Given that, lets consider your visual system, comprised of:
An analog collimator (lens) sending focused analog signals (light)...
...to light sensitive transducers (photoreceptors) ...
...Using the Visual phototransduction cascade [1] to turn light into voltage.
That signal is then transmitted through the optic nerve, pre-processed for reaction and early processing, and then hits the visual cortex where it flops around and other sensory systems combine their voltages trying to make meaning of things in increasingly abstract ways.
That is basically the same process used for all other senses as well, but just using different sensor methodologies
However the actual data uses the same medium of chemical-voltage transport throughout the PNS and CNS.
So using this construct, you need a hardware system that can remap pathways between sensors without rigid architecture and via processes that are emergent based on environmental changes.
I propose then that a liquid electro-chemical substrate that is materially additive (accretes more permanent power flows) depending on how much power (P = Volts * Amps) is flowing between nodes. That substrate then fills a cube covered in anodes and cathodes that represent individual sensor inputs/outputs.
It's basically recreating a human nervous system with a gel like substrate as the "brain" that can "rewire" itself infinitely as sensors are changed or otherwise.
I have a drawing of this but haven't uploaded it anywhere. If you're interested I'll scan it in
Yep. Only in my case, my day-dreaming about this to date had me considering using digital components (digital switches, multiplexers, ADC/DAC's, etc) to build the configurable routing part. Not to say that I'm convinced that is the right approach, it's just some stuff that was bubbling around in my head.
I have a drawing of this but haven't uploaded it anywhere.
That would be great. And if you'd like to move this discussion to a different forum (aside from comments on an HN thread) feel free to ping me directly @ [email protected]. Or keeping it here is fine. Either way's good with me.
Also for anyone following along these are the images
My perception is there has been a steady stream of analog circuit work in recent decades (not quite pure "analog computing", but related to it) due to the integration of analog signaling into cell phone chips that otherwise are digital (see https://en.wikipedia.org/wiki/Mixed-signal_integrated_circui... ); not sure if any of that is relevant to your explorations.
It's funny you say that. Over the past few months I've ordered copies of almost every book on analog computing that Amazon has, planning to do a deep dive into the "current" state of that, with an eye towards figuring out if there's some "there, there" to explore. I say "current" in scare quotes because there aren't a lot of recent books on the topic, given that analog computing mostly died out decades ago. There does seem to be a bit of a limited resurgence of interest recently though.
Transformers IMO are basically pointing towards time series predictions as being the root of epistemological grounding for "intelligence"
I've also been spending some time studying neuroscience and the literature on "spiking neural networks" based on a vague hunch that that and the analog stuff might tie together into something interesting.
All of that said, I do think that it makes sense to think about hybrid analog/digital systems since digital has so many advantages, especially in terms of flexibility and programmability. The question then, is something like "what's the best architecture that lets you best take advantage of both analog and digital styles of computation"? I won't pretend to say I know the answer to that, but that's a path I'm trying to explore a bit.