THE PUZZLE OF PUZZLES

I don't think I was ever very interested in puzzles as a child, by 
which I include things like crosswords and Sudoku when I was older. 
I was vaguely interested in creating them, but I doubt I ever got 
anyone to actually try and solve one I'd made, so that wasn't very 
rewarding either.

Lately in some evenings, which are now too cold to abandon my spot 
on the couch under the heater, I've been slowly drawing up a 
revised copy of the messy schematic for my mother's late-delivered 
Christmas present, eventually destined for my website if I ever get 
that far. It occurred to me, not for the first time, that designing 
electronic circuits really is a puzzle (if you don't just copy 
example circuits from datasheets anyway), and one I do enjoy. 
Equally designing circuit board layouts is something I really get 
into, trying to crack the task of cramming all the components into 
the smallest area while connecting them all correctly with one or 
two layers of copper tracks.

I guess some people feel like this about computer programming too, 
and I sometimes feel a hint of that, but it wears thin much too 
quickly for me. In particular the chore of learning and remembering 
all the rules of the programming language quickly brings me down. 
People who try to learn new programming languages for fun are a 
mystery to me, it pains me just to switch between the few I do use 
and inevitably struggle to reignite my memories of their syntax. 
Also there's the different technical aspiration in programming to 
make things clear and readable. Circuits are often a battle of 
optimisation over clarity and I take some special glee at mapping 
out a combination of connections which is bizarre to the eye yet 
balances itself in a perfect rhythum of operation. Machine language 
programming is more like that, and I tried some of that once, but 
of course to make things that are really useful it soon grows to be 
too much to handle, and in the case of that particular project I 
gave up trying to debug it in the end. Of course you have to know 
the rules of some libraries in order to run a useful machine 
language program on an OS or using modern microcrontroller hardware 
peripherals anyway, lest every project becomes a life's work 
reimplementing the deep dark complexities of modern computer 
hardware and software, which are only more obscure rules to follow 
in the end.

I guess what I enjoy is more of a creative puzzle, without the 
drearily fixed rules of computer programming and hardware. Yet 
still to achieve a practical outcome, not just a pointless artwork. 
Of course there are rules to electronics, the laws of physics 
itself, but they're a challenge to tease and exploit rather than an 
ideal of order imposed onto you by others. It's just a shame that, 
much as I've tried, that sort of electronics isn't really 
marketable in isolation anymore. Everything in electronics is going 
into and/or coming out of a computer these days. The computer 
itself usually being a layer cake of complexity that can hardly be 
untangled completely by anyone alone. The computer puzzles always 
leave with a question mark still here or there, sometimes 
impossible to answer, even when it might be the cause of a problem. 
Then even at the end there's new hardware and/or software to be 
compatible with - it's not done, even when it does everything it 
needs to at that moment in time, nothing's ever permanently solved.

Perhaps AI code generators are an expression that I'm not alone 
with this frustration for a system that expresses code without such 
rules. But AI is not like the laws of physics in electronics, its 
behaviour can't really be pinned down, and the attempts to do so 
have apparantly already birthed a new field of "prompt engineering" 
describing rules effectively for a new, endlessly evolving, 
programming language that your new AI pre-compiler might or might 
not actually obey or interpret consistently. That's really the 
worst of programming as it already existed.

As an aside, I do think there's a way for a sort of AI to be 
helpful in another way. You could have a sort of fuzzy 
probabilistic code execution where one writes a program manually 
but an AI constructs test conditions around it then runs adaptive 
probabilistically-weighted random variations upon it, endlessly 
varying the code and reducing it to a near-perfect optimisation 
still fitting the broader logic of the original code. A super 
optimiser really, which doesn't just optimise the code itself, but 
the idea behind it, while automatically testing behaviour alongside 
the original code to ensure conformity. But that's another software 
puzzle, so I don't really want the big, tedious, (not to mention 
endless) task of trying to work on it myself. I'd sooner try that 
"big website idea", which I can't bring myself to really start work 
on either.

Alternatively I've got some ideas around analogue computing that 
I'd really like to play around with, but there'd never be any money 
in that, so I don't get around to it either. Instead I now pick 
most of my puzzles from the in-between zone, from where I don't 
really enjoy most of the work on them, and they don't really make 
me enough money in return. Then when I find some spare mental 
energy I tinker with some project done "my way" (often similar to 
everyone's way, circa 1970/80s, complete with graph paper and 
pencil), like someone with a hobby just working a normal 9 to 5 
passionless job for a faceless corporation, and making ten times my 
income. Yet in that role my creative frustrations would be 
accompanied by my usual social ones. In the end whether to take 
that as a hint about a necessary lifestyle change is another puzzle 
I don't like to try and solve...

 - The Free Thinker