[noise] Reworking PSK usage

[Jason A. Donenfeld]

On Sun, May 7, 2017 at 11:52 PM, Alex <alex at centromere.net> wrote:
>> On Sun, May 7, 2017 at 10:41 PM, Alex <alex at centromere.net> wrote:
>> > How would that generalize for pskN?
>>
>> psk0: first token after e of first message.
>> pskN for N>0: last token of Nth message.
>
> Does this imply that a psk token located anywhere else results in an
> invalid pattern?

No, it just means it's unnamed. Trevor's original email and a
follow-up discussed this in depth, if I recall correctly. psk0 is for
people who want the psk protecting _everything_. Otherwise, for people
who need it to protect slightly less for various reasons, the most
commonly useful places for it will be at the end of a particular
message. That doesn't mean other placements of the psk token are
invalid. It just means that they're [presently] unnamed, like so many
other potential Noise combinations.