The Curveball

An NSF-funded focus group of IT educators and practitioners, working in 2003 to redesign how technology programs prepared graduates for real work, invented a term for a deliberate disruption injected into a learning exercise: a curveball. The definition was precise — any event that would simulate the real world, such as modifying requirements, gaining or losing team members, or introducing a new constraint that forces adaptation. Their finding was equally precise: a capstone course that did this once was not enough. The curveball had to be threaded throughout a program, repeatedly, until adaptation became instinct. They were describing, in 2003, the permanent condition of professional work in 2026.

What the curveball was designed to do

Simulate the thing that traditional education most carefully avoids: the situation changing while you are solving it. Curriculum is a controlled environment. Work is not. The gap between them is exactly where judgment lives.

Why the capstone fails

A single project at the end of a program exposes learners to the curveball once, when they are already nearly done and the stakes of real failure are absorbed by the institution. Formation requires repeated exposure — not one disruption managed, but the habit of adaptation built over years.

Anchor: the distinction between practice fields (safe, simulated) and real fields (actual client, actual stakes) from the same research.

What adaptive expertise looks like

Bransford's term for the capacity to handle novel problems, not just routine ones. The practitioner who has only ever followed procedures cannot adapt when the procedure fails. The practitioner who has been interrupted, redirected, and held accountable through disruption can.

AI as the permanent curveball

Every project now includes an AI-shaped variable: the tool that produces something technically plausible and contextually wrong, or that automates the step the practitioner needed to do in order to understand the next one, or that generates output no one can verify without the judgment to know what to look for. AI is not a productivity tool that simplifies work. It is a curveball that changes what work requires.

The formation question

You cannot develop adaptive expertise by delegating adaptation to a tool. The curveball works only if the learner is the one responsible for the response. A practitioner who hands the disruption to AI and accepts the output has not been disrupted. They have been insulated — which is another word for unprepared.