Re: [perpass] A reminder, the Network is the Enemy...

[Phillip Hallam-Baker <[e-mail]>]

On Fri, Dec 6, 2013 at 11:12 AM, David Conrad <[e-mail]> wrote:

> On Dec 5, 2013, at 7:27 AM, Phillip Hallam-Baker <[e-mail]> wrote:
> > A better approach is to design the system so that it takes a defection
> by more than one party. Instead of relying on just the ICANN root KSK
> require a TLD to be signed by three out of five trusted national cryptolabs.
>
> Trusted by whom? E.g., trusted like NIST now? (No disrespect of folks at
> NIST intended: just observing some may no longer view them as trustable)
>

I mean trusted in the technical sense of relying on them (albeit to a
qualified degree).

And it is important to note that trusted does not mean the same thing as
trustworthy, a point I raised at one of the early trusted computing group
efforts (only Microsoft seemed to take note or maybe they came up with the
understanding independently).

If you have a three out of five scheme one should choose five labs that are
very likely to collude. The UK, US, Australia, Canada and New Zealand would
not be a very good choice. The UK, France, Russia, India and Brazil would
be a rather better one. Of maybe you would want to have the EFF or the like
in there (if they could set up a secure facility and maintain it at
acceptable cost).



> I personally believe a better approach is to make the operation of the
> system extremely public and documented such that it doesn't matter who is
> involved since the risk would be too high that attempts at compromise would
> be observed. This is what ICANN tried to do with the root KSK (one can
> argue whether they succeeded).
>

These do not need to be exclusive, nor was that my proposal. I would expect
any national cryptolab to follow the established industry practices.


-- 
Website: http://hallambaker.com/