[FRIAM] Clocks

[Marcus Daniels]

It might be interesting to train a LLM on cellular automata rules to teach it 
how to design or detect reversible vs. destructive processes; to understand 
causality in operational terms.



From: Friam <friam-bounces at redfish.com> On Behalf Of Jon Zingale
Sent: Sunday, July 28, 2024 12:36 PM
To: friam at redfish.com
Subject: Re: [FRIAM] Clocks



Marcus,



Not bad first steps. It knows how to describe the construction of a clock (an 
atomic beam, recognition of electron states, etc...), now for it to complete 
the task. What additional steps would I need to prompt, write, describe, 
etc... for it to find examples not already in the convex hull of its training? 
What about the reliability condition? Once it has its atomic clock "in hand", 
how do I set it loose to discover additional reliable clocks? What I liked 
most about your (transduction?) post on qualia is the emphasis on doings free 
from natural language.



Steve,



I thought about the coupled resonances angle, which to some extent got me 
wondering whether the machine would need to have/be a reliable time-keeping 
mechanism itself. Also, the lever. I was curious if this was too low hanging? 
People often joke that if you know you have a hammer then everything is a 
nail, but what if you need a hammer? It seems like I might be able to generate 
(via genetic algorithm or something?) an endless unit test file, but I suspect 
that the tests would quickly become inconsistent.



Anyone,



One thing I am finding interesting about polycomputing is the lack of original 
intent an organ, process or artifact needs to have for another agent to find 
any use in that organ, process or artifact. Exaptation seems related to 
function seeking behavior, of course it also need not be.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://redfish.com/pipermail/friam_redfish.com/attachments/20240728/fcd83cba/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 5594 bytes
Desc: not available
URL: <http://redfish.com/pipermail/friam_redfish.com/attachments/20240728/fcd83cba/attachment.p7s>