Reinventing the wheel factory

See here for the idea behind writing kata.

Software has eaten the world.

Maybe it's just the view from the inside talking, but I am perpetually annoyed that there are endless websites all doing essentially the same thing that all employ some number of (expensive!) developers on an endless treadmill to deliver essentially the same products.

Let's say you're interviewing for a software job. You talk to a dozen companies (pretend also that it's not 2025 and you can actually talk to software companies instead of having your resume disappear in an avalanche of AI-tuned slop resumes before it all gets unceremoniously swept into a dumpster),^[1] of which three are e-commerce sites. "Why do you want to work here?" the third one asks, and at this point the sting of your two prior rejections is still crisp, and so you trot out some milquetoast lie about being passionate about whatever-it-is they're selling, despite their frontend (and indeed, their entire architecture) being fundamentally identical to the previous two.

Having failed to convince them, you cycle back for another round of applications, asking yourself all the while, how can I fake enthusiasm when there's so little to be enthusiastic about?

Software is remarkable: we invented functions and libraries to allow us to re-use code, so that (in theory) we wouldn't have to reinvent wheels, and for the past ten or more years the largest growth industry in the world has been wheel reinvention.

You go to a storefront, and their search is broken, or their layout doesn't refresh correctly on navigation, or their shopping cart is broken, or a feature works in one view and is completely useless in another (ostensibly identical) view. And, if you have an ounce of self-awareness, you ask yourself "how can we be getting all this so wrong?"

Lest you think this is only about e-commerce, don't be mistaken: the entire web, and probably all software, is broken.^[2] It just so happens that e-commerce is one of the oldest justifications for companies having an online presence, and now, 30 years into the internet revolution, it's still a fundamentally unsolved problem.

At one point I thought no- and low-code would be the solution. The lowest common denominator websites all have the same general features: user accounts, search views, item drill-downs, and so on. In the narrower version, just considering e-commerce, you also have things like invoicing, customer support, and shopping carts. But you don't even have to try to find some feature that is broken on one website that works elsewhere.^[3][4]

No- and low-code promised that you could write components in one place and use them again and again, moving the frontier of abstraction up several layers from the rote toil of normal webdev. They have largely failed to deliver, with a couple exceptions: while companies like Shopify have managed to capture a significant share of the e-commerce market, a non-trivial portion of these sites are still bespoke.

Even worse, constantly shifting demands (mobile first, native apps, changing web standards) have ensured that the goalposts keep moving down field even as developers try to stave off boredom by any means available: writing new frameworks and libraries that do essentially the same things as existing tools; adding additional layers of complexity to their sites' architectures like a manic greek pastry chef building the world's tallest baklava;^[5] or chasing industry fads like a junkie looking for the next high even as the core product rots.

What have we gotten for all this? Some high salaries, for sure; but these are limited to a small fraction of the population. The average consumer has mostly gotten slow, broken, and confusing websites, and (extrapolating the through-line) the continued atrophy of local commerce and city centers.

"Just one more framework, bro; just one more standard, and we'll solve software forever," says the industry, oblivious to its own amnesia and long history of failing to solve coordination problems.

Are there other examples of software that attempts to solve a class of problem completely? What does that look like?

The answer is, sadly, yes; it's called SAP and it's an absolute nightmare of complexity. There are people who make their entire careers writing business logic in languages you've never heard of ("wow, you have 20 years experience with ABAP! one question: what's ABAP?"), and most users would practically rather slit their wrists that deal with SAP in all its glory and complexity.^[6][7]

The long and short of it is: software, as an industry, is a towering monument to man's inability to coordinate at the scales required for modernity; a shoddy, rickety, crumbling monument, twisted at impossible angles like lovecraftian architecture.

But, there's good news on the horizon: soon, we'll be able to sweep away the sin of all this wasted effort and lack of coordination, and all the good money sent after bad, as AI takes over writing software.

Hold on, I'm being passed a note...^[8]