On 10/05/2009 05:45 AM, Hauke Laging wrote: > Am Montag 05 Oktober 2009 schrieb Daniel Kahn Gillmor: >> 0) you only talk about digesting the e-mail part of the address. what >> about the human-specific name? Would this need to be digested also? >> Why or Why not? > > From a technical point of view that nearly does not matter so one could > leave this up to each user. i don't think you can leave this up to the user effectively, unless you want clients to need to query for a potentially-lengthy series of different digested addresses. If something like this is going to work, you need to have one widely-adopted digest, and to change it *very* infrequently, so clients would never need to try more than two digests in a search. > But there is a second argument, privacy. This is valid for the email > hashing, too. In an ideal world the key server data could not be used for > anything their "owner" does not want it to be used for. If you want to advocate for a standard for digested User IDs, I think the general privacy argument is a much stronger one than the spam argument. For example, consider a Bordurian dissident who is being persecuted by her government. She may very well want to exchange secure, private e-mails without people being able to search for her by name, and without publically binding her name to the e-mail address she's using (though she may be willing to confide that information to her associates). she may also want to take advantage of the revocation, certification, and expiration features of the WoT (all features which are enhanced by using public keyservers) without exposing the names of her personal contacts to direct review or searching on the public keyservers. Beyond names and e-mail addresses, there are other things that could be stored in an OpenPGP User ID which people might not want to be publicly enumerable or searchable, though they might want the associated material to be found if someone else already knows the relevant name. For example, i work on the Monkeysphere project, which provides a PKI for OpenSSH servers using the WoT and public keyservers as a PKI (for certification, revocation, etc). We've had some potential users express concern about using it because they don't want their infrastructure to be mappable or enumerable from the public keyservers. https://labs.riseup.net/code/issues/show/1181 Using a privately-held keyserver could solve this problem for a certain class of these users (those who maintain a "walled-garden" network, for example), but this requires an authentication/authorization policy for access to the private keyserver, it creates another point of failure in the system, and it doesn't work well at all for people who prefer a federated approach over a "walled garden" approach. Allowing these folks to take advantage of a broader keyserver network for distributing their data might make this use case more feasible. > I do not know the openpgp key format. Would it be easily possible to add > the signed information whether the UID of this key may or must not be > uploaded to a key server in cleartext, at best distinguishing between name > and email? OpenPGP User IDs are defined as UTF-8-encoded textual data, and can in principle range from 0B to 4GB (though in practice i've never seen one even as long as 1KB). There's no reason that they couldn't be used to store the digest encoded as a text string. If you wanted that to work smoothly, though, you probably want to have a single unique form so that they could be found. > One small additional point: This hashing approach would be used for all > published keys, not only for key servers. I guess that most PGP users have > their public key on their web site. right, otherwise someone could download the key from a web site and post it to the keyservers. Note, however, that most people who publish their keys to their web site have already provided an easy way for people to come up with their name and e-mail address. Those people's contact info has probably already been transferred into a mailing list that has been sold to a spammer, so the cat is probably out of the bag for them (as it is for everyone who participates in public mailing list discussions). A proposal like this would really only be useful for people who prefer to avoid having their User ID publicly searchable in any way. But such people exist, and it would be good to allow them to use the nicer features of the WoT as well without divulging their information directly in the User ID packets. --dkg
Attachment:
signature.asc
Description: OpenPGP digital signature