Briar Patch

Don't Believe Everything You Prove

Mon, 10 Aug 2026 00:00:00 +0000

In many ways, this is a companion post to my complaint about John Backus’ Turing lecture. If you travel far enough along the smarter-than-thou asshole path, you’ll usually find the end of the line somewhere around formal proof methods.

The idea is that if we can elaborate a proof from a well-verified set of operations and a minimal collection of axioms, the thing we’re trying to demonstrate has to be correct. It’s the ultimate trump card for those seeking a pure, unassailable foundation for their program (or, increasingly, their philosophy of life).

Please Do Not Admire Its Purity

Fri, 31 Jul 2026 23:52:26 -0700

The title, if you didn’t catch it, is a reference to the movie Alien.

Specifically, it’s a reference to the part (spoiler alert for a movie from 1979) where the crew discovers one of their fellow crewmates, Ash, is actually an android. They discover this when he tries to kill Ellen Ripley, our protagonist, and after he’s clubbed to death with a fire extinguisher, the crew reconnects his severed head for a final interrogation.

The Death of Design Patterns

Sat, 04 Jul 2026 23:46:51 -0700

Every now and then, I see Design Patterns floated as a kind of next-step resource for programmers after they’ve gotten comfortable with a language, typically an object-oriented (or at least object-aware) one. I don’t particularly have a problem with OOP as a concept, though I agree with its critics that inheritance is overused in teaching materials by several orders of magnitude.

But as for Design Patterns itself, I personally think these days that it only serves as a window into 90s programming philosophy - a historical relic of a bygone age, and one that I’m not all that interested in revisiting (unlike my passion for Software Tools from the 70s).

GITR: Resources

Sat, 20 Jun 2026 23:53:52 -0700

Instead of running my mouth out the gate, I want to start by pointing to what other people have said, and resources that have undoubtedly influenced me. The bulk of this will be in book form, because that’s the type of media I prefer to learn from, but I’ll do my best to include websites, courses, and videos where I can.

I probably won’t put everything here at once. Instead, my intent is to have this post be one I can return to as time goes by, adding to the list and updating my views on what’s helpful and what isn’t.

Growing Into the Role

Sat, 20 Jun 2026 23:43:11 -0700

Intro

For a long time now, I’ve been thinking about how I would structure something like a course on software engineering. One thing that’s important to me is to emphasize practical aspects and “common sense” tools and techniques, because often what ought to be common really isn’t.

I feel that gap even more strongly because I came into writing software from a non-CS background, and in the early days it was just a means to an end for me. But flailing around without knowing much about even simple things like version control led to way, way too many mistakes and setbacks. In the end, I was almost mad when I found out there were much easier solutions to the problems I was having, and since I learned the hard way, the lessons stuck with me that much more strongly.

A Gun to Your Head

Sun, 07 Jun 2026 13:35:25 -0700

We Live in a Society

While I think I’ve made my position on voluntarily using LLMs for software engineering pretty clear, I realize that not everyone has the luxury of choice in our current corporate environment. In fact, I’m in that boat myself, with management breathing down my neck to show some sort of participation in the LLM craze.

As a result, I’ve spent a lot of time thinking about how someone whose career may be tied to an AI mandate outside their control can survive the current era without totally compromising their values.

Something Nicer

Sat, 06 Jun 2026 18:23:35 -0700

I realize I keep bouncing between screeds and fun little project summaries like a meth-addled squirrel, but I don’t want everything I write to be depressing or angry. It would be far too easy to do that, in all honesty, so I’m making myself look on the bright side.

In this case, you might remember me saying something about the flexibility of MUCK descriptions in my post about the line editor project. It went a bit like this:

LLMs Are Bad at Software

Mon, 01 Jun 2026 22:27:41 -0700

I am incredibly tired of “reasonable” takes about LLMs being used to develop software. “Well, they’re useful for-” No! They’re unethical plagiarism machines that exploit some of our worst addictive impulses while being trained to flatter us so we’ll spend more time and money on an ecologically destructive and inefficient method of approximating text sequences. They are insidious and I’m going to lay out how they screw up your coding and your critical thinking in all sorts of ways.

Line by line

Fri, 29 May 2026 00:55:04 -0700

Ancient History

Since I talked about building things last time, it’s only fair I share the things I’ve been working on. A lot recently has been inspired by old tech. Some of it is probably nostalgia for an era I didn’t really get to experience, but some of it is genuine delight. Lately I’ve been excited by the cutting edge of early-90s internet - the MUCK.

The hell is that? Well, it’s a less grind-oriented variant of the multi-user dungeon (MUD).

Get Weird

Wed, 20 May 2026 21:59:39 -0700

Now What?

Maybe you can feel it. The enshittification. The suits breathing down your neck to adopt a broken technology meant to replace you and deny you access to information save through their filters and their control. An assualt on the craft and creativity that used to be part and parcel of being a weirdo tech enthusiast.

OpenAI’s Super Bowl ad was an unwitting(?) tell - give us all your curiosity, all your attempts at learning, all your triumphs of insight, and we’ll give you back a chat box. That’s all you need. That’s all you are. It does everything you can. Better. Faster. More obediently.