I, for one, have "terminal value" for traveling back in time and riding a dinosaur, in the sense that worlds consistent with that event are ranked above most others. Now, of course, the realization of that particular goal is impossible, but possibility is orthogonal to preference.

The fact is, most things are impossible, but there's nothing wrong with having a general preference ordering over a superset of the set of physically possible worlds. Likewise, my probability distributions are over a superset of the actually physically possible outcomes.

When all the impossible things get eliminated and we move on like good rationalists, there are still choices to be made, and some things are still better than others. If I have to choose between a universe containing a bilion paperclips and a universe containing a single frozen dinosaur, my preference for ice cream over dirt is irrelevant, but I can still make a choice, and can still have a preference for the dinosaur (or the paperclips, whatever I happen to think is best).

I actually don't know what you even mean by my values dissolving, though. Sometimes I learn things that change how I would make choices. Maybe some day I will learn something that turns me into a nihilist such that I would prefer to wail about the meaninglessness of all my desires, but it seems unlikely.