From: Jeff Davis (jdavis@socketscience.com)
Date: Thu Nov 05 1998 - 17:57:16 MST
On Tue, 03 Nov 1998 12:24:58 -0500
Michael Lorrey wrote:
>My question is more environmental.<snip> If we increase or decrease in a big way the amount of heat being put into the atmosphere at that point, what is the effect that is going to have on the weather? <snip>...we could have some serious problems with this concept.
Problems are always a possibility. An accurate global climate model to predict environmental impact is, inevitably, to be desired. Then one can change the "half-empty" situation--uncertainty that the enterprise might jump up and bite you--to a "half-full" situation where you seek to achieve both the energy-generating benefits and a concurrent purposeful climate modification benefit. Make the weather better by design,... if possible.
Continuing, he writes:
>Solar power generating stations suffer from one serious fault: peak
>demand incompatibility. being able to store power from this intermittent
>source for supply to the grid at periods of peak demand (when the power
>is also worth the most) is of utmost importance.
It seemed to me likely that there would be multiple sites world wide--Sonoran, Saharan, Saudi Peninsula, Central Australian, and many others not quite so large--which would obviate any demand-timing mismatch, ...
as MaxM suggested when he wrote:
>I do believe though that it will be a fundamental flaw to keep energy
>production centralised like that. But that is not a fundamental flaw of the
>selfreplicating idea. Maybe we should just view it as an ecosystem with an
>excess production that we can harvest and move to other parts of the world.
Continuing with Michael Lorrey's comments,
>As for the amount of power it generates, the area of Arizona should see
>a solar flux level equivalent to over 250 terawatts. Given that the only
>solar technology currently available which has any kind of near-market
>cost effectiveness is thin film amorphous,
Stop right there.
At this point you begin to overlook the essential transformative character of self-replication, the basic premise of my post. Conventional market economics, and the related cost factors, go the way of the buggy whip when automation reaches the level of exponential machine self-replication.
Cost factors always originate in the labor component of a product or service, which, in the case of a self-replicating machine system, is confined to the design and construction of the "seed". The exponential growth of the system distributes that initial cost over the entire system. Not only that, but the manufacturing system which builds the solar array, then exists, free of any cost whatsoever, and ready to produce anything within its capabilities, in whatever quantity, very rapidly, at a cost based solely on reprogrammming. To put it bluntly, a macro-sized (in contrast to nano) machine self-replicating manufacturing system makes anything that current technology can make, in quantities that saturate demand, and at a "cost" which, shrinking to insignificance, frankly, challanges the definition of the term.
So single crystal vs. amorphous is a non-issue.
>... we are
>talking about 20 terawatts of power. This is a significant amount, but
>not a significant percentage of the world power demand.
However much is useful to make, you make. See above, specifically, "quantities that saturate demand".
On Wed, 4 Nov 1998 11:45:03 -0800 (PST)
Joe Jenkins wrote:
>Although I share much enthusiasm for self replicating machines, I am
>also skeptical of the benefits of cheap centralized energy. I am
>currently paying about $400 a month for power to my home. I don't
>remember the source or exact numbers, but only a small percent (around
>10%) of that is being used for paying the cost of energy. The
>majority is for distribution, maintenance of power lines, meter man,
>and general overhead. Even if the energy was free, my bill would
>still be around $360 and that's not exactly a major increase in
>quality of living. A real benefit would be an energy box in everyones
>home that requires occasional refills of cheap chemicals. Now that
>would make a difference.
I gave this some thought and realized, of course, you're right. The distribution system and administrative infrastructure would be the big cost factors. I can think of two ways to address this, given a self-replicating machine system.
Have the self-rep system address itself to manufacturing, installing, and maintaining the power DISTRIBUTION system, as well as the power generation system. All physical elements of the power distribution system--cables, poles, switches, conduits, etc.--can be manufactured on the same "cost-insignificant" basis. Programming the manufacturing system to produce these things is the only cost, but because power distibution systems are global in scale, the programming cost is spread over a very large system. (It seems paradoxical, but apparently, larger is cheaper. Other examples possessing this advantage are communication systems, pipelines, and all roadways and rail systems.) Insofar as people are still needed in the system, there will be conventional labor costs, but much reduced.
The other approach is distributed photovoltaic power generation, ie., home rooftop power. After all, if the system can make thousand square mile quantities of solar cells for power production out in the boonies, then it can just as easily make rooftop systems, in quantities to saturate demand. At negligible cost, of course.
Joe Jenkins implied as much when he later wrote:
>Household micro power plants
>encourage competition and eliminate the massive infrastructure needed.
----------------------------------------------------
An interesting feature of a self-replicating system, is that exponential growth makes component size irrevalent. Drexler's molecular scale system is just as powerful (more powerful, actually, because of the benefits of atomic precision) as a macro-sized system, dispite the fact that the replicators are infinitessimally smaller. You all know why. A few more generations more or less of replication, and the bulk of one system equals the bulk of another, and with it productivity.
What this means is that when you go to build a self-replicating system, you build it as small as possible without introducing excess cost due to miniaturization. If you build an oven to melt silicon to make chips, given a choice between one the size of a toaster and one the size of your thumb, you go with the one the size of your thumb.
Final note:
The central issue here is the transformational nature of self-replication. It is NOT about Arizona-sized solar power systems. That is a distraction which entered the picture because the exponentially growing automated manufacturing system (the self-rep system) naturally requires a power source; both during its growth phase from the initial "seed", and at maturity. Because the final size of the system must be capable of saturating demand for everything every human on the planet will want, it must, logically, be mega large. Therefore, its power system, must be mega large. And since power is needed by human society now and will be needed in the future, and since the first product the self-rep system makes for its own use is a power system, it is only logical that the first product it should make for society is excess power and more power systems. Don't be distracted. A robot making a power system, even a big one, is boring. A robot making EVERYTHING for free (almost free, OK? Don't bust my chops!) is
the economic singularity.
Nano ain't here yet. Cryonic reanimation ain't here yet (though suspension IS here, and SUCCESSFUL REANIMATION IS A NEAR CERTAINTY). Indefinitely perpetuated human lifespan ain't here yet (except by cryonic suspension, THE SUCCESS OF WHICH IS NEAR CERTAINTY). Human equivalent AI ain't here yet, nor SI, nor enhancements, nor uploading. But self-rep and the economic singularity that it will bring, marking the dawn of the transhuman era, is waiting to be built today,...this afternoon even.
Best, Jeff Davis
"Everything's hard till you know how to do it."
Ray Charles
This archive was generated by hypermail 2.1.5 : Fri Nov 01 2002 - 14:49:44 MST