Taste and Connection

I love the concept from baseball statistics of Wins Above Replacement. It doesn’t ask how good you are in the abstract — it asks how much better you are than the typical player available in the talent market. Every team has a bench full of replacement-level talent: competent, available, cheap. WAR is framework for evaluating the value of a player - how much are they worth paying for compared to someone who will work for (closer to) the minimum.

AI is producing replacement-level work across white-collar disciplines — writing, coding, design, financial analysis, research. Output that used to require years of specialized training arrives in seconds for the cost of an API call. If your value is “I can do this competently,” you have serious competition now.

This is a pattern we’ve seen many times before in history. It’s led to turbulence, unrest, and to extraordinary gains in quality of life for most people.

The Capacity Unlock

“Computer” used to be a job title. Before electronic computing, orbital mechanics meant rooms of people with pencils and paper, grinding through calculations by hand. Space missions were bespoke because the math was bespoke — each launch constrained by how many human hours you could throw at the trajectory problem. Remove that constraint, and the number of objects launched into space went from dozens per year to thousands. We didn’t get more ambitious. Ambition was no longer gated by arithmetic.

The same pattern is unfolding in software. Paul Ford recently argued that there are millions of software products that should exist but don’t — dashboards, trackers, internal tools that would save hours of frustrated clicking — because the people who need them could never find the budget or the developers. When AI makes it possible for a non-technical person to iteratively describe their problem and design a proof of concept or actual working application that solves their unique problem, that backlog starts to clear.

The pattern extends beyond code. Research in science and business — literature reviews, financial modeling, competitive analysis, experimental design — has always been constrained by how many skilled humans you could put on the problem. AI doesn’t replace the thinking, but it closes the gap between question and first answer. Problems that sat in the queue because nobody had the capacity are finally getting worked.

Each is the same story: remove a capacity constraint, unlock problems that were too expensive to attempt.

The Selection Problem

Small teams and individuals are accelerating in ways that weren’t possible two years ago. A “bad first draft” of almost anything — a financial model, a clickable prototype, a scaffolded application, a research brief — can arrive in minutes. The barrier between “I have an idea” and “here’s something you can react to” is collapsing.

But abundance creates a new tension. When anyone can generate a thousand variations of anything, the bottleneck shifts from production to selection. Someone has to look at those outputs and decide which one is actually good. Someone has to point the machine in a direction worth pursuing.

That someone is exercising taste.

Taste and Connection

Taste — the ability to discern what’s worth doing, what “good” looks like, and what standard to hold — becomes more valuable as production gets cheaper. When execution is abundant, judgment is scarce.

This is where the human element isn’t just preserved — it’s amplified. Kevin Kelly described it as 1,000 True Fans: you don’t need to reach everyone. You need to find and connect with people who trust your taste — who value how you curate and create, who follow you because of what you are uniquely capable of creating and building. There is a ton of value in being able to understand and describe a problem, to evaluate potential solutions, and then to share your work (curation or creation) with others who value your point of view. There is value in being able to create custom work to solve narrow problems.

I nerd out learning about engineering, building, and designing our lived world from great communicators, who earn money making educational materials from people like me. I can help my children learn about saving and spending with fully custom software. I can give my wife a sense of calm and control at the end of her work day with a fully custom dashboard to see and manage her personal life at a glance in our kitchen.

In a world where the cost of producing something digital is falling, the dominant strategy is building things that deliver value, and that send a clear signal about what you value. That how you build and deepen personal and professional connections.

Unsolicited Advice

I rarely give unsolicited advice, but these are rare times. There has never been a better time to start creating and putting work in front of people who will value it. The tools to make a bad first draft of almost anything digital are accessible to anyone with a laptop and curiosity. The cost of testing an idea — in code, in writing, in design, in business — has never been lower.

Build your portfolio. Build your audience. Build the skill set that compounds as the tools improve.

The capacity constraints that gated who could create and who could solve are falling. We’re going to clear a lot of problems that have been waiting for someone to get to them. We’ll make some new ones along the way.

Postscript

As I went to hit publish on this, I listened to the latest Ezra Klein show podcast with a co-founder of Anthropic which covers similar themes, but also how to best work with AI as it exists today - it’s a great listen.

---