Contents

01. Experiments starting (Tim Boykett)
02. Fungible computons (George Homsy)
Return to Contents

----------------------------------------------------------------------

Date: 12 Jun 1998 04:50:22 -0400
From: tim@bruckner.stoch.uni-linz.ac.at (Tim Boykett)
Subject: Experiments starting


The attached document hopefully captures enough
of the details for you all to be able to
filter something out of the up-and-coming data streams

tim


- ---------------------------------------


Note on Lab Data


As a byproduct of the CTL lab series, data sets will be created with
ridiculous regularity. These datasets contain as many biomechanical data
points as possible, taken from the subject of the experiments in the
process of being experimented upon. It is hoped that, in some sense, these
datasets contain the summed data of the experiment, that no other
information will be necessary in order to be able to explain or understand
what has happened.

In collaboration with Nic Baginsky we are developing a process for the
automatic evaluation of these data sets, in principle the goal is to create
a scientist who is not influenced by any prior knowledge of the
experimental subject whatsoever. Not even the knowledge that this person is
a person should be apparent to this scientist. This automated scientist,
presently based upon a Kohonen model, attempts to maintain an objective
stance with regard to the experiment, and from this vantage point to cast a
net over the data. Given experience with a series of subjects, we hope that
the scientist will be able to begin to make generalisations about humans,
in a very scientific fashion.

Of course such a scientist, while possibly highly objective, and
definitely beyond manipulation for political, academic or other reasons,
cannot claim to have the one true answer. The multitude of possible
interpretations of data, the immense possibilities for other
understandings, not to mention misunderstandings and reinterpretations, is
overwhelming. For this reason, we are accumulating the data in
interpretable ways and logging these data-wads on the net, where we invite
pseudoscientists of all stripes to call past and cast their opinions on the
table.

The data sets are as follows:
- a subject data sheet, containing facts about the experimental subject.
We have tried to obtain as many relevant data points as possible, it is
however not always in keeping with the tenets of confidentiality that we
release all details to the public. For those researchers who need access to
more specific data points, biometrics or belief structure analyses, we can
only offer the chance to come in contact with us directly so we may mediate
the exchange.
- a graphical rendition of the objective scientist's (who unfortunately
remains nameless, please forward suggestions) analysis. For those amongst
us who comprehend more visually, this may be the key. For the others, we
can only offer the raw wad:
- a data log of biometrical points, including specifics on the process of
the objective scientist's progress. These measurements are plotted as note
values on 16 channels of midi data. This seems to be an efficient method
for the recording and distribution of experimental data. The exact
allocation of data flows to midi channels may vary from one experiment to
the next, details posted with each specific experiment. We invite the
examination of these data from all possible points of view. With
developments in the (re)interpretation of midi data, including video
effects (Imagine from or Gem
from Mark Danks ) and machine control
(Matt Heckert and Fam.
Brandt
) as well as music, this data can be interpreted in a
multitude of different ways. We invite music critics to analyse the
harmonic structure, dynamic systems professionals to look into the
dimensionality, stock market analysts to investigate the correlations with
the Dow and fishing boat operators to offer a barometric interpretation.

The first data sets have already been installed under
, interpretations are being sought
constantly. Please feel free to pass this pointer on to astrologers,
mathematicians, analysts, meteorologists, psychics, physicists, economists
and any other interpretational/methodological placeholders you know.

- -------- ----------------------
\ / TIME«S UP
\ / Industriezeile 33 B --------------------------------------------
\/ A-4020 Linz Tim Boykett
/\ ph:+43/732-787804 tim@bruckner.stoch.uni-linz.ac.at
/xx\ fax: +43/732-795742 --------------------------------------------
/xxxx\ http://www.timesup.org
- -------- ----------------------


Return to Contents

----------------------------------------------------------------------

Date: 12 Jun 1998 19:15:28 -0400
From: George Homsy
Subject: Fungible computons


Hi; Geo here. The bio-computation effort is not an attempt to do computations
which are useful in themselves, but rather an effort to bootstrap bacteria
as little molecular machine shops, to have them assemble things for us.

Those "things" could be microstructured materials, designer drugs, or
maybe someday, molecular scale computers.

Computers which run at millihertz may not of themselves be useful for doing
what we now think of as computation, but I'd love to have a trillion
little slaves who would be happy to work for a bit of sugar and a warm
place to stay, and who would reproduce themselves (including their program)
every twenty minutes if I wanted them to. How about you?


On another track, when I was working at QuickLogic, a manufacturer of
field programmable gate arrays, we were bandying about the idea of
programmable gate arrays which could reconfigure themselves on the
fly in response to computational needs. Basically, a CA type approach,
but with a model of parallel execution of distinct "programs" which
would vie for computational resources in much the same way as competing
species vie for turf (although hopefully in a non-destructive manner).

In this way, you could hook up large arrays of these things together,
maybe adding more processing power as you can afford it, or as Tim
mentioned, perhaps borrowing it from friends or renting it.


On still another track, there is an effort afoot here at the AI lab to
write programs for what they call an "amorphous" model of computation:
Unsynchronized, blind, unreliable, identitiless processors, randomly
distributed in a communication space. Imagine a 3D crowd of people
in the dark. All they can do is yell and hope someone hears them.
The only thing they know initially is what their instructions are, and
that (probably) everyone else has the same set of instructions.

Now, the question is: What kinds of things can you do with such a
simple computational model? Some of the results obtained so far are
quite impressive, given the wimpiness of the model: Reliably point-to-
point communication, local clock synchronization by a "firefly"
algorithm, distance measurement, local curvature measurement, and
production of various biologically and electronically relevant
structures, to give a short list. Email me for pointers to the
web page.


Cheers,
--Geo

Return to Jun. Index