Here's a little blast from the past.
I was ripping a few 1983-4 vintage controllers apart yesterday,
and came across some core memory boards.
Each board represents 4k x16 of non-volatile memory.
I was amazed to think how these things were put together,
or how much they would have cost to buy when new.
The way it works is this:
There is a simple matrix of NS and EW wires which carry a "half current",
and when polled create a "whole current" which magnetises the (doughnut shaped)
ferrite toroids.
However, there is a third wire which seems to pass through all the toroids,
the purpose of which I'm not sure. Does this carry a sort of ambient current to
keep the thing alive? Therefore, is this memory truly non-volatile?
Next question: When we talked about this style of memory back at tech in the 70's,
they use to employ rooms full of dear little old ladies with knitting needles (true!).
So, how did they make them this small?? Trained fleas?

The other thing that this demonstrates is that with all the
rod-packing/sphere-packing theorising we can do,
it's what it can be used for that counts.
The thing that distinguishes us from the slime is in observing
what quirks of the natural world can be utilised to our mutual benefit.

(I used to have a 25x magnifier. When it turns up again,
I'll be able to get in much much closer.)