[OkayPhysicist]

Humans are terrible at doing what's best for them. They are pretty good at following local gradients, though. Smoking might kill you in 30 years, but right now it lets you fit in with the cool kids, or feels good once you're hooked. Not brushing your teeth might be terrible for them and your gums, eventually, but right now it saves you from having to do something.

At any given decision point, people are more likely to pick the option that provides some benefit to them. That looks very different from consistently picking the choice that is eventually best for them.

[jpadkins]

One reasoning flaw I've seen in this type of discussion is the assumption that the person has the same value system as you / the experts. In your example, it is assumed that the subject values a very long life. Maybe they don't, maybe they value smoking way more than a long life.

I largely agree with you, but I would tweak it to say "Humans are decent at doing what's best for them given their own values and knowledge".

[listenallyall]

Smoking is typically a bad example, IMO, because it really takes a lot to actually kill you^. Like 50 years, usually (even 30, in your example, is on the low side). Further, there really are no visible downsides among smokers in their first 10 years or so. Meanwhile lots of other bad habits - hard drugs, alcohol, over-eating, even just sloppiness or laziness - often have real, visible, negative effects pretty quickly.

^as in any situation, there is always the <1% of outliers