RE: Observer-Moment Measure from Universe Measure

From: Hal Finney <>
Date: Wed, 15 Jun 2005 16:00:45 -0700 (PDT)

Jonathan Colvin writes, regarding the Doomsday argument:
> There's a simple answer to that one. Presumably, a million years from now in
> the Galactic Empire, the Doomsday argument is no longer controversial, and
> it will not be a topic for debate. The fact that we are all debating the
> Doomsday argument implies we are all part of the reference class: (people
> debating the doomsday argument), and we perforce can not be part of the
> Galactic Empire.

Well, I don't want to open up discussion of the DA. Suffice it to say
that good thinkers have spent considerable amounts of time considering
it and don't necessarily think that this reply puts it to bed. has an exhaustive discussion.

[Regarding measure and size]

> I find these conclusions counter-intuitive enough to suggest that deriving
> measure from a physical fraction of involved reasources is not the correct
> way to derive measure. It is not unlike trying to derive the importance of a
> book by weighing it.

Don't be too eager to throw out this concept of measure. It is
fundamental to the Schmidhuber and Tegmark approach to the multiverse.
It allows deriving why induction works as well as Occam's razor.
It explains why the universe is lawful and has a simple description.
It allows us in principle to calculate how likely we are to be in The
Matrix or some such simulation vs a basement-level universe. It is
quite an amazing quantity of results from such a simple assumption.
I don't think you will find anything else like it in philosophy.

As far as the specific issue of measure and size, suppose you agree
that making copies of a structure increases its measure, but you object
to the idea that scaling up its size would do so. Years ago I came up
with a thought experiment that adopted the position you have, that size
doesn't matter. (That's what my wife kept telling me, after all...)
>From that I proved that copies didn't matter either, which wasn't too
appealing. Today I would say that my premise was wrong. Size matters.

Here is a simple example. Suppose we have a book in a computer memory.
Now we make two copies of the book in memory. Perhaps you will agree
that this increases its measure. Maybe the measure doubles, or maybe
it doesn't go up quite that much, but it does increase.

Now suppose we arrange the two copies interleaved in memory. Instead of
"It was the best of times, it was the worst of times..." in two places,
we have "IItt wwaass tthhee bbeesstt ooff ttiimmeess...". Is this
still two copies, or is it one copy with extra big data representation?
The difference is not so clear. This should suggest, maybe there isn't
any difference in terms of measure of the two cases.

I have several other examples I have used as well. In one I have a
pair of electronic computers running exactly the same program side by
side, in lockstep. This is two instances and arguably the program has
larger measure. Now I attach wires between the corresponding parts
of the electronic circuits that make up the computers. Point A in the
left computer is attached to point A in the right computer, and so on
for every circuit element. These wires have variable resistance that
can be smoothly changed from full conductivity to perfect insulation.

When the wires are insulated, there is no interaction between the two
computers and there are two copies. When the wires are conductive, the
corresponding circuit elements are electrically joined so the system acts
like a single computer that is twice as big. By varying the resistance
we can smoothly go from one case to the other.

Imagine a conscious program running on this system. When there are two
computers, perhaps there are two conscious entities, each with their
unique identity. When there is only one computer, there is only one
consciousness. Yet we can switch smoothly between the two. We can go
from two people to one and back! How much sense does that make?

I wouldn't put it like that today, but if we just focus on measure,
we again go from a pair of computations to a computation that is twice
as big. It makes sense that the two cases would have the same measure.

>From these and other thought experiments I find that it is not as
surprising as it seems at first to imagine that increasing size increases
measure. The fact that it follows immediately from Schmidhuber's simple
principle only came to me recently. I find it interesting and provocative
that these two independent arguments lead to the same conclusion.

Hal Finney
Received on Wed Jun 15 2005 - 19:51:12 PDT

This archive was generated by hypermail 2.3.0 : Fri Feb 16 2018 - 13:20:10 PST