[extropy-chat] and the nano/holo fun goes on...

[Adrian Tymes]

--- Emlyn <emlynoregan at gmail.com> wrote:
> I figured out you could do a TB for $1000 a couple
> of weeks ago... it
> blows my mind.

Amusing point: just last week, I completed some
contract programming for a startup (so we're not
talking billions of dollars), and one of the test
conditions for this program was making sure it could
work with multi-yottabyte disk allocations.

The details of what it was doing with that much space
are confidential, unfortunately.  But my software did
pass the test.  (Lesson learned: if you're tracking
bytes in Perl, watch out when you pass around 909 TB.
At around 910 TB - 16 decimal digits - Perl starts
insisting on displaying integers as scientific
notation, with some loss of accuracy.  There are
libraries to help get around this, if one remembers to
use them.)

> Don't worry Kevin. We'll keep finding ways to use
> all that mighty
> hardware, I promise.

For much of my career, I've been practicing techniques
to not use excess memory, writing small and fast
programs.  Make sure they could operate in parallel
cleanly, so if my client needed a lot more of them
(say, customer interface software for e-commerce sites
which start getting a lot more customers per unit
time), they could simply buy more hardware and load
the same software.

But I do have to wonder.  In many industries, large
enough quantity has often taken on a quality all its
own.  How many nodes are in, say, a rat's brain?  What
was the memory capacity, again?  If we could write a
perfect emulation of that, it seems likely that merely
scaling up (with tweaking to represent any
specified-by-genetics initial weights and external
connections) could imitate a dog's brain, or a chimp's
brain, or a human's, especially if wired up to a robot
with the same approximate sensor/motor bandwidth.
(How many rods and cones in each eye, again?
Megapixel, or more?  Certainly more than a low-end toy
camera's, the ones used by many research project
androids today.)