IT in the horse latitudes

The Reg reports: Silicon Valley staff 'gloomiest' in US. Some 27% report fearing for their jobs, compared to a mere (?) 18% in the U.S. as a whole. A major reason: offshoring, underscored by a Google ad floating enticingly in the sidebar to the right of the article text, offering Romanian programmers for $8 per hour. What do you get? "Under nice looking interfaces you will find quick applications designed to meat all your requirements." Ooh, meat my requirements!

Well, requirements are important. In the early 90s, I read an article about the abysmally low success rates of projects, with an excellent analysis of the critical factors. This article identified two major factors: vendor failure to gain and maintain upper management support, and a failure to acquire and maintain accurate end user requirements. Technology, programming skill, methodology, choice of programming language - all these were second-order effect. Nothing mattered nearly as much as figuring out how to keep the idea sold to bosses, and how to get their underlings to use the damned thing.

Another Reg article, When the customer is always wrong attributes the ongoing failure rate of software projects - still around 75-80% - to failure on the part of the client to invest in managing the relationships with vendors. Well, that's your basic failure to acquire and maintain upper management support, and it certainly lends itself to more screwups in acquiring and maintaining end user requirements.

I got out of software in part because of the frenetic runarounds and the high failure rate despite all the energy expenditure. It looks like nothing much has changed, despite a huge workforce restructuring. One problem might be that the money saved in offshoring tends to drag down budget estimates for project management on the client side as well. This is bad thinking, if so. If anything, the money saved by hiring cheaper offshore programmers should be put toward staffing the client side of project management more heavily. But that's not how things work in the real world much of the time - if something is cheaper to acquire, management feels that the transaction costs should be cheaper as well. Ah, but there's no free lunch.

IT will be nettlesome in any case - upper management support and accurate end-user requirements are often in collision. "I automate people's jobs - it's what I do," I used to tell my own clients in despair. If upper management wants to cut costs by automating jobs, and their employees know it, then your best source of accurate end-user requirements - the client's workforce - is pitted in an adversary relationship against the people paying for the project - their bosses. You can lead horses to water, but you can't make them drink. Small wonder that software projects seldom end well.