Segal introduced the diagnostic in You On AI as a practical tool for distinguishing flow from productive addiction during AI-augmented work sessions. The behavioral markers he identified — generative versus demand-clearing question patterns — map onto what cognitive psychology calls metacognition, the capacity to monitor and evaluate one's own cognitive processes.
Pang's contribution is to connect the diagnostic to its neural substrate. The self-referential processing that produces metacognition is a DMN function. It requires the same unfocused attention that creative association requires, and it is suppressed by the same continuous engagement that suppresses creative association. This produces a structural problem: the cognitive system that would allow the builder to recognize that she should stop working is the same system that continuous work disables.
The practical implication is that the diagnostic cannot be relied on as the sole structural defense against productive addiction. It works early in sessions, when cognitive resources are adequate. It degrades as sessions extend. By the point when stopping would be most valuable, the diagnostic has often failed. This is why Pang argues for pre-committed structural boundaries — the four-hour rule, the fixed end-of-session walk — rather than relying on in-session self-assessment.
In the contemplative computing framework, the diagnostic could be externalized into the tool itself. An AI that tracked prompt patterns — detecting the shift from divergent to convergent, from exploratory to optimizing, from novel vocabulary to repetitive patterns — could surface signal quality information the fatigued user can no longer reliably generate. This would make the diagnostic robust to the cognitive fatigue that compromises its self-assessment form.
Edo Segal introduced the generative-versus-demand-clearing distinction in You On AI (2026). Pang synthesized it with the metacognition literature and DMN research.
Question patterns as proxy. The shape of questions reveals the cognitive state of the questioner more reliably than felt experience does.
DMN-dependent. The diagnostic depends on the same self-referential processing that continuous work suppresses.
Self-defeat. The diagnostic fails precisely when it would be most useful, as extended work degrades the metacognition it depends on.
Externalization opportunity. Tool-level tracking could make the diagnostic robust to fatigue.