Taurus is one of the larger IT project failures in terms of the scale of the time and cost overruns. The clear message is the importance of stakeholder management, otherwise requirements inevitably become far larger than can be accomplished with time and budget constraints. With the Taurus system the approach to requirements management appeared to be to attempt all requirements – of course, all this did was to balloon duration and costs until the project was ultimately killed by warring factions of management consultants.
Taurus stands for Transfer and Automatic Registration of UncertificatedStock. Basically a way to automate the back end processing of London Stock Exchange transactions, since the prior Talisman system was stretched.
Taurus was first estimated to take 6-18 months and ultimately took 10 years, without success. Budget overruns were similarly impressive, with initial estimates in the range of 6-18 million pounds and the final figure being somewhere between 400 and 800 million. I believe the reason for the range of estimates is because the project having multiple phases and directions to it, with two initial proposals (Taurus1 and a committee recommendation) being scraped until a final and very ambitious proposal was attempted.
The core problem was 280 financial institutions in putting requirements and the economist John Kay describes on his blog.
“It would take a full page of the Financial Times to list interests that needed to be placated and demanded to be represented. The registrars feared that a centralised computerized system would put them out of business. Listed companies wanted to make sure they could find out easily if arbitrageurs were on their share register. Private client brokers wanted to protect their customers’ rights to hang share certificates on their walls and make cheap crossings on the English Channel with their concessionary shares. The government insisted that communications should be as secure as the Government Communications Headquarters and investors’ funds as well protected as the gold in the Bank of England.
And so it went on. And on, and on, through interminable committees and working groups.”
Ultimately Taurus was scrapped and CREST was implemented, Daniel Davies argues CREST succeeded due to regulatory stability, but I think that is overlooks the stakeholder chaos that Taurus experienced independent of regulatory change. Rory Linwood argues that requirements management was the crux of it here. CREST didn’t give everyone everything, but it made the tough decisions early and was successfully implemented within 3 years.
Similar issues with the challenge of design by committee and unclear reporting chains can be seen in the project failure of the Typhoon Eurofighter weak requirements management as also evident in the FBI’s virtual case file project.