As far as I can tell killing/not-killing a person isn’t the same not-making/make a person. I think this becomes more apparent if you consider the universe as timeless.
This is the thought experiment that comes to mind. It’s worth noting that all that follows depends heavily on how one calculates things.
Comparing the universes where we choose to make Jon to the one where we choose not to:
Universe A: Jon made; Jon lives a fulfilling life with global net utility of 2u.
Universe A’: Jon not-made; Jon doesn’t exist in this universe so the amount of utility he has is undefined.
Comparing the universes where we choose to kill an already made Jon to the one where we choose not to:
Universe B: Jon not killed; Jon lives a fulfilling life with global net utility of 2u.
Universe B’: Jon killed; Jon’s life is cut short, his life has a global net utility of u.
The marginal utility for Jon in Universe B vs B’ is easy to calculate, (2u—u) gives a total marginal utility (i.e. gain in utility) from choosing to not kill Jon over killing him of u.
However the marginal utility for Jon in Universe A vs A’ is undefined (in the same sense 1⁄0 is undefined). As Jon doesn’t exist in universe A’ it is impossible to assign a value to Utility_Jon_A’, as a result our marginal (Utility_Jon_A—Utility_Jon_A’) is equal to (u - [an undefined value]). As such our marginal utility lost or gained by choosing between universes A and A’ is undefined.
It follows from this that the marginal utility between any universe and A’ is undefined. In other words our rules for deciding which universe is better for Jon break down in this case.
I myself (probably) don’t have a preference for creating universes where I exist over ones where I don’t. However I’m sure that I don’t want this current existence of me to terminate.
So personally I choose maximise the utility of people who already exist over creating more people.
Eliezer explains here why bringing people into existence isn’t all that great even if someone existing over not existing has a defined(and positive) marginal utility.
As far as I can tell killing/not-killing a person isn’t the same not-making/make a person. I think this becomes more apparent if you consider the universe as timeless.
This is the thought experiment that comes to mind. It’s worth noting that all that follows depends heavily on how one calculates things.
Comparing the universes where we choose to make Jon to the one where we choose not to:
Universe A: Jon made; Jon lives a fulfilling life with global net utility of 2u.
Universe A’: Jon not-made; Jon doesn’t exist in this universe so the amount of utility he has is undefined.
Comparing the universes where we choose to kill an already made Jon to the one where we choose not to:
Universe B: Jon not killed; Jon lives a fulfilling life with global net utility of 2u.
Universe B’: Jon killed; Jon’s life is cut short, his life has a global net utility of u.
The marginal utility for Jon in Universe B vs B’ is easy to calculate, (2u—u) gives a total marginal utility (i.e. gain in utility) from choosing to not kill Jon over killing him of u.
However the marginal utility for Jon in Universe A vs A’ is undefined (in the same sense 1⁄0 is undefined). As Jon doesn’t exist in universe A’ it is impossible to assign a value to Utility_Jon_A’, as a result our marginal (Utility_Jon_A—Utility_Jon_A’) is equal to (u - [an undefined value]). As such our marginal utility lost or gained by choosing between universes A and A’ is undefined.
It follows from this that the marginal utility between any universe and A’ is undefined. In other words our rules for deciding which universe is better for Jon break down in this case.
I myself (probably) don’t have a preference for creating universes where I exist over ones where I don’t. However I’m sure that I don’t want this current existence of me to terminate.
So personally I choose maximise the utility of people who already exist over creating more people.
Eliezer explains here why bringing people into existence isn’t all that great even if someone existing over not existing has a defined(and positive) marginal utility.