Tech CEOs are Violating a Fundamental Law

By Damon Kiesow | Working Systems | June 10, 2026

The primary goal of any successful product is the minimization of friction—essentially stripping away the time, effort, and financial cost for the end-user.

The Principle of Conservation

In the realm of system design, this phenomenon is governed by Tesler’s Law. Formulated by Larry Tesler at Xerox PARC during the 1980s, this principle suggests that complexity is a constant.

The Axiom: Every single product contains a core activity that possesses a non-negotiable, irreducible amount of effort. This inherent complexity cannot be deleted; it can only be shifted.

Mathematically, we can view this as: $\text{Complexity}_{\text{Total}} = \text{Complexity}_{\text{User}} + \text{Complexity}_{\text{System}}$

The Great Shift

In the modern era, tasks that once required significant effort have become one-click experiences.

Action	Former Friction (User)	Current Friction (System)
Hailing a Cab	Standing on street/calling dispatch	Complex GPS & matching algorithms
Buying a Book	Visiting a store/browsing aisles	Massive logistics & warehouse networks
Finding Info	Manual library research	Hyper-scaled data centers & indexing

Consequently, the obsession with removing consumer friction has become the foundational ontology of Silicon Valley—the primary lens through which they interpret the world.

The Generative AI Delusion

Now, the emergence of Generative AI is feeding the egos of tech executives. They have begun to believe they can apply this "one-click" optimization to the very structure of their companies by eliminating automating their staff.

The "lone genius" founder now imagines a world where they can:

• Conceive an idea in the morning.
• Write a few GenAI prompts.
• Deploy a full app by the afternoon.

They ask: Why pay for UX designers, software engineers, or product managers?

The Category Error

This mindset is a delusion based on a fundamental category error: confusing the software/service (the output) with the socio-technical system (the business and its users).

def company_optimization(workforce):
    if workforce == "Humans":
        return "Meaningful Value"
    elif workforce == "LLM_Prompts":
        return "Infinite Regress / Hallucination"

This approach fails for two primary reasons:

1. The Discovery Problem: The most grueling part of development is identifying human needs. Determining the what, why, and for whom cannot be automated; it is only uncovered through the friction of human deliberation.
2. The Conservation Problem: Because complexity is conserved, offloading the intellectual heavy lifting to a Large Language Model doesn't remove the burden—it simply creates an infinite regress of oversight and correction.

The Bottom Line: There is a perverse logic in trying to engineer a way around the human condition. A company cannot optimize away the humans on the inside while expecting to provide genuine value to the humans on the outside.