[Fwd: Re: [Fwd: Singularity in twenty to forty years?]]<br />
From: <br />
"Paul D. Fernhout" <<a href="mailto:pdfernhout@kurtz-fernhout.com">pdfernhout@kurtz-fernhout.com</a>><br />
  To: <br />
Bryan Bishop <<a href="mailto:kanzure@gmail.com">kanzure@gmail.com</a>><br />
  Date: <br />
04/29/08 09:16 am<br />
-------- Original Message --------<br />
Subject: Re: [Fwd: Singularity in twenty to forty years?]<br />
Date: Mon, 14 Feb 2000 15:42:05 -0500<br />
From: Paul Fernhout <<a href="mailto:pdfernhout@kurtz-fernhout.com">pdfernhout@kurtz-fernhout.com</a>><br />
Organization: Kurtz-Fernhout Software<br />
To: <a href="mailto:ray@kurzweiltech.com">ray@kurzweiltech.com</a><br />
References: <007001bf6b33$2c4cd480$559969c7@ray><br />
<br />
Ray Kurzweil wrote:<font color="green"><br />
> Thanks for sharing this with me, I read it, and did find it of significant<br />
> interest.  I may quote from it in a future writing,</font><br />
<br />
Ray -<br />
<br />
I'd be curious which parts you found most interesting or of value.<br />
Feel free to quote from it if you wish.<br />
<br />
I didn't want to reply before reading your book. So, I've been occupied<br />
with reading "The Age of Spiritual Machines" in my spare time this past<br />
week or so.  I have many things I could say (it is well written and<br />
convincing for one). But I will focus on what I see as the most<br />
important issue you raise.<br />
<br />
First, I agree in general with the premise related to exponential growth<br />
of computing and many of the implications. That isn't due to your book,<br />
but your book definitely convinces me more of the truth of this. I don't<br />
believe I had been aware of your "Age of Intelligent Machines" book<br />
before. http://www.kurzweiltech.com/machines.htm<br />
<br />
In 1985, I had just graduated from Princeton (undergraduate), working<br />
with George Miller (Marvin Minsky's advisor in the Snarc days) doing<br />
cognitive science related to AI. I wanted to build an AI that could<br />
preserve homeostasis in a space habitat.<br />
<br />
Afterwards I went to live in Pittsburgh for a year and basically hung<br />
around CMU at the Robotics Institute. I was a visitor in Hans Moravec's<br />
Mobile Robot Lab at CMU/RI as he was writing "Mind Children" (proposing<br />
replacement -- downloading was a minor option then). I also was visiting<br />
with Red Whittaker's Field Robotics Laboratory (Red is a proponent of<br />
person in the loop -- robotics as augmenting human abilities). Of the<br />
two visions, I found Red Whittaker's more desirable (MOSH to the end I<br />
guess). I remember when one of the people working with Hans explained<br />
how we would be able to download our minds into robots and explore the<br />
universe. My reaction was mostly that technically I believed this was<br />
possible (in some way) but it was quite possible things would go wrong<br />
(with either transference or a generation of competitive machines).<br />
<br />
While your "The Age of Spiritual Machines" and Moravec's "Robot" both<br />
serve to convince me more strongly of the premise, I still retain the<br />
reaction that it is likely things will not work out so well for humanity<br />
as one might hope, without explicit action to construct alternatives.<br />
Some of this is based on my studies in Ecology and Evolution. I pursued<br />
a PhD but only got an MA from SUNY Stony Brook -- the program wasn't<br />
ready for ALife studies or ecological simulations as study topics there<br />
in 1990.<br />
<br />
In my opinion, your analysis of the evolutionary dynamics of a world<br />
wide web of downloaded humans is flawed because it ignores fundamental<br />
aspects of ecology and evolution. Specifically, here are two issues<br />
about your conclusion:<br />
a) it assumes humans in a different environment will still act human<br />
with classical human motivations (as opposed to dissolve into an<br />
unrecognizable set of bits or simply locking in a pleasure loop) because<br />
to a large extent environment elicits behavior, and<br />
b) it ignores evolution and its implications in the digital realm<br />
(especially the enhanced pace of evolution in such a network and the<br />
implications for survival).<br />
Of these, the most important is (b).<br />
<br />
Evolution is a powerful process. Humans have evolved to fit a niche in<br />
the world -- given a certain environment which includes a 3D reality and<br />
various other organisms (including humans). Humans have an immune<br />
systems (both mental and physical) capable of dealing with common<br />
intellectual and organismal pathogenic threats in their environment.<br />
There is no easy way to translate this to success in a digital<br />
environment, because the digital environment will imply different<br />
rewards and punishments for various behavior, and evolve predators and<br />
parasites which these immune systems have never been exposed to before.<br />
Human style intelligence is valuable in a human context for many reasons<br />
-- but sophisticated intelligence is not necessarily a key survival<br />
feature in other niches (say, smaller ones the size of roaches, hydra or<br />
bacteria). In short, the human way of thinking will be inadequate for<br />
survival in the digital realm. Even augmented minds that are connected<br />
to the network will face these threats and likely be unable to survive<br />
them. You discuss the importance of anti-viral precautions in your book,<br />
but I think you are rosily optimistic about this particular aspect.<br />
<br />
At best, one might in the short term construct digital environments for<br />
digital humans to live in, and defend these environments.  However, both<br />
digitized human minds and immensely larger digitized human worlds will<br />
be huge compared to the smallest amount of code that can be self<br />
replicating.  These digital "bacteria" will consume these digital human<br />
minds and worlds because the human minds and worlds will be constructed,<br />
not evolved. Human minds will be at a competitive disadvantage with<br />
smaller, quicker replicating code.  Nor will there be any likelihood of<br />
a meaningful merger of human mind with these evolved and continually<br />
evolving patterns.<br />
<br />
I could endlessly elaborate on this theme, but in short -- I find it<br />
highly unlikely that any mind designed to work well in meatspace will be<br />
optimal for cyberspace. It will be overwhelmed and quickly passed by in<br />
an evolutionary sense (and consumed for space and runtime).  It is<br />
likely this will happen within years of digitization (but possibly<br />
minutes or hours or seconds).  As an example experiment, create large<br />
programs (>10K) in Ray's Tierra and see how long they last!<br />
<a href="http://www.hip.atr.co.jp/~ray/tierra/tierra.html">http://www.hip.atr.co.jp/~ray/tierra/tierra.html</a><br />
<br />
Our best human attempts at designing digital carriers (even using<br />
evolutionary algorithms) will fail because of the inherent<br />
uncompetetiveness of clunky meatspace brain designs optimized for one<br />
environment and finding themselves in the digital realm. For a rough<br />
analog, consider how there is an upper limit of size to active creatures<br />
in 3D meatspace for a certain ecology.  While something might survive<br />
somehow derived from pieces of a digitized person, it will not resemble<br />
that person to any significant degree.  This network will be an alien<br />
environment and the creatures that live in it will be an alien life<br />
form.  One might be able to negotiate with some of them at some point in<br />
their evolution citing the commonality of evolved intelligence as a bond<br />
-- but humanity may have ceased to exist by then.<br />
<br />
In short, I agree with the exponential theme in your book and the growth<br />
of a smart network. We differ as to the implication of this. I think<br />
people (augmented or not) will be unable to survive in that digital<br />
world for any significant time period. Further, digital creatures<br />
inhabiting this network may be at odds or indifferent to human survival,<br />
yet human civilization will likely develop in such a way that it is<br />
dependent on this network.  The best one can hope for in the digital<br />
realm is "mind children" with little or no connection to the parents --<br />
but the link will be as tenuous as a person's relation to a well<br />
cultivated strain of Brewer's yeast, since the most competetive early<br />
digital organisms will be tiny.<br />
<br />
Once you start working from that premise -- the impossibility of people<br />
surviving in the digital world of 2050, then your book becomes a call to<br />
action. I don't think it is possible to stop this process for all the<br />
reasons you mention. It is my goal to create a technological alternative<br />
to this failure scenario. That alternative is macroscopic<br />
self-replicating (space) habitats.<br />
<a href="http://oscomak.net/">http://oscomak.net/</a><br />
However, they are no panacea. Occupants of such habitats will have to<br />
continually fight the self-replicating and self-extending network jungle<br />
for materials, space, and power. (Sounds like the making of a sci-fi<br />
thriller...) And they may well fail against the overwhelming odds of an<br />
expanding digital network without conscience or morality. Just look at<br />
Saberhagen's Beserker series <a href="http://www.berserker.com/">http://www.berserker.com/</a> or the Terminator<br />
movies.<br />
<br />
It will be difficult for you to change your opinion on this because you<br />
have been heavily rewarded for riding the digital wave. You were making<br />
money building reading machines before I bought my first computer -- a<br />
Kim-I.  But, I think someday the contradiction may become apparent of<br />
thinking the road to spiritual enlightenment can come from material<br />
competition (a point in your book which deserves much further<br />
elaboration). To the extent material competition drives the development<br />
of the digital realm the survival of humanity is in doubt.<br />
<br />
Still, you are a bright guy. If you study ecology and evolution in more<br />
detail, I think you may change your conclusion, or at least admit the<br />
significant probability of a bad outcome, and that we should plan<br />
accordingly.<br />
<br />
If you do change your opinion in the future, and wish to fund work<br />
related to helping ensure humanity survives the birth of the digital<br />
realm, please remember me.<br />
<br />
MOSH to the end I guess!<br />
<br />
-Paul Fernhout<br />
Kurtz-Fernhout Software<br />
=========================================================<br />
Developers of custom software and educational simulations<br />
Creators of the Garden with Insight(TM) garden simulator<br />
<a href="http://www.kurtz-fernhout.com/">http://www.kurtz-fernhout.com/</a><br />
<a href="http://www.kurtz-fernhout.com/oscomak">http://www.kurtz-fernhout.com/oscomak</a><br />