Before I get my teeth into the BBC Trust’s service review (I feel obliged to sooner or later) I enjoyed the collision of cultures (or contrast of cultures I guess) evident in two announcements made last week. In the first one Google announced that the company’s personal health platform thingie Google Health now works with medical records systems at various US hospitals (they started in Cleveland).
Obviously this is just another step in the advancing Googlification of Everything but it’s also interesting because of the way it contrasts with the second announcement, which was from mega-government IT contractor Fujitsu (which used to be ICL) that they’ve got into a terrible pickle and have finally had enough of the vast and (by the sound of it) out-of-control government IT disaster-in-the-making that is the NHS medical records system.
The former (Google and the hospitals) says: use light-weight, consumer-grade tools, put control in the hands of users and not administrators and concentrate on incremental methods, standards and interoperability. The latter (the £12 Billion NHS system) says: build grim, centralised and monolothic systems on a military-industrial scale, exclude open, incremental or agile methods because of your 1950s risk model and hope for the best.
So the big question is: how many of these epic, national-scale contracting disasters do we need to see before we change our approach and try building important national systems by assembling existing code and services in a smart, non-dogmatic way? My guess: at least another ten years. Contractors (BT In particular) are queuing up to replace Fujitsu in the NHS job because the money is just vast. A real web 2.0 type approach to the project would cost 10% of the bid price for the whole thing and would get dozens of executives fired.
In the meantime, I think everyone involved (at the NHS and Fujitsu at least) should read this fascinating presentation about the re-engineering of the BBC’s online identity system from Brendan Quinn and Ben Smith (thanks to Jem Stone for the link). To quote:
1. Each project must have a clear customer and a real benefit
2. Don’t repeat yourself
3. Be as simple as possible
4. Be as open as possible
5. Start simple, then iterate
6. Don’t optimise prematurely…
7. …but build to scale
8. Test often
10. Let it die
If the BBC, which is a pretty big institution—although I’ll acknowledge it’s an order of magnitude smaller than the NHS—can build like this then the NHS could too. I wonder if there is any radical thinking of this sort going on there or is it life in the bunker for all concerned?