Tuesday, August 19, 2008

When Projects Go Wrong

Don Paskini has linked an excellent article on IT projects and management consultants explaining just why there seem to be quite so many lurking in the corridors of power. It's a should serve as a very sobering read for any Tory who imagines that as they take control of their department they'll be able to go "consultants, out, now!".

As someone with a fair amount of experience working on IT projects, I agree wholeheartedly with the article and much of the sentiment. The general impression of large government projects (IT and otherwise) in the media is that they frequently go vastly over budget and take forever to deliver. Often the changes in delivery time and cost can often be explained. So, I thought I'd have a crack at explaining why projects sometimes appear to go wrong.

Changing Costs, Changing Deadlines
What most people don't realise is that when an IT system is initially designed, the customers are not totally aware of what it will do. They will have a rough outline, but the fact is that it's very difficult to envisage the final shape of the project at it's inception. If this sounds strange, consider the following example.

A NHS Trust has developed a system that automatically schedules home visits for nurses, every day a nurse with home visits to perform recieves a list of patients, the nurse them gets the appropriate records and goes off to visit the patents. As the system takes shape a number of additional requests get thrown into the mix.

  • Managers suggest that it would be good if visits were grouped together by distance so that nurses completed their visits more quickly.
  • Doctors suggest that rather than have nurses take the patient record, the patient list should include a list of what checks and treatments are required, with the results then recorded afterwards.
  • Accounts suggest that it would save form filling if the travel distances were automatically sent to them for calculating mileage payments.
So what we have is a simple idea that has grown larger over the course of the project. Modern development methodologies recognise that this kind of occurence is not uncommon, meaning that changes can be accomodated without the need to rewrite the system. This project may have been happily recieved by it's users and have delivered a number of efficiency improvements, but on paper looks like it was delayed and over budget.

Another good (although non IT related) example of bad reporting on the subject is the 2012 Olympics, although very little building work has taken place, we get reports of spiralling costs. Whether the costs have overrun or not will not be known until 2012 when everything is built. Most of the cost announcements so far have related to spending plans rather than costs.

Constantly Changing Justifications
Opposition politicians like to bash ID cards because apparently the government is constantly changing it's reasoning for it. One week it's terrorism, the next week it's identity theft. The truth is rather different.

Most government agencies have problems verifying identities, when someone gives them their details they simply have to take that person's word of it. Some checks might be made, but this is often a long and complicated process and in a lot of cases, not worth the hassle. A national identity register, creates a single identity record that is checked in detail when it is created. This gives government agencies far more accurate identity verification, it also gives them the ability to better check their own data against that held by other agencies.

The upshot of this is that government agencies have far more accurate data and much greater ability to act in a more joined up fashion. The result is benefits accross a very large number of departments. To sum up, it's not that the government keeps changing the justification, it's that there are many different justifications.

No comments: