Now that the PGP key management "bug" is public, I'd like to comment on some source code issues and follies. The source for versions in question (starting from 5.*) has been available for more than two years. While many crypto experts intensely bullshit about the importance of the source code to counter "security through obscurity", it appears than none really looked at the sources closely. The net effect is that for several years now it is possible to add encrypting keys to ones whose owners are stupid enough to publish them on the public key servers. And it is not impossible that several more bugs. equally serious, are waiting to be discovered. Availability of the source code created the false sense of security. [I think that it makes a very basic sense not to disclose any more info than absolutely necessary. A "public" key part does not really have to be public at all. Transporting it over secure channel adds positive authentication to PK (I never grasped the idea that I should trust someone I do not know to establish secure communication ... but that is another issue.)] What will the next bug be ? Poor randomness in session keys, maybe. Who ran Diehard on PGP-generated session keys ? I know that I did not. Spooks will wellcome any PGP-bashing, of course, as it diminishes the trust and, ultimately, the use. But it appears that we have been sold the false premise that source code by itself is somehow insurance against the snake oil. Obviously it is not. Source code inspected and walked through by several professionals - yes. But how many experts are willing to put hundreds of hours to do a public service, when they can do something else and charge for it. It would be interesting to see how many actual man-hours were burned inspecting PGP 5.* and later sources before the latest discovery. pick one: a) 5 b) 50 c) 500 I spent about 15 minutes on PGP 6.2, looking primarily at PGPNet. 4.75 hrs more is needed for a) to be the answer.
