Michael St Jules 🔸 comments on Can you control the past?

[Michael St Jules 🔸]

I think we should be able to find lots of examples in the real world like Smoking Lesion, and I think CDT looks better than EDT in more typical choice scenarios because of it. The ones where CDT goes wrong and EDT is right (as discussed in your post) seem pretty atypical to me, although they could still matter a lot. I think both theories are probably wrong.

What matters in Smoking Lesion are:

• Variables $A$, $B$ and $C$, with $B\ne A$ and $B\ne C$.

  • In Smoking Lesion, $A=Lesion$, $B=Smoking$ and $C=Cancer$.

• $B$ is what you’re deciding on.

• $A$ causes both $B$ and $C$ (positively), and this is the only way these three variables are related (or also $B$ prevents $C$ (causes $notC$ or reduces $C$), but overall $C$ is more likely conditional on $B$, without accounting for $A$.

• $C$ and $B$ have causal effects with opposite utility, conditioning on each value of $A$, but $C$‘s far outweigh $B$‘s (e.g. even just some cost to choose $B$ over not $B$ if binary, or choose to increase the value of $B$) conditioning on each of $A$ and not $A$, but $C$‘s far outweigh $B$’s. (And maybe some of this is stronger than necessary.)

• You have enough uncertainty about $A$, and to block the Tickle Defense, you also have enough uncertainty about any intermediate outcomes between $A$ and $C$.

For example, IQ and vegetarianism tend to be positively correlated, but I think this is likely almost entirely explained by higher IQs (or something close to it) causing vegetarianism (or lower IQs perventing it), or possibly something that causes both, not a large positive effect of vegetarianism on IQ. In this case, $A=C=-IQ$ (negative of IQ, which we want to minimize, to maximize IQ) and $B=notveg$. Of course, the last condition seems likely to not be satisfied for many people, but others may have enough uncertainty about their own IQs, because they’ve never taken an IQ test or observed anything that’s sufficiently correlated with IQ.

And to take things further, childhood and adult IQ are correlated, so an adult can increase their expected childhood IQ by becoming vegetarian.

There are probably others with IQ we could find.

Maybe we can take $A$ to be health consciousness, but there are probably many things ($=B$) health conscious people are more likely to do that don’t actually help.$C$ would be future health outcomes.

It seems like consistently following EDT risks justifying useless or even actively harmful self-signaling.

[PhantoMinecrafter]

Alright… So, let’s imagine, that I have a vaping habit which I may believe is not harmful at all. I have an option A, that states: “If I will believe vaping is not harmful—I won’t suffer in my future, as if I would smoke normal cigarettes”.

Then I die, respawn and have an option not to smoke/​vape. However, (if I am dying with the memory) I can memorize that vaping was cool. So here I have option B, which states:”I miss the melon flavor so much...”.

Here comes the Universe and gives me the link in the youtube advertisement regarding vaping and other stuff, which, for instance, maybe this one https://​​vawoo.co.uk/​​ (random from google, not spam, sorry, just as an example).

And here I have two more options:
C) To suffer from not smoking and try to force me to forget about that, even though the google system is smarter as we think and will continue to attack me with advertisements, forcing me to buy their subscription.
D) Forget about everything and then start to smoke again. But without the belief, this is not harmful.

I mean, this is life. People created so much stuff in order to kill themselves very slowly.