On 01/10/2015 11:35 am, Peter Lebbing wrote: > On 01/10/15 10:33, Bob Henson wrote: >> There might be a possible exception where there is no individual >> person to meet - the verification signature with software, say. When >> you have downloaded the software from the same, known website for >> some time it might be reasonable to sign the verification key - if a >> tad pointless if it is only really a checksum. > > Well, it doesn't help me at all to know that the developer of said > software indeed has "David Niklas" on his passport. That gives me no > more confidence in the integrity of the software than if he had a > different name. All I need to know is that that piece of software that I > previously trusted has had an update written by the guy or girl I trust, > regardless of his or her name.[1]
That's what I was implying when I described it as a possible exception. > I don't understand "it's only really a checksum". The key property is > that it's signed by the same developer each and every time. A checksum > has very different properties, but I might simply misunderstand you. If the program has been altered the signature will fail, will it not? > >> Someone who I had never previously even heard of once signed my old, >> now revoked key - were that person someone "known" to be nasty, it >> would have degraded my key's value. > > No, it should not degrade the key's value. Unfortunately the key's value > is in the eye of the beholder, and that eye is often not fully aware of > the lack of implications an untrusted signature has. An untrusted > signature has precisely one implication: useless baggage. It neither > increases nor decreases the value of the key it has signed. > > One of the people who's key I've signed at a keysigning party gained a > signature by Adolph Hitler. Enter Godwin's Law. Anyway, he revoked the > key. I can understand that. It just looks bad when someone uses the web > interface of a keyserver to look up his key. But it doesn't degrade his > key in any way other than what is a misperception. Only trusted keys > matter. Untrusted keys can be wholly ignored. Even if they are from the > Führer. > >> The best it could have been is totally meaningless. > > It /is/ totally meaningless. And we should educate users that it is > meaningless. Agreed. But a new user who has yet to be educated would baulk at trusting a key signed by Genghis Khan or Atilla the Hun - however they perceived it, they might well refuse to acknowledge the signature as valid and would certainly not sign it or assign it user trust - that's human nature. Human beings are essentially illogical. :-) _______________________________________________ Gnupg-users mailing list Gnupg-users@gnupg.org http://lists.gnupg.org/mailman/listinfo/gnupg-users