On Fri, Sep 2, 2016 at 11:12 PM, John Levine <jo...@taugh.com> wrote: >>But I'm not sure what native would look like. After Lavabit, would the >>type of folks who use pgp actually trust our implementation if they >>couldn't see it and verify it? > > In my experience there are two kinds of PGP users. One is the hard > core who go to key signing parties with their passports in their > pockets. The other is the casual ones who get keys from keyservers > when they send moderately touchy stuff. > > The latter group would probably be OK with your implementation. The > others would not, so they'd have to use POP/IMAP/SUBMIT and do the > crypto at home. > >>Also, the spam problem becomes challenging in that environment... > > For the latter group, you can ask them if it's OK to use their keys > for spam filtering and they'll probably say it is. For the former > group, it's a problem. Of course, they're only likely to exchange > encrypted mail with a tiny set of friends, so perhaps you could say > that the sender's key isn't in someone's address book, rate limit it > down to one or two messages per day. That gives an opportunity for > initial contact, at least until the spammers figure out that their > botnets have plenty of CPU to invent a new identity and a new key for > every spam.
In addition to what John said, I think a very useful first step component would be for Google (Microsoft too!) to run an internal PGP keyserver (if you don't already have one) and then use it reject signed msgs that fail a basic sig test. You don't need anyone's private key, and you could sync your keyserver the same way all the other keyservers do. This would go a long way towards true message integrity. -Jim P. _______________________________________________ mailop mailing list mailop@mailop.org https://chilli.nosignal.org/cgi-bin/mailman/listinfo/mailop
