Tech CEOs are breaking the law

Great products and services minimize friction—reducing the effort, time, or cost to the consumer. In system design, this is governed by Tesler’s Law (the “conservation of complexity”), coined by Larry Tesler at Xerox PARC in the 1980s.

The axiom states that at the core of every product is an activity with a non-reducible, non-negotiable amount of effort attached to it. That work must be assigned in some share to either the system or the end-user.

Smartphone apps are the ultimate expression of this adage. The act of hailing a cab, buying a book, or searching the world’s knowledge banks are now one-click transactions for consumers. All of the effort has shifted upstream to complex algorithms, logistics networks, and hyper-scaled data centers. And so the relentless reduction of consumer friction has become Silicon Valley’s guiding ontology—the foundational lens through which they filter reality.

And now the rise of Generative AI is preying on the petty vanity of tech CEOs, leading them to believe they can "one-click" optimize their own companies by automating away their workforce. Who needs product managers, UX designers, or software developers when the self-described “lone genius” founder can dream up an idea in the morning, craft a few GenAI prompts, and launch a new app in the afternoon?

But this is both a delusion and a category error which confuses a socio-technical system (the business and its potential users) with the software and services produced by that business. It fails on two fronts:

First, solving for human needs is always the hardest part of software development. Figuring out what to build, why, and for whom cannot be automated; it is discovered only through the friction of a deliberative process.

Second, because complexity can only be “conserved,” never eliminated, outsourcing that intellectual burden to a Large Language Model doesn't solve the problem, it simply introduces an infinite regress.

There is an unfortunately perverse logic to Silicon Valley’s attempts to engineer its way around the human condition. Tech workers are complicated, expensive, and contentious. But so is the target market. A business cannot optimize away the humans on the inside while expecting to meaningfully serve the humans on the outside.