I wanted to provide my thoughts on this.
First, in the interest of full disclosure, Russ and I conferred on stage about
the questions, and he had my OK to go ahead.
Now, of course the questions were at a high level, and with full hindsight they
could have been more neutrally formulated. At the end of the day, they are what
they are. And should be given the value they deserve.
I'd actually like to argue that the IETF position on this topic is something
bigger, something where the plenary discussion and hums played a supporting
role but they are not the sole determination. Here's my take-away from this
week:
"The IETF considers pervasive-monitoring as a security issue and is willing to
work to address it."
Nothing more, nothing less. Most working groups that I went to were addressing
this topic in one way or the other, going through application by application,
doing careful work to understand what options we have to improve security, and
weighing the various trade-offs in different designs. The proof of the pudding
is in the eating. "We need to address it" vs. "We are putting in the cycles to
address it". When I look at the discussions throughout the week, it is very
clear to me that we are putting in the cycles.
As Carsten said:
As always, hard work follows, and the devil is in the details. But that
doesn’t take away from the unanimity.
And indeed there are a lot of details and trade-offs to worry about.
Opportunistic encryption, for instance, has been discussed at length this week
and the variants and trade-offs are far from clear.
I think the next steps are what is important. And this is a long term effort.
Here are some of the things we should be doing:
- work on the general guidance in this area ("consider it as an attack",
"recommended ways to apply opportunistic encryption", "threat model changes",
...)
- work on the specific protocols and application areas (http, xmpp, etc)
Jari