Re: SSL/TLS everywhere fail

On Fri, Dec 4, 2015 at 4:10 PM, Poul-Henning Kamp <phk@phk.freebsd.dk>
wrote:

> --------
> In message <245746A1-A574-4D19-A8AA-8AC5270D0D8F@lukasa.co.uk>, Cory
> Benfield w
> rites:
>
> My point is that TLS anywhere made this *possible* for them.
>
>
​And were a majority of traffic in plaintext​ this would have been
impossible?  Harder?  I think maybe easier, instead.   Less expensive, to
boot.  Sure, passive analysis is easier yet, but MiTM of plaintext is
easier than MiTM of TLS on its face.


> The correct response would have been to roll out more or less
> *exactly* what the encryption draft contains, along with a wide
> number of diverse key-management schedules.
>
>
​There may also be more than one correct response.​  Doing more than one
thing may be required to get to a desired state, after all.


That way the Kazakh government, no matter how much they desired it,
> would not be able to do a MiTM on their entire country, because they
> would not be able to get any tool from anywhere to implement it.
>
> Their political choice would then be reduced to:
>
>    A) Pass traffic, collecting only the "envelope",
>
> or
>
>    B) Block all traffic that cannot be inspected.
>
> Even in Kazakstan B) would be politically suicide, so the result would
> *at worst* be A).
>
> A) Would have been a *much* better situation than what is now happening.
>
> Now that they *have* implemented total MiTM, we fight the uphill
> battle where they can (and probably will) block any other kind of
> encrypted traffic than TLS, so rolling out our stronger weapon just
> got impossble.
>
>
​And your evidence for this is what, exactly? ​Combined with TLS, how does
your stronger weapon get weaker?

Ted