Simple steps to making better software

Ed Hasted, writing for the Guardian has written a piece on "Simple steps to making better software."

Those with a passing familiarity of software engineering will recognise a restatement of a well-known truth, that successful, novel, large software systems are built incrementally, starting with very small pilot system and gathering user feedback, developer experience and design insight.

Sometimes it seems to me that the public sector does so badly at building large software systems because of the nature of the tendering process used---they assume that because what they want is (for example) a identity database to track and record the details and activities of all citizens with only vague and uncertain requirements, that they should put out a tender for one, rather than a handful of tiny, interoperable pilot schemes, at a fraction of the price. Such clusters of small projects are currently being used by the JISC e-learning programme with great success. While the strike rate may not be 100%, those projects which fail to deliver what was expected, or which highlight difficulties in the field do not hold up other projects, and because all the projects are required to use and test against open standards, those that do deliver can interoperate.