Re: PhD-thesis on Observational Selection Effects

From: Russell Standish <>
Date: Tue, 4 Jul 2000 11:14:24 +1000 (EST)

Jacques Mallah wrote, in response to Nick Bostrom

> Now for some (hopefully constructive) criticism. First, you say your
> paper presents the first mathematically explicit “observation theory”. No
> doubt a standard literature search would confirm that, convincing the old
> fogies. On the everything-list and some member’s web pages, though, the SSA
> is old hat. I checked and I have to admit that it was you who introduced
> the term “SSA” onto the list. By now though there are two long-opposed
> camps, the ASSA (correct) and RSSA (idiotic). Despite the length of your
> paper, I noticed you didn’t discuss your view on that.
> The idea, however, is one I (and no doubt others) understood well before
> I even heard of the list. It’s how I’ve always read Everett; to me he was
> the first ASSAer. You’ve seen my web page discussion of the anthropic
> principle (; it has
> significant overlap with your paper. I might like to expand it into an
> article sometime, maybe by throwing in a refutation of the RSSA. Come to
> think of it, Don Page’s work (
> touches on some of the same things.
> Really, this is fine because we all need to work. I guess I’m just
> wondering how this is all supposed to be done properly, but still letting us
> all look good :)
> coin. Nor is there any theoretical justification for not regarding all
> observer-moments to be in the same reference class. (As you can tell I am
> partial to the ASSA.)

I finally had a bit of a chance to glance through Nick's thesis. I
guess Nick stayed out of the ASSA/RSSA debate for good reasons.

His SSSA introduced SSA over observer moments, so one could rightly
say that the ASSA and the RSSA are both examples of SSSA. Note the
SSSA clearly states that the reference class includes all observer
moments that don't differ by any relevant respect. With the ASSA, the
reference class is all observer moments belonging to a particular
individual (rooted at that individual's birth, lets say). With RSSA,
the reference class is the set of all future moments adjacent to a
given observer moment. All of this assumes that a temporal partial
ordering in fact exists between observer moments.

With such different reference classes, it is no suprise that they
should address different questions. Arguments arise when questions are
incorrectly posed.

With the ASSA, the measure attached to a particular observer moment is
related to the complexity of that observer moment, ie how much history
is contained within it. It can also be computed by integrating the
RSSA values from birth over the history that leads up to that observer
moment. Several histories may need to be integrated over in the case
of forgotten events.

It would seem to me that when discussing expected self-observed age,
the ASSA is the wrong reference class, here age is a relevant
variable. However, the ASSA treats all observer moments of a given
individual throughout its lifetime as equivalent. Surely this is wrong.

Nevertheless, for other questions, the ASSA would be an appropriate
reference class.


Dr. Russell Standish Director
High Performance Computing Support Unit, Phone 9385 6967
UNSW SYDNEY 2052 Fax 9385 6965
Room 2075, Red Centre
Received on Mon Jul 03 2000 - 18:11:39 PDT

This archive was generated by hypermail 2.3.0 : Fri Feb 16 2018 - 13:20:07 PST