What they did
King located site-level signals in the leaked modules and lined them up against specific Google denials. He documented siteAuthority, host trust levels from NOT_TRUSTED to FULLY_TRUSTED, and the hostAge attribute.
What they found
A siteAuthority metric is stored as a compressed quality signal referenced in a Q* ranking system, and hostAge carries a note about sandboxing fresh spam in serving time. Both map onto things Google spokespeople had denied in public.
How much weight to give it
The documents confirm these fields are named and defined. They do not show how much siteAuthority counts, whether it drives current ranking, or that hostAge produces a live sandbox effect. King inferred a contradiction with Google's public messaging.
What it doesn't prove
This is field existence set against public denials, not proof of active use or of equivalence to the denied concepts. A stored siteAuthority is not necessarily the domain-authority score Google denied, and Google contested the interpretation, so this reading is genuinely disputed.