This is why orienting around the concept of teleology makes sense, right? Where the end goal is what you optimize for. That accounts for all n-order effects, even the unknownable ones (your method either works or it does not. If it “should” work but does not work in practice, you abandon it)
Teleology seems to be making a comeback (saw a recent Sabine Hossenfelder video on it). Funnily enough I stumbled on a decade old yudowsky essay dismissing it but I think all the objections are answerable now. Basically: the future does determine the present if you consider correct predictions of the future as knowledge in the same way knowledge of something happening physically far away from you is knowledge. You can make decisions in the present based on your ability to correctly predict the future.
This is why orienting around the concept of teleology makes sense, right? Where the end goal is what you optimize for. That accounts for all n-order effects, even the unknownable ones (your method either works or it does not. If it “should” work but does not work in practice, you abandon it)
Teleology seems to be making a comeback (saw a recent Sabine Hossenfelder video on it). Funnily enough I stumbled on a decade old yudowsky essay dismissing it but I think all the objections are answerable now. Basically: the future does determine the present if you consider correct predictions of the future as knowledge in the same way knowledge of something happening physically far away from you is knowledge. You can make decisions in the present based on your ability to correctly predict the future.