[tor-talk] Why would authorities fall back to "1"?

[Nick Mathewson]

Whoops; my phone used the wrong from address.  Let me try that again.

I tried to say:
> On Oct 27, 2012 11:50 AM, "Sebastian G. <bastik.tor>" <
bastik.tor at googlemail.com> wrote:
> [....]
>
> >
> > Let's say more than 2/3 support "14" and one supports "13". The last one
> > wouldn't use "14" because it does not support it, but why would it fall
> > back to "1"?

To be honest, the whole "falling back" design here is a bit pointless.  If
2/3 of the authorities support and will use method 14, then there is as far
as I can tell no reason for anybody to generate a method-1 consensus, or
indeed any consensus whose method is not 14.  That's because such a
consensus would get signed by at most 1/3 of the authorities, and therefore
wouldn't have enough signatures to get used by anybody.

Now, we could imagine weird exceptions here, involving perversely
fragmented sets of authority versions, and clients who believe that the
authority set is much smaller than it truly is.  But for the most part, I
think falling back at all has little point.

That said, it might be smart to fall back to something other than 1, if we
can find a good reason to fall back.  The reason 1 was specified when I
designed the system was because only version 1 was specified as
must-support.

-- 
Nick