PVKK and Iron Nest both look great, but they should have put the gun on a train. Many "railway guns" were aimed in traverse by driving along a curved track.

So you get all the fun of a Big Gun, but you also get to drive a train. You know you want it.
https://en.wikipedia.org/wiki/Railway_gun

I really need to do some exercise, but the AQI right now suggests it would be a net negative.

People forget that procrastination is a superpower. I can pack for a 10-day trip in under 5 minutes.

Of course there is also a cost. I am incapable of unpacking from a 10-day trip in anything less than 5 weeks.

Aaaargh. I have fixed all the really difficult cerebral problems in my code, and I really want to release it, but I have one remaining problem that is almost fixed but has some dumb off-by-one or something and I can't find it.

The temptation to ship it anyway with a note saying "please email me when you fix this dumb thing" is very strong.

Some truncated titles there, but I think you'll figure it out.

Most heinous rhyme you secretly admire:

Me: "So you remember when Half Life 2 came out..."

Coworker: "Er... hang on... 2004... no, I wasn't sentient at the time."

Me: <crumbles to dust>

Normally when this happens, I get to the bit where I go "oooooh yeah of course that's not true - duh" and then I fix it and it works.

But no - I was doing the right thing all along. I just needed to sacrifice my time to the algorithm gods I guess.

Well, whatever - works now. Happy.

Odd. I was working on a corner-case in my rendering algo. And I was pretty sure I understood the problem, and what the solution was. But I just couldn't get it to work. Tried it a few ways. Huh. Well maybe I don't understand the problem?

So I wrote some debugging and small steps to prove/disprove each step. Bit by bit, stepping through it, and then I got... it working. But it's using the algo I originally wanted. I already understood the problem! Wut?

Maybe I had a dumb typo or something.

I just linked someone to a paper of mine that is 20 years old called "Linear-Speed Vertex Cache Optimisation." And I *only just now* realised the title is stupid. It should be either "Constant Speed" or "Linear Cost", i.e. it's O(n). "Linear Speed" sort of implies O(n^2) which is hardly impressive.

Oh well, enough people ignored that and actually read the paper, so that's fine. It's just wacky that I never noticed my crappy choice of words before.

UNFORTUNATELY. Yeah, you knew I was going there. Unfortunately, the current implementation of AI is bullshit. It doesn't open up doors, it just gives people a crutch. Everything you do, it did for you. You learn nothing. You cheat yourself and the game. I don't know how to fix that.

So maybe MAYBE something good will come of all this AI bollocks. Just like something good came of all that BASIC bollocks. Which is the love of coding. Real coding. Not using BASIC - that was just the seed. Flattening the learning curve. Getting the love and the need, if not the details, into kids.

But what they did was induct a whole generation of kids into the idea that you could make a computer do what you want. Once you understand that, and like it, and discover in your soul that this is what you want to do with your life, the exact mechanism is irrelevant. You'll learn real languages!

So this idea that it's going to revolutionize the high end of coding is very funny. Zero evidence for that. But what it might do is the same thing that the 80s home computers, all of which had a version of BASIC built in, did. They were terrible machines, and BASIC is a terrible language.

And the effects on the pros at each of those steps are well-documented. They were OK about C, wary of C++, did not like C#, and fuck no we're not using JS. The robustness and control dropped sharply, and hey have you checked with your local greybraid what they think of vibecoding recently? Exactly.

It just seems that the people who suddenly discover AI coding are the same people who discovered JS, who discovered C#, who discovered C++, who discovered C, who discovered macro assemblers - all just shifted by a decade. The productivity gain is there, but it's kinda linear.

But we have been making that same steady progress, opening up code for newbs, reducing the learning curve, for many decades. They're called programming languages. We never stopped inventing them. Copilot is just another programming language. And if you think JavaScript is shonky crap... hoooo boy...

I agree - it's nice that you, a newb, can now kinda code. Not by yourself, obviously - you need the AI's help, and will forever. But cool - you can now make an Arduino do a thing, and you couldn't before. That is cool. I am in favour of that (though I think a book would help you more).

(apologies if you already read this rant on Bluesky)

I am so tired of people who can't code telling me about how AI helps them code to 1% of the ability of a professional, which sure is 10x for them.

It's like me telling people that the invention of music theory will make actual musicians obsolete because now I, too, can write chords that sound OK.

What's the current most permissive license for code and algorithms? I thought it was MIT, but apparently there's problems with it? CC0 doesn't feel quite right - isn't that more for art/assets?

STB seems to have two licenses - MIT and "Public Domain (www.unlicense.org)". I guess I could just copy what STB does?