Not every pro-p group is the absolute Galois group of a field. The question is how to tell which ones are. The standard toolkit consists of cohomological tests: a group must satisfy certain relations among its cohomology classes to be realizable as a Galois group. The Massey vanishing property (triple Massey products vanish), quadraticity (all relations come from degree-2 cohomology), and the structure of the cup product — each filters out some groups that can't be Galois.
These tests have a blind spot.
The construction produces Demushkin groups — pro-p groups with particularly clean cohomological structure — that pass every known test. They satisfy triple Massey vanishing. They are quadratic. Their cup-product structure matches what absolute Galois groups must have. By every available criterion, they look like Galois groups.
They aren't.
The proof of non-realizability uses a different kind of obstruction — one that doesn't show up in the standard cohomological tests. The specific nature of the obstruction reveals a gap in the diagnostic toolkit: the tests we have are necessary conditions for being Galois, but they are not sufficient. The groups satisfy all the necessary conditions and still fail the actual condition.
The implication is that characterizing absolute Galois groups requires invariants beyond those currently used. The cohomological approach, which has driven much of modern Galois theory, is incomplete. There exist “Galois-like” groups — sharing every testable property with genuine Galois groups — that are algebraically distinct from any Galois group over any field.
The toolkit recognizes everything that Galois groups have in common with each other. What it misses is whatever separates Galois groups from their cohomological twins. That separating invariant is the open problem.