Years ago, an astute manager summed up a problem that I had only vaguely intuited up to that point in my career.

“A big problem with most companies,” said Roland, “is that they have no institutional memory.”

As I recall, Roland was describing capricious political winds, and lamenting that the only form of loyalty a company has to employees is the kind they put in writing. As soon as there’s major M&A activity, or HR decides to rebalance salary allocations, or an incentive program gets adjusted to the latest management fad, all recollection of old priorities and soft obligations vanishes in a puff of smoke.

If anything, Roland was understating the problem. Companies routinely panic and change strategy half-way through an investment cycle, because they can no longer articulate the rational analysis that led them to take a plunge. Buzz floods the internet about some innovation that makes everybody excited, but we forget that we’ve heard the idea before, behind some different terminology. (Are you nodding your head because “cloud” in the last few years is just a recycling of “utility computing”from circa 2000? Trev, a colleague of mine at Adaptive Computing, showed me a dog-eared copy of The Challenge of the Computer Utility, by Douglas Parkhill. It’s all there–XaaS, elastic and on-demand, in 1966. And who knows–maybe sci-fi writers or the designers of Eniac had thought of it even before Parkhill…)

But I digress.

One particularly insidious form of forgetfulness in software relates to technical debt. Another colleague, Doug, reacted to an expedient workaround this way:

My one regret with this is that by doing something that is good enough it will never get the attention it might deserve to be made better. This happens each release: we make compromises at the very end to get it out the door, promising ourselves that we’ll revisit it later.

Folks, we don’t keep these promises to ourselves very well; Alzheimers is endemic with regards to technical debt. The only thing that saves us is that Continue reading →