Bit of an open ended question that one Adrian.
For me the answer lies in the domain of network structure - this can either be network for knowledge (and mindlines instead of guidelines for the BMJ readers out there), or physical networks of equipment.
People have already mentioned Strogatz and his books but his "Exploring Complex Networks, Nature 2001,410,268-76" is accesible in places.
My simplistic take on it:
Effective networks have to have built in redundancy, centralised networks build in that redundancy separately and it's about switching new for old in the case of failure of one part of the system (as in a football team with one manager and 11 players on the field, you can use subs from the bench if one player is injured, but they can't do anything in the meantime, and if the manager fails you need to get a replacement and until you do the whole team is stuck). Such systems are either inefficient or get pared down to a minimum reserve capacity which means they are vulnerable to failure.
Scale free networks (such as the world wide web and power grids) have the ability to shift work around and each element will have some redundancy most of the time. These generally cope very well with intermittent failure (but just occasionally go completely belly up). Because they work on nested nodes they are sensitive to targeted attack ( a fact that internet denial of service attacks exploit)
Too much healthcare computing (IMO) is geared around a centralised network and making it attack-proof. While not wishing to argue that the NHS data repository should resemble SETI, I think there may be some useful insights here.
Chris
--
Dr Chris Burton,
GP, Sanquhar Health Centre, Dumfriesshire. DG4 6BT
CSO Research Training Fellow, University of Edinburgh
Practice 01659 50221 Mobile 07736 020362
http://www.medicine21.com
04/11/2004
On Thu, 4 Nov 2004 15:56:21 +0000, Adrian Midgley wrote:
> I'm attending one of the NPfIT best practice series' of meetings on
> workflow etc.
>
> How does what we know about complexity affect what a computer
> system should do and how we should design them?
>
>
> --
> Dr Adrian Midgley from the office using Open Source Homefield
> Surgery because it is better Exeter UK
> www.homefieldsurgery.nhs.uk 01392 214151 www.defoam.net
> www.oshca.org
>
>
> Homefield Surgery Heavitree Exeter 01392 214151
>
> ______________________________________________________________________
> __ Doctors.net.uk e-mail protects you from viruses and unsolicited
> messages
> ______________________________________________________________________
> _
|