[FoRK] Making robotics a priority

Eugen Leitl eugen at leitl.org
Thu Jul 14 04:48:10 PDT 2011


On Wed, Jul 13, 2011 at 08:04:31PM -0700, Ken Ganshirt @ Yahoo wrote:

> Say it often enough and eventually someone who hears you will believe that
> we are in some sort of "information consuming economy". Or, at least, that
> we're headed for one.   

We're probably heading for one, but it will take a while. Time
frame is probably a century, maybe slightly less. (This is in
terms of wall clock, real AI would change everything but I don't
think we'll be getting real AI before).

> The former is so far from the current reality that it's laughable.  The
> latter might happen but not in a timeframe that I'm interested in (my
> timeframe of primary interest extends at least to the lifetimes of my
> grandchildren who are 7 and 2).   

Cryonics could be an option, but it's currently in failure mode.

See http://chronopause.com/index.php/2011/05/29/a-visit-to-alcor/
and other posts on http://chronopause.com/

You have no idea how much this irks me.

> By the way, your contention that information consumption can easily
> approach infinite bandwidth is a crock.  In order for it to be true it must
> make the assumption that it can occur without the consumption of real
> resources and real energy.   

Total solar output is 4 MT/s of flux, and there's more than enough
matter in the local system to make use of it. We're currently a factor
of million away from Landauer limit, nevermind reversible computation.

> In rebutting someone else's similar comment, an IT pundit recently did a
> guesstimate of the resource and energy costs to run Google to illustrate the
> absurdity of the comment.  Regardless of the efficiencies Google has managed
> to achieve, the consumption rate of all the materials and energy required
> just to build, operate and maintain the information services of a single
> Google Corp is quite significant.   

It takes the entire hardware on the planet to track everything the
space between your ear does. Biology is also quite far removed from
limits of physics of computation, and there's plenty of unreal estate
in the sky.

> Your "information consuming economy" is still totally dependent upon
> significant consumption of Real Stuff.   

Of course. We're very far removed from the possible limits, but not in
terms of doubling steps.

> If you have envisioned some way to gather, store, share and consume bits at
> low or no resource and energy cost, I'm sure that would make for a very
> interesting thread.  But be sure to classify it appropriately: Speculative
> Fiction.  ;-)

Most of the current advances in HPC are energy-focused. We should see
the transition from electronics to spintronics/plasmonics/photonics over
the course of then next 30-40 years, or so. 3d integration of molecular
scale components should be also in time frame, or slightly later.


More information about the FoRK mailing list