We have a solution for a problem, but nobody agrees on what the
problem is.
Could the peole wishing to have TAM accepted, write down what the
problems to be solved are ?
As one of the people who has
wanted to close the group for some time I am less worried about this particular
extension than others.
I have wanted to see the group close
because it seemed the only way to get closure on some of the drafts. Now that
pretty much everything is in queue and looks set to complete I am not really
concerned about that.
It seems to me that what we have here is a
choice between creating a new group to progress TAM independently and doing it
in PKIX. Either way the same set of folk are involved.
The only reason for not doing TAM in PKIX
is to shut the WG down completely which is going to be a problem until we work
out how we are going to manage long term maintenance of cryptographic algorithms
and protocols. The IETF model might work at the IP layer where you want
finality. But application protocols and in particular security protocols need
ongoing maintenance.
The W3C has exactly the same problem, as
does OASIS.
TAM is a three to five year commitment
minimum. If we are going to do useful work here we have to face the fact up
front. Haste makes for long delays. Every time someone says 'we can't think
about that because we are on a tight schedule' I know that its going to take
twice as long as it needs to.
Starting TAM in PKIX does not mean having
to punt on the maintenance question. Once we get TLS, S/MIME etc over the
algorithm agility issues perhaps we decide that both of them plus the bulk of
PKIX go off to some new maintenance activity and split off TAM into a different
group. It gives us a bit more time to think about it.
What I do have concerns about with TAM is
that the use cases and motivating application are not exactly clear to me. I
think that in order to get to a useful conversation we are going to have to
transpose the discussion into a different domain of application where we all
understand the concrete security risks.
I want use cases of the form Alice buys a
network enabled lightswitch and wants to add it to her network.
I have a second reason for wanting to take
that approach. Experience strongly suggests that military grade cryptography is
much easier than consumer. In military applications you can cover up the
deficiencies by ordering people to use it. Sometimes it works and sometimes the
director of the CIA gives the key to his study containing all his top secret
documents to his cleaning lady...