Complementary-Goods

It is widely known that the Claude Max 20x plan doesn’t buy you just $200 worth of API (Application Programming Interface) credits in a slightly more convenient format. It’s closer to … well, an order of magnitude is what I have heard, but I’ve heard other people say it goes up to $5,000–6,000 worth of API credits.1 In any case, it’s a lot. What’s the catch? Enterprises — not startups, usually — shy away from Claude Max 20x because it has certain data retention properties that they consider too commercially valuable to waive away.2 So instead they either work through the API directly or use something else like GitHub Copilot for Business. ...

I wrote my first Clojure program in anger a few months ago. Only, it’s not accurate to say that I wrote it; AI did most of the writing, and I directed it on what my design goals were and why. Architecturally, you could call it was a stateless anti-corruption layer, yapping between some REST endpoints and some ancient proprietary serialization. I had never written a Clojure program of nontrivial size before. Indeed I had barely touched any Lisp at all since working through SICP during a high school summer vacation, over a decade ago. That didn’t matter. Much more important was emphasizing certain invariants from the start, like the aforementioned statelessness, which made it very easy to unit and integration test, which was the true challenge. ...

I use Claude Code. I like Claude Code. According to Steve Yegge’s Eight-Circuit Model of Claudesciousness, I’m a pretty solid Stage 6, edging into 7 on heavy days. I think most engineers, most of the time, can get most of their work done faster with one of these tools than not, although the force multiplier of that speedup is probably not as extreme as it is in my case. Why is that? ...

Lua and Scheme occupy curiously similar places in my noggin, becuase my primary use case for both has been as scripting languages embedded into other, larger projects. Lua of course is basically designed for this job, but it’s a solid intermediate format to target even if you have no immediate intentions to use the programming language parts at all. A TSV file like 1 2 10.0.0.5 192.168.1.1 80 tcp 10.0.0.6 192.168.1l.1 443 tcp can be given a lot more structure on the cheap by writing a small transpiler of sorts to turn it into ...

I am an ardent capitalist. I am also an ardent Unix philosopher. I have long percieved there to be some interesting tensions between these two wolves inside me. The Unix philosopher in me wants my code to be as simple and flexible as possible, and - not always, but often - to minimize the number of lines of code I need to sling to solve a given problem.1 The capitalist in me wants to cover as many edge cases as possible, make every transition as smooth as possible, and apply as much napalm to the fire of user acquisition and douse its twin flame user churn in liquid death of another kind. What to do? ...

It’s no secret that I’m a fan of economics-inspired solutions to otherwise hard problems. The other day I happened across an old post by GMU economist Bryan Caplan which I think does this very elegantly, for a problem of some interest to me. The vast majority of wills evenly divide the residuary estate between children. Mine evenly divides the residuary estate between (children and grandchildren). I like this a lot. It makes it unambiguously clear where your priorities in the case of your untimely demise lie. It is also self-reinforcing; children of yours who have no children themselves will recieve less, but when it comes time to write their own wills, they won’t have any children or grandchildren to bequeath to anyway. Children of yours who have many children themselves, perhaps inspired by this very policy, might decide in the end they were duped - but it seems far more likely to me they’ll cite it as one of the biggest reasons they ultimately went for 4 instead of 3, or 2 instead of 1, or 1 instead of none. After seeing it work in practice on themselves, they might decide they want to incentivize bringing their own grandchildren into being in kind. ...

Perl 5 went through a long nadir of unpopularity due in large part to its deserved “Write Once, Read Never” reputation. So I was surprised to find out not only is it installed by default on Debian, it’s installed nearly everywhere by default. It’s even the non-shell scripting language of choice on OpenBSD! Perhaps the only thing more impressive than Perl’s utter ubiquity is its longevity. The latest major version of Perl was first released in 1994. It came into existence on this planet less than a year after I did. It’s even arguably more portable than the median shell script - different Unices might use Bash, Zsh, Ksh, or even something newfangled like Fish, but for the most part a Perl 5 program is a Perl 5 program is a Perl 5 program. ...

Peanut butter and jelly are complementary goods, as are cars and gasoline, newer cars and electricity, electricity and basically everything else. We don’t normally think of, say, Docker and Kubernetes as complementary goods in software engineering, because you can get both for the low price of free. Or can you? You still have to invest time in learning both, and as the famous saying goes… Say it takes X hours to learn Docker adequately. If Docker suddenly becomes easier to learn, such that it now takes only X / 2 hours, it’s reasonable to assume Kubernetes will become more popular in tandem, because one of its complements now costs less and is hence supplied in greater quantities anyway. You don’t want a peanut butter only sandwich, do you? ...