The Duty of a Scholar

(simultaneous multithreading) feature they introduced in 2002 could leak information between execution threads via the cache state.

There was a joke at the time that although the National Security Agency had opened up in recent years and started taking their defensive role more seriously, the one aspect of cryptography which they would never discuss publicly was side channel attacks — and this itself constituted a side channel, revealing the importance they attributed to the field. Being active in computer security, and recognizing that the NSA knew a few things about the topic, my immediate priority was to determine whether the information leakage I had identified was sufficient to construct a practical cryptographic side channel. By mid-December I had a proof-of-concept exploit which stole enough bits of an RSA private key for the entire key to be reconstructed, and I turned my attention to the hard part of the problem: Quietly notifying security teams and working together to get the issue fixed before disclosing it publicly.

This became my occupation for the next six months. While my Oxonian contemporaries were making corrections to their theses and applying for jobs, I set those aside (much to the irritation of my supervisors): I had a responsibility, and it took priority over mere career considerations. As it turned out, being unemployed proved useful in early May, when Intel — rather irked by my impending publication — asked Yahoo to order me to withdraw the paper; Yahoo was very happy to reply that, unlike many FreeBSD developers, I was not in fact employed by them and they were not in a position to threaten to fire me. On May 13th, 2005, I presented my paper at <a href=