In what sense is utility fungible? Remember utility is fungible by definition—if 1 positive utilon doesn’t cancel out 1 negative utilon, then at least one of them was not actually 1.
I confused a perceived pattern in humans for a pattern in the world.
Assuming (and it may be so) humans are much more dutch-bookable along -loss/gain and -gain/loss lines than loss/loss or gain/gain lines, and we can project our utility function to remove muddle such that at the end two self-consistent value categories (loss and gain) can’t be made consistent with each other, that’s our problem. This is unlikely, as even if humans are most muddled along this axis, there is no difference in kind between unmuddling between gains and losses and within gains or losses.
Maybe unmuddling just can’t be done, but there’s little reason to believe that it can be partially done with the result being exactly two categories.
Remember utility is fungible by definition—if 1 positive utilon doesn’t cancel out 1 negative utilon, then at least one of them was not actually 1.
I object to holding that tightly to the definition. “Atoms” are divisible...assuming we figured out how to trade all utilities against each other, but only in two inconsistently related categories, “utility” would still be apt if reality was found to lack only that. We would then speak of “positive utility” and “negative utility” using complex numbers or something, able to say 10+5i is “more” (in some sense) than 6+i but not 11+i or 6+11i.
Inconsistency or Dutch-booking is bad regardless of fungibility, because they let you be pumped for arbitrary amounts. If they don’t, then they may simply reflect extreme preferences.
Are positive and negative utility fungible? What facts might we learn about the brain that would be evidence either way?
In what sense is utility fungible? Remember utility is fungible by definition—if 1 positive utilon doesn’t cancel out 1 negative utilon, then at least one of them was not actually 1.
I confused a perceived pattern in humans for a pattern in the world.
Assuming (and it may be so) humans are much more dutch-bookable along -loss/gain and -gain/loss lines than loss/loss or gain/gain lines, and we can project our utility function to remove muddle such that at the end two self-consistent value categories (loss and gain) can’t be made consistent with each other, that’s our problem. This is unlikely, as even if humans are most muddled along this axis, there is no difference in kind between unmuddling between gains and losses and within gains or losses.
Maybe unmuddling just can’t be done, but there’s little reason to believe that it can be partially done with the result being exactly two categories.
I object to holding that tightly to the definition. “Atoms” are divisible...assuming we figured out how to trade all utilities against each other, but only in two inconsistently related categories, “utility” would still be apt if reality was found to lack only that. We would then speak of “positive utility” and “negative utility” using complex numbers or something, able to say 10+5i is “more” (in some sense) than 6+i but not 11+i or 6+11i.
Inconsistency or Dutch-booking is bad regardless of fungibility, because they let you be pumped for arbitrary amounts. If they don’t, then they may simply reflect extreme preferences.
People work for money.
Is there a specific model of human utility you endorse? People prefer A to B, B to C, C to A etc.