For people interested in modelling intertemporal allocation from a SF-ECON Lab perspective, there is a priceless (and simple!) article on Nature that I highly recommend: The ergodicity problem in economics by Ole Peters. The idea is to not maximize the discounted expected value of wealth, but rather “what will happen to my physical wealth as time goes by”. This is because the expected value is computed as an average of what happens to the same individual, who is taking a gamble, across multiple universes in which the gamble influencing her wealth (a random variable, a priori identical in all parallel universes) had different a posteriori realizations.
In ergodic systems, the discounted expected values (the discounted value of the realization of the random variable average across all these parallel universes) and the average over time of the individual’s wealth are the same. But most systems are not ergodic; systems in disequilibrium are almost always non-ergodic.
I will leave you to the article to enjoy the argument. For now, I want to flag to people who deal with this sort of decisions (for example in financial economics) that this might be a good way to model them.