On Thu, Jun 09, 2005 at 02:04:00PM -0700, "Hal Finney" wrote:

> I was working on an essay on the nature of thought experiments about
> copying, but it got bogged down, so I will make this short.  I am trying
> to analyze it based on evolutionary considerations.  Copying is much like
> biological reproduction and we can expect many of the same effects in
> a society in which copying is a long-standing and widely used technology.

Given the technology required for copying, it's a wash between Darwin and

> The most important effect is that making copies will be desirable.
> Just as genes try to reproduce themselves, so will people once that
> becomes possible, and for the same reason: successful reproducers occupy
> more of the universe's resources (i.e. have higher measure) and so these
> habits tend to become more widespread.

Definitely. Given the nature of propagation across spacetime, there will be a
selection for fastest replicators/travellers in the propagation wavefront.
Convergent evolution there, big time.

> When we consider thought experiments involving copies, it is important to
> understand these effects.  It is truly different to make a set of copies
> than to experience a probabilistic event.  Making copies increases your
> measure in the world; flipping a coin does not.  The decisions you will
> make in the two cases are different as a result.

I am not my copy, after subjective chronon has passed after bifurcation.
I can only observe my own branch, so does the other copy.

> One thought experiment was to consider two choices: flipping a coin and
> being tortured if it came up a certain way; versus making several copies
> and having one of them be tortured.  Assuming the copies are all going to
> survive, clearly the latter would be the one selected by evolution.

I have not fully understood the thinking behind torture/non torture 
gedanken. If a randomly selected branch gets tortured, I very much do not
want to become that branch. If there are two scenarios, one with lots of
branches, and one with far less, and still only one gets tortured, clearly
the former is preferrable, if a choice has to be made. The probability of me
being tortured next is lower that way. If all copies but the tortured one are
synchronized then there are still only two outcomes, and I will be tortured
with a 0.5 probability.

So far, everything is obvious. What am I missing?

> Copying is such a bonus that it swamps consideration of quality of life.
> In a world where people have adapted to copying, they would work as
> hard to make a copy as they would in our world to avoid dying (each one
> changes measure by plus or minus 100%).

As copying takes a cost, it's the same old rat race. Furthermore, the same
technology allows you to keep remote backups, with incremental syncronisation
and dead-man-switch instantiation. These are not copies, being static images,
until instantiated.

> It might be objected that this approach does not shed much light on what
> our expectations would be or should be about what we will experience when
> we go through these transformations.  I agree with the perspective that
> there is truly no "fact of the matter" about what it is like to have one
> of these things happen.  All we can really do is look at the experiences
> and memory of each person, at each moment.  No one will disagree about
> what each person at each moment remembers and how many of them there are.

Clones are people with the same memory, until bifurcation. Synchronized
clones are just one person, and can't differentiate in thinking/perception by
the sync boundary condition.

> That is really all there is, factually.
> Our attempt to make these novel situations fit our conventional
> expectations don't work because we currently have an implicit assumption
> of mental continuity which is violated by copying experiments.  There

Subjectively, mental continuity is preserved (notice that this is necessarily
true for meat people, yet we still do not consider them zombies, glossing
over lacunes). Continuity really doesn't mean much, given Korsakoff patients,
or confabulated memories.

> really is no meaningful and non-arbitrary way to map our current ways
> of thinking about the future to a world where copying is possible.
> But what we can do is really just as good: we can predict how people
> would and should behave.  Which preferences will they have in these
> thought experiments?  How hard will they work to achieve one option versus
> another?  Evolutionary theory provides guidelines and examples we can use
> to understand how people will behave if and when copying becomes possible.

If you have to copy, you have to split your possessions. Computational
substrate isn't free, and it will be a scarce resource, given that copying a
pattern is far quicker to build the substrate to contain it, even with
molecular fabrication.

Eugen* Leitl <a href="http://leitl.org";>leitl</a>
ICBM: 48.07100, 11.36820            http://www.leitl.org
8B29F6BE: 099D 78BA 2FD3 B014 B08A  7779 75B0 2443 8B29 F6BE

Attachment: signature.asc
Description: Digital signature

Reply via email to