I used to think of a cool project, sketch out what was needed, and start buying.
Invariably my original idea was half-baked - it wouldn't work as first thought, so had to be modified, which meant more buying.
Eventually I'd get it working, at least in part, but in the meantime extra addons would be seen as needed for extra coolness, which meant more buying.
But by that time other projects would have been thought of, which would take time and buying, so the first project would drift to the back of the bench, just waiting for the time and money to get finished and installed.
So now, the back of the bench is full of really cool in principle but half-baked in practice half-finished ... stuff. Half of which I don't recognise, it's so long ago I started them.
Now I think I've mastered the problem.
Instead of the buying phase, I just make a shopping list on my laptop.
And Write a short description of what it's supposed to do.
Then a circuit diagram, and a rough outline of the code.
Of course the code wouldn't work as-is, so work needs doing on it, and that throws up problems with the circuit as drawn, but they do eventually approach congruence, and may even show the way to extra coolness.
So we approach the dreaded buying phase.
But hang on - how much can we check out before we buy? We don't yet need those extras we were going to buy if we can make an LED or a buzzer stand in for the Harrier's rudder, or another Pi pretend to be the Pegasus engine control.
The more we can do virtually the less we have to buy.
So now my laptop is getting full of half-finished half-assed projects, most of which I don't recognise.
The latest project was started when my front door got changed from plain wood to all-singing-and-dancing fancy plastic.
The door-bell was wired through the wooden surround, but that has to change.
So a Pi-zero W with a button, connected by WiFi to another inside with a speaker and amp. Simple you may think.
Previously I would have bought on day one, but now ...
Apart from the obvious coolness of a camera attached, and of course solar-powered with battery back-up, we'd like it to play a tune depending on what facial recognition would find.
But what sort of sound should it make? A buzz is no good - we'd really like something that sounds like bagpipes, or maybe church bells.
What makes a church bell sound like it does?
So now, before we start working out the details of the project, we have to find out what it's going to do.
The first thing to note is it doesn't sound its fundemental - it only has 2nd, 3rd, and higher harmonics.
But there's a lot more to it than that.
I'm going to start with just half a dozen harmonics to see what it sounds like.
But the maths of even that small number gets complicated, because each one affects the rest, as if you've got a point wandering around a six-dimensional space and you want it to slowly (but not too slowly) approach the origin.
So that's where I am now.
Several months of time-wasting, terabytes of software written, and nothing bought! Not even an LED!