Deploying non-critical name constraints to the existing community of subcas reduces the risk for everyone and gives apple time to fix their aged tls / certificate stack.
What economic benefit does it provide that outweighs the reduction in security experienced by users of Safari and countless other non-browser clients by allowing Google to be in possession of a private key that can impersonate the identities of arbitrary TLS servers?
Beyond that, Google operates one of the four most important clearinghouses of Internet trust by managing the Chrome certificate policy. It's a little silly to suggest they're pulling a fast one on the Internet; if you don't trust them, you have bigger problems than nameConstraints.
Google did not create the subsidiary CA problem.
Perhaps if you were to mentally s/Google/that hospitality company customer of Trustwave/ perhaps you could see my point more clearly.
RMH chose Google. For him, it was a brilliant rhetorical strategy :-)
So who would be a good example of an organization possessing a 3rd party sub-CA for me to advance my argument? It's much harder because most of them are already relied upon in other critical ways, or are kept secret under confidentiality.
"That hospitality company issued a sub-CA by Trustwave" is the best I can come up with at the moment.
Many libraries built these into regression tests for their chain engines and the suite covers name constraints pretty good.