things that could prove you wrong; and you don’t really have time to develop an attachment to your wrong ideas. most normal games reward you for staring into the abyss very indirectly because being good at intentional practice makes you do better over the very long run, but you don’t get immediate feedback loops for it, and so it’s easy to just not realize you could be doing a lot better.
Chess. Mistakes in chess usually become noticeable quickly, in just a move or two, and you have no RNG or teammates to blame them on. But to get better you have to acknowledge your mistakes and avoid making the same mistakes again.
i think the problem is that the feedback loop is too long—if you notice a mistake, there is no obvious action, and no immediate feeling of having improved. what you really want is something where you can choose whether or not to notice that you are making a mistake, and choosing to notice gives you immediate positive reinforcement.
How about math olympiads? They do reward you for solving complex problems and require to admit that your first conjectures were hopelessly wrong (unless, of course, you happened to get them right. Alas, this might come with practice faster than the habit of staring into the abyss)
I mean, it only gets to the stage staring into the abyss when you spend 1h+ on one hypothesis and get nothing and are getting desperate and are attached to your idea for proof of A but realize it’s probably \neg A. Mostly how it works is you collect observations then form hypotheses a test a few of those, and mostly you quickly realize what works and what doesn’t. And if I’m stuck and keep doing one thing it’s because I had tried many times to invent something better but I couldn’t. It’s a really, really difficult thing to pull yourself out of this “mode collapse” where you’re banging your head against the wall where there’s clearly a wall, but it’s a different skill from seeing the abyss because 1) it’s easy to notice your approach is lacking something but 2) “not making the mistake anymore” is not blocked by psychology but by g factor or something.
For the thing you’re interested in, how important is the “game” part? (Minecraft and Factorio are both particularly excellent games with rich depth, in a way that pastcasting is not particularly)
The hardest part of “stare into the abyss” is that it’s often about stuff that you’ve wrapped your identity around in a psychologically loadbearing where. When I hear “the Minecraft of staring into the abyss”, I’m imagining something that gets you invested in an overall direction in a complex world, that is the wrong direction, and then have the opportunity to change course on your goal.
I think my Planmaking & “Baba is You” exercise is at least related. (In this variant, your instruction is to form a complete plan for solving a Baba is You level on your first try. This gives people a lot of opportunity to get invested in a set of assumptions and keep building on them. People are usually quite overconfident in a way that felt a lot more “gut punchy” than other calibration training)
by game i mean it in a very very loose sense. video games, board games, card games, sports games, strange workshop activities, etc all count.
for the identity load bearing ness, it seems possible you could create it on a short time horizon. for example, even just arguing about something for 10 minutes can make me feel somewhat invested in my position. having teams in general can create some level of this. i feel like if you stacked a bunch of different psychological tricks you could kind of approximate it. even just getting used to the meta has this—i often find that i stagnate in a game because i learned some suboptimal meta, but i feel some emotional avoidance towards learning better meta because the displeasure of losing is less than the displeasure of learning the new meta; and the feedback loop of winning slightly more often from better meta is not very easily felt.
possibly you can design a game where you constantly have to accept better meta to even progress at all through the game. similar to how it is almost impossible to play Factorio without automation even though it’s technically possible.
i think it’s undesirable to have a game with one big twist that you build up to. for feedback loop reasons you want to have to do it over and over again and consistently get reward when you gaze into the abyss and not get reward when you don’t.
Calibration games such as https://www.quantifiedintuitions.org/? (a) You can choose to be wrong/overconfident, or you can acknowledge you don’t know when you don’t know. Acknowledging is rewarded. (b) The game pushes you to try to be overconfident by making you want to be top 1 (beat other teams). And it hurts to see you ranking if you are failing.
Not sure I buy the premise that (a) is needed or even good? I mean, part of abysses is that they don’t offer immediate feedback. What about a video game where everything is basically one-shot? You can spend as long as you want preparing, including gathering resources and doing science to the environment, and then you get one big shot; if it goes well you win, if not you lose and lose all your progress.
maybe if you were trying to make a game to teach the feeling of having one try to solve alignment, sure. but that’s not the game i want here.
if you want to get better at anything, including gazing into the abyss, then you want to get as many quality reps as possible in a fixed amount of time. a rep is higher quality if the feedback loop is tighter, and if the abyss is more painful to gaze into. if we had mind reading tech what you’d want is prompt the user to reflect on things that are emotionally painful, to detect the moment they push past the resistance to confront the emotion, and dump nicotine into their bloodstream 3 milliseconds later. unfortunately, we don’t have this technology, so we need some other way to do this
I’m saying that the bottleneck isn’t getting the feedback really fast, it’s having abysses to stare into at all. So my proposal is aimed at generating lots of abysses at all.
an idea: a game where there are several distinct but mutually exclusive strategies (eg a shooter where you can be a sniper, or a bullet sprayer, or a tank, etc), where you have to invest a bunch of time into specializing, and then you feel sunk cost about switching to a different strategy; but make the environmental conditions constantly change (in subtle or hard to reason ways so you have to spend a bunch of effort to notice things changing / there is plausible deniability as to whether things changed or whether you were always suboptimal), so that the optimal strategy changes frequently; and make there be strong diminishing returns to further investment in a strategy, which simultaneously makes the sunk costs feel bigger, and makes the initial gains from switching strategies feel very large so when you switch strategies you very quickly start winning.
I wonder if there’s a question-asking game, preferably one-on-one that would encourage this? Something akin to NYT’s 44 questions to make anyone fall in love, but instead 44 questions to stare into the abyss. Getting the right interlocutor and the right questions would be hard to do though.
It’s not a game, but it is a structured activity.
I’m skeptical that you can really get the abyss in small doses. Maybe there’s also a progressive activity where the first exercises are small things to admit about oneself, before progressing to more and more difficult questions.
If you’ve ever had a long match of Go where you are losing from midgame onwards, you will feel quite a lot of these emotions. Go games can last for quite some time, and the fractal nature of your mistakes can be realised to a fairly high resolution. Especially if your opponent is higher rank than you so you are playing with a handicap (“but I had so much ground at the start?!! How did it go so wrong?!?!?!”)
Rain World is survival-platformer whose protagonist is a nimble omnivore tool-user (similar niche to an ancestral human’s). The prospect if exploration is enticing, but you are in the middle of the food chain, and so must balance the need to survive with the your own drive to explore. Your creature must:
evade predators,
find and sufficient food/prey to hibernate,
Take shelter before a lethal rainstorm arrives.
Exploring means doing the above in less time. Regions are gated based on minimum survival streak so each sortie is like a bet on your ability. There are carnivorous plants. It is difficult and stressful. I highly highly recommend it.
in the same way that Minecraft teaches you to exercise agency and Factorio teaches you to optimize, are there any games that teach you to stare into the abyss? the ideal game would (a) reward you on a tight feedback loop for constantly admitting that you were wrong, (b) give you the option to not admit that you were wrong but make that decision acutely hurt. pastcasting is good for (a) but not good for (b) because you are sort of forced to confront being wrong all the time, which maybe teaches you that it doesn’t feel as bad as you might expect, but it doesn’t teach you to intentionally seek out things that could prove you wrong; and you don’t really have time to develop an attachment to your wrong ideas. most normal games reward you for staring into the abyss very indirectly because being good at intentional practice makes you do better over the very long run, but you don’t get immediate feedback loops for it, and so it’s easy to just not realize you could be doing a lot better.
Chess. Mistakes in chess usually become noticeable quickly, in just a move or two, and you have no RNG or teammates to blame them on. But to get better you have to acknowledge your mistakes and avoid making the same mistakes again.
i think the problem is that the feedback loop is too long—if you notice a mistake, there is no obvious action, and no immediate feeling of having improved. what you really want is something where you can choose whether or not to notice that you are making a mistake, and choosing to notice gives you immediate positive reinforcement.
Play against a strong chess engine while allowing yourself to undo as many moves as you like at any time and try to find any winning game?
In general I find that I can trace losses in Go games to moments when I acted unvirtuously (e.g. greedily, impatiently, fearfully, arrogantly, etc).
Go is also a long enough game that one mistake seldom sinks you (as long as you’re willing to give up the sunk cost).
How about math olympiads? They do reward you for solving complex problems and require to admit that your first conjectures were hopelessly wrong (unless, of course, you happened to get them right. Alas, this might come with practice faster than the habit of staring into the abyss)
I mean, it only gets to the stage staring into the abyss when you spend 1h+ on one hypothesis and get nothing and are getting desperate and are attached to your idea for proof of A but realize it’s probably \neg A.
Mostly how it works is you collect observations then form hypotheses a test a few of those, and mostly you quickly realize what works and what doesn’t. And if I’m stuck and keep doing one thing it’s because I had tried many times to invent something better but I couldn’t. It’s a really, really difficult thing to pull yourself out of this “mode collapse” where you’re banging your head against the wall where there’s clearly a wall, but it’s a different skill from seeing the abyss because 1) it’s easy to notice your approach is lacking something but 2) “not making the mistake anymore” is not blocked by psychology but by g factor or something.
Outer Wilds comes to mind. Or The Witness? Or any of the other “figuring out the rules is the game” sub-genre.
For the thing you’re interested in, how important is the “game” part? (Minecraft and Factorio are both particularly excellent games with rich depth, in a way that pastcasting is not particularly)
The hardest part of “stare into the abyss” is that it’s often about stuff that you’ve wrapped your identity around in a psychologically loadbearing where. When I hear “the Minecraft of staring into the abyss”, I’m imagining something that gets you invested in an overall direction in a complex world, that is the wrong direction, and then have the opportunity to change course on your goal.
I think my Planmaking & “Baba is You” exercise is at least related. (In this variant, your instruction is to form a complete plan for solving a Baba is You level on your first try. This gives people a lot of opportunity to get invested in a set of assumptions and keep building on them. People are usually quite overconfident in a way that felt a lot more “gut punchy” than other calibration training)
by game i mean it in a very very loose sense. video games, board games, card games, sports games, strange workshop activities, etc all count.
for the identity load bearing ness, it seems possible you could create it on a short time horizon. for example, even just arguing about something for 10 minutes can make me feel somewhat invested in my position. having teams in general can create some level of this. i feel like if you stacked a bunch of different psychological tricks you could kind of approximate it. even just getting used to the meta has this—i often find that i stagnate in a game because i learned some suboptimal meta, but i feel some emotional avoidance towards learning better meta because the displeasure of losing is less than the displeasure of learning the new meta; and the feedback loop of winning slightly more often from better meta is not very easily felt.
possibly you can design a game where you constantly have to accept better meta to even progress at all through the game. similar to how it is almost impossible to play Factorio without automation even though it’s technically possible.
i think it’s undesirable to have a game with one big twist that you build up to. for feedback loop reasons you want to have to do it over and over again and consistently get reward when you gaze into the abyss and not get reward when you don’t.
Calibration games such as https://www.quantifiedintuitions.org/?
(a) You can choose to be wrong/overconfident, or you can acknowledge you don’t know when you don’t know. Acknowledging is rewarded.
(b) The game pushes you to try to be overconfident by making you want to be top 1 (beat other teams). And it hurts to see you ranking if you are failing.
Not sure I buy the premise that (a) is needed or even good? I mean, part of abysses is that they don’t offer immediate feedback. What about a video game where everything is basically one-shot? You can spend as long as you want preparing, including gathering resources and doing science to the environment, and then you get one big shot; if it goes well you win, if not you lose and lose all your progress.
maybe if you were trying to make a game to teach the feeling of having one try to solve alignment, sure. but that’s not the game i want here.
if you want to get better at anything, including gazing into the abyss, then you want to get as many quality reps as possible in a fixed amount of time. a rep is higher quality if the feedback loop is tighter, and if the abyss is more painful to gaze into. if we had mind reading tech what you’d want is prompt the user to reflect on things that are emotionally painful, to detect the moment they push past the resistance to confront the emotion, and dump nicotine into their bloodstream 3 milliseconds later. unfortunately, we don’t have this technology, so we need some other way to do this
I’m saying that the bottleneck isn’t getting the feedback really fast, it’s having abysses to stare into at all. So my proposal is aimed at generating lots of abysses at all.
an idea: a game where there are several distinct but mutually exclusive strategies (eg a shooter where you can be a sniper, or a bullet sprayer, or a tank, etc), where you have to invest a bunch of time into specializing, and then you feel sunk cost about switching to a different strategy; but make the environmental conditions constantly change (in subtle or hard to reason ways so you have to spend a bunch of effort to notice things changing / there is plausible deniability as to whether things changed or whether you were always suboptimal), so that the optimal strategy changes frequently; and make there be strong diminishing returns to further investment in a strategy, which simultaneously makes the sunk costs feel bigger, and makes the initial gains from switching strategies feel very large so when you switch strategies you very quickly start winning.
I wonder if there’s a question-asking game, preferably one-on-one that would encourage this? Something akin to NYT’s 44 questions to make anyone fall in love, but instead 44 questions to stare into the abyss. Getting the right interlocutor and the right questions would be hard to do though.
It’s not a game, but it is a structured activity.
I’m skeptical that you can really get the abyss in small doses. Maybe there’s also a progressive activity where the first exercises are small things to admit about oneself, before progressing to more and more difficult questions.
idea: a cold shower connected to an IV drip that delivers a microdose of some habit forming chemical
If you’ve ever had a long match of Go where you are losing from midgame onwards, you will feel quite a lot of these emotions. Go games can last for quite some time, and the fractal nature of your mistakes can be realised to a fairly high resolution. Especially if your opponent is higher rank than you so you are playing with a handicap (“but I had so much ground at the start?!! How did it go so wrong?!?!?!”)
how do you avoid just closing the game without going back through all of your mistakes?
If you are playing a real opponent they can review it with you, or a tutor can do the same.
Rain World is survival-platformer whose protagonist is a nimble omnivore tool-user (similar niche to an ancestral human’s). The prospect if exploration is enticing, but you are in the middle of the food chain, and so must balance the need to survive with the your own drive to explore. Your creature must:
evade predators,
find and sufficient food/prey to hibernate,
Take shelter before a lethal rainstorm arrives.
Exploring means doing the above in less time. Regions are gated based on minimum survival streak so each sortie is like a bet on your ability. There are carnivorous plants. It is difficult and stressful. I highly highly recommend it.