While generating all possible minds some random things I thought about while skimming the post:
Is my mind from 1 second or 1 year ago the same mind as my mind right now? Are we interested in Archimedes mind from any specific point in time or just at any point in his life?
The state of the minds we’re generating in our search. Are we generating minds whose state is that of a mind that has experienced one billion years of torture? Is that bad? Stopping torture is good, no?
How many minds will be generated that did not want to be resurrected?
Some of this depends on the technical details of how traversing the space of possible minds works and also something like...can we generate a mind without “running” it? If we can generate a mind, and the mind isn’t running, are we OK with generating the tortured mind if we can inspect it in a not-running state?
If we’re just generating possible minds, how many will say “Yo, I’m Archimedes!” but actually have no relation to our historical Archimedes?
Are we interested in Archimedes mind from any specific point in time or just at any point in his life?
One possible solution is to resurrect the oldest version of the person, but before the onset of any irreversible brain damage. For example, we bring back the version of Archimedes from some minutes before his murder by a Roman soldier. But for many patients, the timing will be tricky to decide (e.g. slowly progressing Alzheimer).
The state of the minds we’re generating in our search. Are we generating minds whose state is that of a mind that has experienced one billion years of torture? Is that bad? Stopping torture is good, no?
I agree, such states of mind should be excluded, if possible. At least because they’re unrealistic, and thus are unnecessary increasing the search space.
How many minds will be generated that did not want to be resurrected?
If we use the method #2, many trillions of such minds will be generated. I think, in most cases, we can repair the will to live without a deep modification of the mind. But I don’t know how to handle the remaining cases. The least we can do is to give them a choice: do you want to exist or not? We gave them the second chance, and they should decide if they want to use it.
can we generate a mind without “running” it? If we can generate a mind, and the mind isn’t running, are we OK with generating the tortured mind if we can inspect it in a not-running state?
I think a mind who is not experiencing any suffering is not increasing the total suffering. For example, a cryonics patient who is still in liquid nitrogen—can’t suffer, because their neurons are not firing. Same for a digital mind who is not running (e.g. just a static file on a HDD, doing nothing).
If we’re just generating possible minds, how many will say “Yo, I’m Archimedes!” but actually have no relation to our historical Archimedes?
There will be many trillions of such minds. I see two possible solutions: 1) we use some smart filtering to avoid generating unrealistic Archimedes, 2) we generate all Archimedes, real or fake, and just let them live their lives.
While generating all possible minds some random things I thought about while skimming the post:
Is my mind from 1 second or 1 year ago the same mind as my mind right now? Are we interested in Archimedes mind from any specific point in time or just at any point in his life?
The state of the minds we’re generating in our search. Are we generating minds whose state is that of a mind that has experienced one billion years of torture? Is that bad? Stopping torture is good, no?
How many minds will be generated that did not want to be resurrected?
Some of this depends on the technical details of how traversing the space of possible minds works and also something like...can we generate a mind without “running” it? If we can generate a mind, and the mind isn’t running, are we OK with generating the tortured mind if we can inspect it in a not-running state?
If we’re just generating possible minds, how many will say “Yo, I’m Archimedes!” but actually have no relation to our historical Archimedes?
Excellent questions!
One possible solution is to resurrect the oldest version of the person, but before the onset of any irreversible brain damage. For example, we bring back the version of Archimedes from some minutes before his murder by a Roman soldier. But for many patients, the timing will be tricky to decide (e.g. slowly progressing Alzheimer).
I agree, such states of mind should be excluded, if possible. At least because they’re unrealistic, and thus are unnecessary increasing the search space.
If we use the method #2, many trillions of such minds will be generated. I think, in most cases, we can repair the will to live without a deep modification of the mind. But I don’t know how to handle the remaining cases. The least we can do is to give them a choice: do you want to exist or not? We gave them the second chance, and they should decide if they want to use it.
I think a mind who is not experiencing any suffering is not increasing the total suffering. For example, a cryonics patient who is still in liquid nitrogen—can’t suffer, because their neurons are not firing. Same for a digital mind who is not running (e.g. just a static file on a HDD, doing nothing).
There will be many trillions of such minds. I see two possible solutions: 1) we use some smart filtering to avoid generating unrealistic Archimedes, 2) we generate all Archimedes, real or fake, and just let them live their lives.