I usually write about different details of doing architecture in an Agile way and other related topics, but the reality of outside world is that Dutch government loses between €4bn and €5bn on failed waterfall IT projects. This is definitely a problem in other countries too.
As far as I know, this is the first time huge failures in governmental IT projects are discussed so openly. A parliamentary committee has questioned today five people from IT industry in the first day of enquiry. The first one was Hans Mulder (a professor at the University of Antwerp) who gives some hard statistics:
Just 7% of the projects with a budget starting at €7.5m can be said to be successful, Mulder told MPs. In total, 70% of projects fail. Of those which flop, 36% fail so seriously the new system is never used and around half are of doubtful value because they turn out to be too expensive, take too long or produce unexpected results, Mulder said. – source: DutchNews.nl
Second was Stephan Corvers (owner of Corvers Procurement Services BV). The main subject was about procurement process. His expected opinion was that failures have little or nothing to do with procurement process itself. According to him, the customer should better describe what is wanted before starting a project. In other words, more of everything what is already being done. The most noticeable is vagueness of words and statements he used when answering the questions. “Frames….bla bla….framework…frames…bla bla”. At some point a MP asked: “Frames?….euhm what?”
Third one was prof. dr. C. Verhoef. He explains clearly some real and painful issues in those failed IT projects. His main point is the lack of proper knowledge in IT industry. Nobody is willingly contributing to failures, but rather “the road to hell is paved with good intentions”. IT people are lacking knowledge to successfully deliver. Another point was lack of third element in procurement process, which is quality. These big IT projects are given to software companies based on cost and time aspects. In all kinds of games played out in this process, something perverse is happening. The companies which are capable of delivering and know the risks usually loose the project to competition, which neglects the risks or play the game of underestimating and compensating later and in the process make the project much more costly or complete failure for the customer.
It was a nice analysis of the problem, but solution was a bummer. His suggestion was to specify even more (quality aspects) before the procurement process starts. From my own experience, these beforehand “ilities” are always made up by some group of architects in their ivory tower. The effect of this is possible creation of even bigger problem – overcomplicated architecture driven by quality aspects that have little to do with reality. Another suggestion was to split up these big projects in many smaller ones.
Fourth interviewed was Tony Wildvank (ex-contract manager for large IT projects in government). Her main point was that we need more architecture and involvement of architects before we start. What she actually says: even more documents, theoretical analysis, and so on, before starting to deliver anything. The rest of her statements were either unwillingness to answer the questions because they were too sensitive or full of vague statements like with Stephan. Statement like: “software projects are same as process of making socks in a factory” is a sign of huge amateurism and lack of knowledge.
Fifth and last interviewed today was Swier Jan Miedema. He is an entrepreneur with 20 years of experience in working for or directly with companies like Centric, Logica, and Ordina. He is the embodiment of the pain many honest IT people have in Dutch government IT projects. Many of his statements were maybe too honest and too personal, but he is one of the few brave enough who speak out in such way. I really wonder how this will effect his future business.
His was really appalled by atrocities of big and well-know software integrators in Netherlands. According to Swier, IT projects in government are controlled by these companies. They define what happens with IT at municipalities.
I love his solution. Vendors must prove they are capable by delivering something in one week. If not delivered, they are not paid. After this, close collaboration starts where vendor continuously proves its capability by delivering frequently. He did not say it, but it was obviously Agile he was talking about.
Altogether, I was disappointed and sadly confirmed in my assumption of extreme amateurism in these multibillion projects. Insanity is mind boggling. A clear example is that nobody thinks of splitting a large project into many smaller ones. Even some of the interviewed people are mere confirmation of this amateurism. Some pretend to know how to successfully deliver such large projects. More architecture, better specification, better control, and so on.
The Dutch IT world seems to be completely divided in people constantly discussing why those big waterfall projects are failing and Agile world. They don’t interact with each other much.