We may think we observe time inconsistency but usually those observations come on the assumption that preferences remain the same through time. If preferences are fundamentally changing through time decisions may look as though they are time inconsistent but in fact we are dealing with, effectively, two different individuals with entirely different preferences, not time inconsistent preferences per se.
Oh and we cannot possibly predict how those preferences will change which would make price-setting for an externality tax bluddy difficult.

Smoking seems to me to be a commitment problem that doesn’t necessarily follow from hyberbolic discounting – although hyberbolic discounting does introduce time inconsistency.

I heard an interesting pointy of view on a podcast recently (can’t remember who the speaker was). It went something like:
If Jane starts smoking when she is 15 there must be some benefit to the smoking – maybe people think she is cool and she likes that. If Jane quits smoking in her early 20s, the probability of smoking related illness virtually dissappears. On those lines it is rational for the current Jane to start smoking.

It turns out it is Jane’s later self that is unwise for not quitting as Jane had previously intended. It is unfortunate that Jane didn’t have a commitment mechanism to ensure that her later self quit or perfect foresight to know that her later self couldn’t quit. But that isn’t necessarily an externality, just an unintended and uncertain consequence of a decision – recognising that while Jane didn’t quit most of her friends did.

If people’s decisions are time inconsistent (and I believe they are) that doesn’t say anything about a role for Government to intervene and encourage people to take time consistent decisions. If we accepted that there was such a role for Government, where would it end? Forced communal exercise while the Government chiefs checked up on you through a two-way tv?

Frankly I don’t think my current or future self would be happier with lost freedom but rock hard buns. Maybe that is just me.

To your questions:

1. I suppose folks could always argue that the first period self self-deceives about the ability of later-period selves to engage in restraint and only have a couple of smokes. Of course, at any period, the current self can hire in some self-restraint mechanisms. It’s not hard. If I were a smoker and I wanted to quit, I’d give $500 to a colleague down the hall and tell him that if he caught me smoking, he gets to keep the money. Ideally, somebody honest but somebody you don’t like enough to be happy to be giving him money.

2. I suppose the argument would be that the gains to the smoking you are less than the losses to the “feeling the effects” yous. That’s question-begging though absent some kind of hedonometer. What evidence we have suggests that smokers overestimate the risks of smoking rather than underestimating them, so that ought to knock out stories that smokers are misled. Hyperbolic discounting still could hold, but why would it hold more on smoking than on anything else? Proves too much.