X-Message-Number: 25523
Date: Wed, 12 Jan 2005 07:42:43 -0500
From: Thomas Donaldson <>
Subject: CryoNet #25513 - #25522

Well, let's see here:

For Peter Merel: 
I mentioned in my last message that I would give a reference. It is
Rodney A Brooks, ROBOT. Brooks has been a leader in the approach
to building animate robots for some time, and in my previous message
I summarized (I hope clearly) just how he did it. If I remember rightly
something that happened years ago, when I was actually involved with
parallel computing, it turned out that a research group as Stanford
had been able to construct a robot able to move along a road very
slowly and painfully, occasionally running into the gutter or a 
branch of a tree --- and while they were making their robot, which
busily computed every step, Brooks had build several creatures
able to scamper about in similar settings. 

For Mathew Sullivan: 
As I said, you may be interested in my editorial in the coming 
PERIASTRON. As for the durability of our brains, I will defy you to
design a brain using whatever materials which will not be subject
to destruction by some means. A suitable helmet and pressure
suit would protect me from the setup you describe (clothes, too, can
be tools. Spacesuits and diving suits are examples). The fundamental
problem which we try to solve when we engineer anything is to find
a design which will perform as well as possible under the PARTICULAR
conditions under which we want to use it --- and we learn very soon
that no design will do well in all possible circumstances. Given that
if you're going to redesign your body so that it can suffer large
impacts and remain whole, you're dressing it in a helmet and pressure
suit, but making those clothes a part of it rather than removable.
Now do you really want to go everywhere in your pressure suit?

For Jeff Soreff:
Computer neural nets are at best only faint copies of a brain and
how it works. And here I go again: the latest PERIASTRON actually
has a discussion of how real brains work, with its primary source
a series of recent articles in NATURE 431(2004) 759-803. EVEN OUR
SYNAPSES CANNOT BE CONSIDERED SIMPLE CONNECTIONS BETWEEN NEURONS.
Even a true neural net (instead of one imitated in a single
computer) consists of multiple processors (one for each node) ---
which makes our brains very highly parallel. The most interesting
and distinctive thing about the approach taken by the neuroscientists
who wrote these articles was their attempt to work out just what
algorithms the different parts of our brain (including single
isolated synapses) performed. If you're a computer person you 
will have to get used to the biological language; however even by
now we have some algorithmic understanding of how our visual 
centers really work. 

No, I personally doubt very much that the kind of complex neural
nets which make our brains have failed or will fail at common
sense reasoning. But to attempt to imitate a highly parallel
(at least 10^11 processors) with a single processor no matter
how fast, or to fail to take account of the simple fact that
our brain connections constantly change, strikes me as a strategy
which is bound to fail.

And finally to anyone here who subscribes to PERIASTRON: As usual,
there were the usual number of typos. The cover sheet had the
biggest one (I try to reuse my cover sheets). It discussed 
subscriptions as if the rate for a single issue was $3.00 US,
and announced that it would soon increase to $4.00 an issue.
Sorry, but that's already happened. A year (6 issues) used to
be $18. It has now risen to $24.00 US. This was caused by changes
in the exchange rate of the US dollar. My apologies if you
think it's still $3.00 when you get your issue.

And apologies to all those who expected an earlier PERIASTRON. Due to 
various personal and nonpersonal interruptions I found it impossible to
get out a separate September issue before November rose above the 
horizon, so this PERIASTRON is really a double issue. It has a longer
editorial than usual, 2 articles, and more Science Reports than 
usual. It also has 2, rather than just 1, short articles: one about
a species of carp able to remain active for 5 days without oxygen, the
other about nanoscale self-assembly. But still, it is coming quite
late. It's now time for me to think about the NEXT PERIASTRON.

And yes, the side you'll see when you get your issue tells you
of the new Updates to the GUIDE. After all, I don't just publish
PERIASTRON.

              Best wishes and long long life for everyone,

                   Thomas Donaldson

Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=25523