The Calorie Theory of Effort
AI is breaking the old signal that effort means visible time. What remains is chosen attention and judgment.
A human body at rest burns about 100 watts. The same as an old incandescent bulb. A full day of you, sitting and thinking and walking around, runs about 2,000 calories, or roughly 2.3 kilowatt-hours. You could run about 7,000 chatbot queries on what your body spent today just being alive.
We don’t think about ourselves that way. We never have. And the mismatch, between how we measure ourselves and how we measure machines, is going to define how we feel about work for the next decade.
What we’ve always meant by “effort”
The waiter who remembers your name didn’t burn extra calories doing it. He spent attention. He chose to direct a finite cognitive resource at you instead of someone else, and you noticed. That’s the part you tipped for.
We talk about effort in terms of energy, but that’s not really what we’re measuring. What we evolved to detect is chosen time spent visibly on us. Calories were a proxy for time. Time was a proxy for attention. And attention was the finite thing we were actually trying to see.
For a long time, those three currencies moved together. Everything that took energy also took time, and everything that took time was a choice not to do something else.
Labor, in its original form, was effort × time. The blacksmith stayed at the forge for ten hours. The farmer plowed for twelve. We could see it. We could count it. We built our entire economic system around the idea that what you did was worth what it cost you in hours of life.
That stack is breaking.
Why AI feels lazy
When someone uses AI to write a birthday note, you can feel it. Not because the words are bad. Often they’re better than what your friend would have written. You feel it because the time is gone. The choosing is gone. There’s no fingerprint of someone redirecting their finite hours toward you.
The “lazy” reflex is a heuristic that worked for millennia under one condition: producing something required time. AI broke that condition.
The reflex still fires. The signal is gone.
This is not a moral failing of either party. It’s a calibration problem. Our intuitions are tuned for a world where output and effort were locked together. They’ve come unlocked, and we don’t yet have a replacement signal.
What companies already know
Most companies already live in this contradiction. They pay you in time and select you on impact.
This isn’t entirely new. Manual laborers used to call knowledge work “not real work.” A carpenter could see the cabinet. He couldn’t see the strategy memo. The cabinet was effort. The memo was air.
The salary, the work week, the eight-hour day, the vesting cliff. All denominated in time. That’s a vestige of labor law and industrial scheduling, and it’s not going anywhere quickly. But inside most healthy companies, the actual selection function is impact. The people who tend to get promoted are not the ones who clock the most hours. They’re the ones whose work moved something. The people who get cut in a downturn are usually the ones whose hours produced the least.
Smart employees notice the gap and play the second game. Most don’t. Most take the salary literally, treat their job as time-for-money, and wonder why they get passed over by people who appear to be doing the same job.
The contradiction is going to get sharper. As production gets cheaper and impact gets more important, the gap between what we measure and what we value will widen. The people who can read the difference will pull away from the people who can’t.
Judgment is the new labor
When everyone can produce anything in seconds, the question stops being “did you make it?” and becomes “should this have been made?”
That’s pure judgment. It’s even more invisible than effort.
I wrote in The Full-Stack Employee about how agents collapse the gap between having the judgment and producing the work. The implication for capability is obvious. The implication for valuation is harder. When the production part is trivial, the only thing left to value is the judgment part. And judgment doesn’t take hours. It doesn’t sweat. It doesn’t show.
A great writer with AI is a hundred times more productive. A mediocre writer with AI is just a faster mediocre writer. Tools no longer compress the gap between high-judgment and low-judgment people. They amplify it. The gap will be visible in outcomes. It won’t be visible in effort.
For knowledge workers, this is the shift.
Finishing in twenty minutes what used to take a week is not a confession. The week was never the value. The choice and the judgment were the value. The week was just the wrapping.
The same is true when we evaluate other people’s work. The important question is not how much time it consumed. It’s what their judgment moved, what they chose not to do, and what the output is worth.
And if you’re producing something people are supposed to feel cared for by, a birthday note, a recommendation, a custom thing for someone you love, the same rule applies in a more personal register. Care is judgment pointed at a person. It says: I chose you, I noticed this about you, I would have done this for you whether or not it was easy. The medium of care is no longer time by itself. It’s chosen attention made visible.
The waiter remembering your name was never about calories. It was about chosen attention. The intuition was right. The proxy was just one we could see.
Now the proxy is breaking, and we have to learn to see what it was pointing at.