[cap-talk] Horton at HotSec '07: How broadly object/capability?

[Karp, Alan H]

MarcS wrote in response to Shap:
> 
> While I wouldn't put it in all caps, I agree with jonathan that the
> implementation details should remain. The code is exquisitely laid out
> in this paper. Some of the readers, like some of the 
> reviewers, will be
> put off by it. But others will find it a profound and rewarding
> experience. Better to knock the socks off half the audience 
> than to get
> a ho-hum from the entire audience.
> 
It depends on the level of expertise of the audience.  The most
important thing to communicate is why the problem you're solving is
interesting.  Experts in the field will know the answer, so you can give
this part short shrift if you're speaking to such an audience.  I don't
believe that's the case here.  Yes, many of them will have heard of
capabilities, but I doubt that more than a few have ever put any serious
time into thinking about them.  Unless MarkM learns otherwise from
someone on the program committee, I think he should target the great
unwashed.

The next most important thing to get across is the insight you've
brought to the solution.  In other words, explain why nobody solved this
problem years ago.  Here the same explanation can satisfy experts and
non-experts.  For this part, I think getting the concepts across is more
important than providing all the details.  This part of the talk is
where I think precision often gets in the way of clarity.

Finally, you need to convince the audience that what you've done
actually solves the problem.  This part needs detail.  Even here,
though, it's fair to refer to a more complete document for the finer
points.

Jonathan is right.  MarkM will win if he can start an argument that will
extend over the break.  Since he's only got 10 minutes, I doubt that
spending his time walking through the protocol, much less the code, will
generate much heat.  I think the best way to do that is to challenge the
existing approaches and make bold claims about capabilities vs ACLs.  He
can refer to "Demolished" and point out that Horton demolishes another
myth.  I'd list the myths, including the accountability myth.  Then I'd
describe Horton and invite people to see me later to learn about the
others.

That's the talk.  What about the paper?  Protocols are hard.  They give
me a headache.  I need a good reason to put in the effort to understand
a new one.  That makes the motivation critical.  Since we have limited
space, I'd sacrifice details if something has to go.  We can always
refer to the web site for those who are interested.  I don't think our
goal is to get the HotSec attendees to understand the details of Horton.
It's to get them to tell their colleagues that Horton might be something
that they can use.

________________________
Alan Karp
Principal Scientist
Virus Safe Computing Initiative
Hewlett-Packard Laboratories
1501 Page Mill Road
Palo Alto, CA 94304
(650) 857-3967, fax (650) 857-7029
https://ecardfile.com/id/Alan_Karp
http://www.hpl.hp.com/personal/Alan_Karp