Marginal frivolity
Until November 2025, the complexity of software made it a high-stakes business. Naturally, the very large amounts of capital required to write and maintain these systems made the work serious. We’re gonna change the world … because you kind of had to in order to achieve the return on investment needed to justify the investment costs.
Software used to be simpler, more fun and….stranger. I remember flying toasters in After Dark, whipping the llama’s ass with Winamp and essentially all of MySpace. These were weird products and reflected some of the eccentricities of the people who designed them. But I think as the market realities of venture capital scaled up and the smartphone gold rush ushered in a generation of unbundled, zero-to-one apps that were gamified and optimized, we lost a good bit of the frivolity.
The first time I wrote code was 1994 on an SGI IRIX machine and it’s indelibly etched into my brain as one of the most magical things I’d ever seen. Back then you could just FTP html, gifs, and eventually css, and your website was live. It was incredible and fast. Contrast that with today. You have multiple environments, integration providers, git and gh rules, deploy actions and all the ci/cd trappings – plus an incredible amount of deployment automation that surrounds the core pipeline. This can be true for a very simple website! If I had to go through all those loops today, I wouldn’t touch the stuff. It was fun because it was simple.
Thirty years later, software can be fun again.
Unless you’re a large enterprise and/or using the more costly APIs, the cost of writing code is still near zero relative to human labor. And so is the operational complexity, which the model and harness can sufficiently handle. The obvious outcome is that we’ll generate more code than we need. The more interesting outcome is one we should actively hope for: more frivolity.
Most individuals and companies haven’t had spare cycles for the work of fun. To be unexpectedly creative outside the bounds of the problem being solved was seen as both costly and out of step with how software was made. The discipline of craft was anchored in thoughtful interface design, not how expressive the experience was. That changes with zero-cost software development. Listen I don’t need my banking app to be sassy. But you’re already seeing this in agentic customer support and other LLM-led scenarios.
I hope we see more of it, especially in the small and indie space. We need more fun, strange software. There’s no more serious work than play.