Question 1

Why does McCann argue that post-hoc governance layers (like content filters or moderation systems) are structurally doomed to fail?

Accepted Answer

McCann grounds his argument in Rice's theorem (1953), which proves that no behavioral layer added on top of a Turing-complete system can fully govern its effects—the gap between what a system can do and what governance covers is mathematically undecidable. This reframes AI governance from a fixable implementation problem into an architectural one: bolting oversight onto an expressive system will always leave edges uncontrolled, making compliance dashboards and safety theater inevitable when expressiveness and governance boundaries are defined independently.

Question 2

What is 'coterminous governance' and why should product directors care about it?

Accepted Answer

Coterminous governance is McCann's proposed solution: separating computation from effect at design time rather than layering oversight afterward. For product directors, this translates a formal theorem into a testable architectural criterion—it's a shift from accepting that every moderation system will fail at the edges, to designing systems where safety constraints and computational expressiveness are unified from the start, backed by mechanized proofs rather than policy frameworks.

Question 3

How does this work connect to decisions about where to invest in AI safety and compliance?

Accepted Answer

McCann's analysis suggests that significant post-deployment safety investments (content filters, moderation layers, compliance dashboards) are fighting a theoretically unwinnable battle. This redirects product strategy toward architectural decisions made before systems go live—where computation and governance boundaries can be coterminous—rather than assuming better oversight tools or organizational processes will solve structural problems that formal logic guarantees will persist.