> But it seems to me there would be an improvement:
> simultaneously search several OP clinics, no engaged phones
I thouhgt the point of the internal market was that if the phone
wasn't answered at Addenbrooke's within four rings, you would try
Newmarket, and then Bedford ... so this would give hospitals an
incentive to answer the telephone.
> BTW, still on the security subject, what's your opinion of the encryption
> arrangements that sites like GA Direct offer? I was tickled to see the
> wee broken key (in the bottom left hand corner of Netscape) become whole
The protocol used by netscape - SSL, or secure sockets layer - was full of
holes to start with and could be trivially attacked. They have tightened it
up recently but the NSA limits them to 40 bit keys in the export versions of
their browsers. This means that unless you bought your copy of netscape in
the USA and iullegally exported it from there to the UK, your traffic can be
broken by trying all 2^40 possible keys. A colleague of mine, Piete Brooks,
tried this as an experiment and broke it in 31.8 hours. For details, see
http://www.brute.cl.cam.ac.uk/brute/ at our site.
A more subtle, and to my way of thinking more serious, problem with SSL is
the underlying trust model. This is that all transactions take place between
browsers (which are in effect retail customers) and servers (which are in
effect shops). I don't think that this is expressive enough to support the
sort of web-based medical records that could be useful in healthcare. The
missing features are:
- digitial signatures on web pages so that you can rely on the origin
and contenets of (say) a lab report or referral letter
_ encryption mechanisms that support access control lists, so that
you can build access controls into heterogeneous distributed systems
in a robust and extensible way.
To sum up, netscape's SSL is fine for electronic shopping - which is what it
was designed for. For medicine, it would be better than nothing, but is so
far short of the ideal that I couldn't recommend it as a strategic solution
(even PGP would be much better).
The sort of products on which a medical trust infrastructure could be built
include SDSI (http://theory.lcs.mit.edu/~rivest/publications.html) which was
unveiled at Crypto 96 last week. This has been adopted by Microsoft who aim to
ship a product by the end of the year, so it is probably in all our futures!
Ross
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|