Monday, December 19, 2005
The government is a different country ...
... They do things differently there. Interesting comment on my post below about integration: "The challenge you have not tackled is how to spend the money wisely - given the government's dismal track record of successful IT delivery, especially in the eGov space, how do you think they can successfully spend that amount of money and still deliver benefit/value? If you think it is possible, just look at the mess over in the Dept of Health - give them £10bn and they still can't get GPs to use the technology!" You're right, I didn't tackle it. It's not an easy topic to tackle in a few lines, but that's no reason to duck it. Here's a few points to provoke a debate: - IT projects on a large scale are fraught with difficulty and risk, whether public or private. For instance, Sainsbury's supply chain system (half a billion wasted?), the FBI's Virtual Case File project ($200 million wasted?) and so on. I could list a few dozen. If I tried to list the most successful IT projects over £100 million in cost, I think I would struggle to find more than a handful. So perhaps lesson one is, don't try to do anything that huge - you will get it wrong, it will cost more than you thought and it will be ugly. But how then do you build systems that cater for 60 million customers. That's lesson two: - If you are going to build something huge, then put the very best people you have on it and run a portfolio that only has a couple of these at any one time - because you can't resource more than that effectively. Departments in government don't have the kind of people in the numbers needed to pull this off, which argues strongly for departments actually not having any of them. Delivery of complex IT is a specialist skill and should be seen that way. An in-house government IT SWAT team (it's been argued for before, more than a few times) may be the answer although no entry from such a team is likely to be silent, surgical or short-term. - Whatever you are delivering, make it incremental rather than all out. This has been one of the central tenets of IT delivery in government since the SPRITE report by Ian McCartney, is listed as one of the 7 (8?) reasons why projects fail by the NAO and is assessed whenever an OGC gate review is being carried out. Easier said than done - if you're bringing in a new tax credits regime for the very first time, how do you roll it out incrementally? Well, perhaps for one thing, you don't turn off all of the old systems overnight. Legislation is a difficult thing to bring in a piece at a time, but the risk of no fallback needs to be included if there's going to be an all out approach. - Do things together, not apart. By having a portfolio of projects across government or an organisation, aligned to business process and overall goals, you can see where you're doubling up or where you're not putting in any effort. This would show up, say, just how many portals are being built or how many CRM efforts are underway. Aligning business processes across an organisation - even a small one - is challenging, doing it across multiple organisations (i.e. government departments) is enormously hard. But if you don't start, you're never going to show it can be done. So a Department for Receiving Money would be an obvious place to start as would a Department for Paying Money. HR, Finance, Payroll type systems are usually the first place to start with shared services - they're identified as low risk but, actually, are probably harder to do than external systems. The internal blocking is likely to be much greater with,e.g. an expenses system - believe it or not there are people who believe having the name of the person claiming in the top left instead of the top right is a meaningful differentiator between systems and deployments (just like where the search bar goes on a website - people get very wedded to it). - Finally, the stakeholders need to be right there. The assertion that the DoH have spent £10 billion and haven't even got the GPs to use it perhaps illustrates that. The NHS is absolutely huge - no organisation touches its scale in a single geography. People talk about Walmart (geographically spread, local systems), the Red Army (no longer with us) or the Indian Railway (look at their safety record) as being equivlent and, of the 3, Walmart is probably the only comparator - but it has a CEO, a board and a line of control. The NHS has none of these. Whilst I think the NHS IT programme could be singled out for many failings, coming up with a trite "I'd have done it this way and it would have been fine" is not easy. The level of integration is enormous - of process and technology, but particuarly the process change that comes with the new technology. Maybe GPs are not fans of change nor of technology. The NHS is littered with pilot after pilot that has never been rolled out - arguing that incremental delivery has been tried 1001 times but has never been shown to work. So an approach similar to big bang was adopted ... and, so far, is hasn't worked (at least as far as the press is concerned - I suspect that there have been many victories along the way). Like I said - all to provoke a debate. Responses welcome, agreeing or disagreeing, as long as they're accompanied by proposed approaches that could be tried.
Posted by Alan at Monday, December 19, 2005