In some EA/rat circles (especially in the Bay Area), the memes of ‘updating fast’, ‘bias to action’, and ‘avoiding sunk-cost fallacy’ are quite strong, perhaps influenced by start-up culture. These are often implicitly or explicitly invoked when starting new projects, deciding on priorities, or life planning.
However—on the opposite end of the meme-space—falling prey to ‘the neighbour’s grass always being greener’, i.e. overly trigger-happy updating leading to reduced focus and execution capabilities is an obvious failure mode of a bias to action. Always re-questioning yourself, what you are doing, and what you stand for could be termed ‘naive bayesianism’: In theory, it leads to an accurate map of the territory, in practice, it may be counterproductive to the goal of actually winning.
Obviously, there seems to be a trade-off between the two modes of thinking, and I can think of examples of situations where one or the other seems better. Kahnemann gives good examples of where inaction due to sunk-cost-fallacy, endowment bias, or similar leads to sub-optimal outcomes. On the other hand, an EA constantly changing their career and city due to updating on information about what’s the best possible career field to go into is an example of over-updating potentially being counter-productive at some point. Another one would be changing your partner, friend group, or social circles due to wanting to be in the best possible environment (which is common advice here) too often with obvious failure modes.
I would be interested in your views on meta-heuristics on in which circumstances one or the other mode of thinking seems more appropriate and how you are modeling the action/inaction trade-off.
One such thing is of course the classic explore/exploit trade-off, which seems to descriptively model a subset of action/inaction trade-offs quite well (with action being exploring, inaction being exploiting what you currently have) but does not seem that powerful or handy when it comes to (prescriptive) statements about how to ideally behave.
Besides, explore/exploit falls short when considering that the dutifully bayesian acts of constantly questioning the status quo, re-visiting the whiteboard, and deeply re-thinking life choices (which would be required to competently decide if exploring is worth it) are costly actions per se, too. They cost valuable time, mental energy, and may even negatiely affect mental health. This means that at some point, enough should be enough, ‘inaction’, i.e. not questioning but just continuing with the status quo, may be the preferred option.
I would be thrilled to hear your perspectives on this. It’s possible that there already some texts/sequences on this that I haven’t read so far, I would appreciate links to these, too! (Searching for bias to action, status quo bias, etc. didn’t turn up anything worthwhile but I might have missed something.)
Heuristics on bias to action versus status quo?
In some EA/rat circles (especially in the Bay Area), the memes of ‘updating fast’, ‘bias to action’, and ‘avoiding sunk-cost fallacy’ are quite strong, perhaps influenced by start-up culture. These are often implicitly or explicitly invoked when starting new projects, deciding on priorities, or life planning.
However—on the opposite end of the meme-space—falling prey to ‘the neighbour’s grass always being greener’, i.e. overly trigger-happy updating leading to reduced focus and execution capabilities is an obvious failure mode of a bias to action. Always re-questioning yourself, what you are doing, and what you stand for could be termed ‘naive bayesianism’: In theory, it leads to an accurate map of the territory, in practice, it may be counterproductive to the goal of actually winning.
Obviously, there seems to be a trade-off between the two modes of thinking, and I can think of examples of situations where one or the other seems better. Kahnemann gives good examples of where inaction due to sunk-cost-fallacy, endowment bias, or similar leads to sub-optimal outcomes. On the other hand, an EA constantly changing their career and city due to updating on information about what’s the best possible career field to go into is an example of over-updating potentially being counter-productive at some point. Another one would be changing your partner, friend group, or social circles due to wanting to be in the best possible environment (which is common advice here) too often with obvious failure modes.
I would be interested in your views on meta-heuristics on in which circumstances one or the other mode of thinking seems more appropriate and how you are modeling the action/inaction trade-off.
One such thing is of course the classic explore/exploit trade-off, which seems to descriptively model a subset of action/inaction trade-offs quite well (with action being exploring, inaction being exploiting what you currently have) but does not seem that powerful or handy when it comes to (prescriptive) statements about how to ideally behave.
Besides, explore/exploit falls short when considering that the dutifully bayesian acts of constantly questioning the status quo, re-visiting the whiteboard, and deeply re-thinking life choices (which would be required to competently decide if exploring is worth it) are costly actions per se, too. They cost valuable time, mental energy, and may even negatiely affect mental health. This means that at some point, enough should be enough, ‘inaction’, i.e. not questioning but just continuing with the status quo, may be the preferred option.
I would be thrilled to hear your perspectives on this. It’s possible that there already some texts/sequences on this that I haven’t read so far, I would appreciate links to these, too! (Searching for bias to action, status quo bias, etc. didn’t turn up anything worthwhile but I might have missed something.)