[tptacek]

Your "1. Going through the DNS process" starts with records the USG controls! Yes, "the best they could do" would be to defeat the entire system. What do you think happens in a TLS MITM? They USG isn't trying to sign your keys and publish your addresses!

This is not, of course, the only problem with DNSSEC. It's also an archaic 1990s cryptosystem built around 1024-bit PKCS1v15 RSA, which by default makes every DNS record in the system public, trivially dramatically amplifies DNS traffic, and does all this without actually securing DNS lookups from browsers, which still run the old insecure DNS protocol to talk to DNSSEC-enabled caches.

It's a silly system, has been since the USG paid TIS to design it in the 1990s, is nearing two decades delayed, and isn't going to happen. Look at what Chris Palmer from Google has to say about it. Whatever the opposite of "betting on it" is, that's what Chromium is doing with DNSSEC. We should get to work designing a modern alternative.