 |
May 14, 2013, 06:35:57 PM |
|
Every talk will be widely witnessed and videotaped so we can get some reasonably good security by simply putting out PGP fingerprints in our slides. Yeah, some fancy attacker could change the videos after the fact, but the talks themselves will have wide audiences and a lot of opportunities for fraud to be discovered. That means it'd also be reasonable for people to sign those keys too if you are present and are convinced you aren't looking at some impostor. (of course, presenters, check that your PGP fingerprints are correct...)
Remember that PGP depends on the web-of-trust. No single measure in a web-of-trust is needs to be absolutely perfect; it's the sum of the verifications that matter. I don't think it matters much if you have, say, seen Jeff Garzik's drivers license as much as it matters that you have seen him in a public place with dozens of witnesses that would recognize him and call out any attempt at fraud.
Secondly remember that many of us are working on software where an attacker can steal from huge numbers of users at once if they manage to sneak some wallet stealing code in. We need better code signing practices, but they don't help without some way of being sure the keys signing the code are valid. SSL and certificate authorities have advantages, and so does the PGP WoT, so use both.
FWIW I take this stuff pretty seriously myself. I generated my key securely in the first place, I use a hardware smartcard to store my PGP key, and I keep the master signing key - the key with the ability to sign other keys - separate from my day-to-day signing subkeys. I also PGP sign emails regularly, which means anyone can get a decent idea of if they have the right key by looking at bitcoin-development mailing list archives and checking the signatures. A truly dedicated attacker could probably sign something without my knowledge, but I've certainly raised the bar.
|
