Jim Lawson wrote: > Suppose in 2005 Kate gets her key signed by Sam at an overseas > keysigning party. Kate's key has an expiration date of 2015. In 2009 > it becomes clear that she needs to refresh her key in order to move to > the new digest algorithm. So, she generates a new key, expiration 2019, > and signs it with her old key. She then sends the new key to Sam, who > she will probably not meet again in person any time soon, and asks him > to sign the new key and send the signature back to her. > > Is this bad form? Should Sam sign the new key, based upon the trusted > signature from Kate's old key, even though he is not meeting Kate and > re-checking her ID and fingerprint? Or, should he insist on an > in-person meeting (or some other form of authentication) ?
Good question. Let's look at the big picture. Key signing builds a Small World network--a network of dense clusters. The Sam-Kate connection is valuable because it binds the "Vague" cluster to the "overseas keysigning party" cluster. I think we are transitioning now while it is still reasonable to trust SHA-1 based keys, and so it is still reasonable to sign Kate's new key. On the other hand, the network can heal if Sam-Kate gets lost. It is much more important that we establish strong connections with our new keys among our various "vague", "family", "friends", and "work" networks, and to keep adding out-of-network key signatures whenever we get the opportunities. I'm still signing email, including this message, with my 1024-bit DSA key. Instead of just quietly jumping to a new key, I'm going through the transition process "live" with you guys so that we can get our corner of the world moving on the transition. CONFESSION: I failed to answer the question. I jujutsued it into restrained advocacy. I say restrained, because I could have easily used the phrase "long tail" :). -- Anthony Carrico
signature.asc
Description: OpenPGP digital signature
