Nick Szabo  🔑

SSL is woefully inadequate against large classes of MITM. Chain from PGP-signed git commit (etc.) to browser much more secure.

@jgarzik To make PGP better, need also a reliance statement: ie you rely on PGP to say what, exactly? Not MITM'd? Not malware? Not private?

@iang_fc @jgarzik This signing, SSL or PGP, would provide integrity of the channel from the signed code to browser -- no more than that.

@iang_fc @jgarzik Previous schemes "failed" because we added expectations crypto doesn't actually provide, e.g. "identity" of source.

@NickSzabo4 @jgarzik users are rational, expect to rely on claims from "crypto". Without reliance, what's the point? What are we saying?

@iang_fc @jgarzik SSH is good model. Leave judging of "identity" & non-crypto integrity, which is dependent on local knowledge, up to users.

@NickSzabo4 @jgarzik If, app signer as a brand, then we need a cryptographic signature over a concept. That I grok cf Ricardian contracts ;)

@iang_fc @jgarzik Crypto protects syntax not wet semantics. Doesn't prevent trademark spoofing or phishing.