[Cryptography] PGP-Signed Email

[Jason Richards]

grarpamp:
> Jason Richards:
>> is there any evidence or even any suspicions that email
>> sent to a list like this is likely to be or has been maliciously
>> modified in transit, without otherwise being detected, and therefore
>> would benefit from signing?
> 
> ex: people are broadcasting "donate to me" addresses while using
> shitty VPN's, proxies, caches, exits, TLS MITM, services, etc.
> These can and do get rewritten by malactors / bitrot.

Agree that this is possible, however is there any evidence that it
actually occurs, or is even likely. (I do agree with your last point:
threat models are important, and there are some cases where always
signing is important; but it should then be used in conjunction with
other opsec practices to mitigate the attack vectors you list.)

>> I understand that there is a perceived value in signing all emails,
>> thereby establishing a habit which "proves" that an unsigned email
>> did not come from the apparent sender
> 
> Yes, few sig 100% of time, so this proves nothing, even if so.
> And sorting out repudiation assertions / miss-sigs is often not too
> easy.
> 
> A greater value of sig is establishing key history to email address /
> context, of any frequency, so that key somewhat more easily
> trusted... vs cold asking for a key from some address in a "now"
> situation when key is needed, and address may in fact be compromised,
> where key / person may not yet be, etc. Suprisingly low frequency
> needed to establish that.

Is there actually a need to /sign/ each email to lay the groundwork for
demonstrating ownership of a key? Or is is good enough to just include
the ID/fingerprint in each email? If there are three years' worth of
unsigned email listing a specific key followed then by an email signed
by that key, is that somehow less secure than having three years' worth
of email signed with that key.

> No crypto is panacea... requires threat models, context, assertions,
> analysis, proofs, etc.

Indeed! I agree completely with this statement.

J