How not to do it

The Government's development of the universal credit system will become a case study to be examined in future decades as a classic example of systems development getting deeper and deeper into the mire, while its project sponsors maintain a wilful blindness to the absence of progress.  The Guardian charts the latest inanity:

Despite a scathing NAO report in September, Duncan Smith has been insistent that the project remained on time and on budget. In December he revealed a new plan for delivering the project.

Sources indicate that tensions between government departments spiked after Duncan Smith refused to restart the embattled project afresh – a move that would have incurred massive write-off costs and political embarrassment.

Duncan Smith is understood to have insisted on a "twin track" approach – keeping current universal credit development going to prove that claimants could use the service before the 2015 election – while ordering money and time to be ploughed into a web-based system that did not rely heavily on jobcentre staff to fill in claimant benefit details.

According to the newly approved plans, hundreds of thousands of benefit claimants will then be transferred from one design of universal credit programme to the other once the digital design is ready some time after the general election.

Plan A won't work, Plan B won't work, and transferring data from Plan A to Plan B will be nightmarish.  It would be laughable if it were not so serious.