01 — Faster Than You Can Read
Production and comprehension were coupled for fifty years. They have come apart — and the gap between them widens on its own.
Somewhere in your organization this week, more was produced than anyone read. Not skimmed — read, in the sense of followed closely enough to vouch for. Code that shipped, memos that circulated, analyses that moved a decision: generated faster than the people accountable for them could take them in. This is not a lapse in diligence. It is arithmetic, and the arithmetic runs only one way.
For most of the history of building things, two quantities moved together so reliably that no one thought to name them apart: how much you produced, and how much you understood. They were coupled by a plain physical fact. You could not ship what no one had written, and writing something forced a person to understand it — at least well enough to get it written. Output passed through comprehension on its way out the door. The bottleneck was also the guarantee.
That coupling has come apart. The cost of producing an artifact — a function, a contract, a report — has fallen close to nothing, and it is still falling. The cost of understanding one has not moved at all. Comprehension is bounded by how fast a person can read and reason, and no one has been issued a faster mind. So the two quantities, joined for fifty years, now run on two different clocks. One of them accelerated. The other is the clock it always was.
The machine reads at the speed of light and writes at the speed of thought. You still read at the speed of reading.
Picture the two as curves. Volume — the amount your organization is responsible for having gotten right — climbs steeply, and the slope steepens as the tools improve. Comprehension — the amount your organization can actually account for — climbs at the rate it always climbed, which is to say the rate at which competent people read carefully and hold what they read. The two curves start together. Then they don't. And here is the part that gets missed in the excitement over the first curve: the distance between them is not a level. It is a gap that widens on its own, every week, whether or not anyone is watching it, because the thing setting the pace is the faster clock.
For a while the gap is comfortable. The output is good — often very good — and it passes the checks you have. Nothing announces that the share of your own systems anyone could explain, defend, or safely change is quietly shrinking. The work looks the same from the outside. It is being understood less.
None of this is an argument against producing more. The gain is real, and there is no giving it back. It is an argument that the second quantity stopped coming free. For fifty years understanding arrived as a byproduct of production — you got it whether you wanted it or not, because production could not happen without it. That subsidy has ended. Comprehension is now something you either pay for deliberately or go without, and going without does not feel like anything until the day it does.
Which means the gap is the thing to watch, and almost no one is watching it — partly because it is new, and partly because it does not sit on any instrument anyone currently owns. The dashboards measure the fast curve. They were built in a world where the slow curve could be inferred from it, and in that world the inference was sound. It is not sound anymore.
A gap that grows on its own is not a state you are in. It is an account you are running. And accounts accrue.