See Parfit (1984), Appendix I, esp. pp. 496-9

  • Insofar as the badness of death is a function of the non-satisfaction of desires, it’s mostly the nonsatisfaction of local categorical desires, not global categorical desires, that make it bad. 
    • Parfit argues (Parfit (1984), Appendix I, esp. pp. 496-9) desire satisfaction theories focusing on global desires are more plausible that summative theories that sum the frustration and satisfaction of both global and local desires. This is because they’d count being addicted to a plentiful drug and desiring and taking it daily as outweighing the satisfaction of a global desire not to be an addict. You can’t tweak theories to only count the satisfaction of desires one wants to have, because this wouldn’t count the frustration of a desire not to be in pain as bad. 
      • Could possibly say the global desire is really a desire not to be and addict, not not to become an addict, so each day is net negative, rather than eventually being outweighed. 
      • A better move is to only count the frustration of conditional desires as bad, and their satisfaction as neutral. This won’t be icky negative utilitarianism bc. Categorical desires are still treated symmetrically.