Re: Mind Survival Strategies (was "Mind machines, a badly ..")

Robin Hanson (hanson@econ.berkeley.edu)
Mon, 09 Nov 1998 10:08:58 -0800

Hal Finney comments on Doug Bailey's idea:
>> There would be a gatekeeper AI that would be in charge of one backup. After
>> a 24 hour period (objective time), the gatekeeper would boot-up my backup
>> and let me scrutinize the current me (the end state of the last 24 hours of
>> mind morphing). If the backup me thinks the new me is acceptable, the
>> gatekeeper overwrites the new me over the backup copy and the cycle begins
>> again anew. If the backup me thinks the new me is unacceptable (e.g.,
>> suicidal, psychotic, just plain "weird") then the new me is overwritten with
>> the backup me and the cycle begins anew.
>
>This is an interesting approach. I can conceive of cases where it might
>not work well, but they are not very convincing. Maybe you could drift
>into an uptight, restricted mindset, and the only way out of it is by
>some kind of mind-blowing, consciousness-expanding experience, the very
>thing which would be rejected by your Judge. More serious might be a
>subtle memetic trap, like some kind of cult where you get sucked into
>it gradually. Now your Judge is helping to enforce your new mindset
>and it's going to be harder to break free.

My problem with this approach is that it seems to assume an unrealistic degree of individuality. You're going to have to wait more than a day to see if you do or don't like some new version of yourself. But if you wait a week or a month, then rejecting this new version becomes pretty expensive. This version has been interacting with the rest of the world, which is not going to be keen about backtracking all their interactions with you.

Consider that you will have close friends and associates, and they will be evolving with you. If you and a friend evolved together in a way you both like, but then an old you disapproves while an old her doesn't, do you revert or not? What about your co-workers and clients and customers? How often will they tolerate your not recalling interations with them?

It seems to me that old version review has a place, but that one can't afford to invoke it very often, and that the cost of doing so increases with time, even though it probably takes some time to see what changes you like.

Robin Hanson

hanson@econ.berkeley.edu     http://hanson.berkeley.edu/   
RWJF Health Policy Scholar             FAX: 510-643-8614 
140 Warren Hall, UC Berkeley, CA 94720-7360 510-643-1884