The Last Generation of Explicit Logic

Before computers were machines, they were people — hired to execute, not to reason.

Through the 1930s, 40s, and 50s, rooms full of women sat at desks performing calculations by hand. Mathematicians, many of them — and brilliant ones. The constraint wasn't their capability. It was the job. At NASA, at Los Alamos, at the Bureau of Standards. Katherine Johnson computed orbital trajectories. Dorothy Vaughan managed entire teams of them. They were called computers. That was the job title, and the job description was simple and absolute: receive a specification, execute it precisely, return the result. No judgment, no interpretation, no deviation. They were valued for exactly one quality: the ability to suppress their own reasoning in service of perfect fidelity to the specification.

When the machines arrived, they inherited the job description wholesale. Alan Turing defined the digital computer as a machine intended to carry out any operation a human computer could perform.

Every generation of technology since then added new layers of abstraction, but never changed: logic must be specified to be executed. The machine does exactly what you tell it. Nothing more.

That assumption is so deep most people in technology don't know they hold it.

And the job description just changed for the first time in 70 years.

With a reasoning model, you express what you're trying to achieve and the model works out how to get there. Not executing your logic. Inferring it. The specification becomes optional.

That's not a faster computer. That's a different kind of thing entirely — one that has more in common with the architect who designed the calculation than the computer who executed it.

But here's where it gets complicated. We spent generations learning to be precise, and for good reason. Explicit logic is auditable. Testable. When a specified system fails, you can find exactly where it went wrong. When a reasoning model fails, you have probabilities and educated guesses.

The temptation now is to overfit our reasoning systems the same way we overfitted our specifications: to constrain them so tightly with rules, guardrails, and instructions that we recreate the very rigidity we were trying to escape. To specify the reasoning itself into submission.

That's the trap. Overspecified reasoning doesn't generalize. It just fails in more interesting ways.

Which brings us to the question technology has never had to ask before: which things still deserve to be specified, and why?

Should a medical diagnosis system reason its way to a conclusion, or should every step be specified and auditable? Should a financial system infer intent or execute rules? Should the software that flies a plane think, or obey?

The answer isn't obvious. That's the point.

— Karl