Thursday, August 31, 2006

Zero Faith in Zero Defects

EWeek ran an interesting story today on Oracle's Fusion Applications. You can read the article here. Among other things, the story broke the news that Oracle has instituted a zero-defect policy for the first release of the Fusion Applications. I personally got a bit of a chuckle over the zero-defect policy. Based on experience, I have zero faith in zero defects - especially for an initial release.

Zero defects is a very worthwhile and commendable goal, but one practically impossible to obtain. The nearer a project comes to delivery, the more pressure is applied to compromise on the zero defects goal. As any project manager worth his or her salt will tell you, the key issue in project management is balancing the classic triple constraint: cost, schedule and quality. Choosing an absolute standard for any one of these constraints will increase the value of the other two. In the case of zero defects, costs increase and the schedule grows longer.

Now couple a zero-defect policy with the promise that the Fusion Applications will be released in 2008. We now have some absolute boundaries on not one, but two constraints: zero defects delivered by the end of 2008. Assuming that the quality and schedule constraints can both be obtained (and I'm only conceding this point for the moment), that leaves us "wiggle room" with only one constraint - cost. Simply put, hitting both the schedule and quality goal will likely cost some pretty big piles of cash. Hmmm, saving cash versus software quality...which concern do you think will take priority?

What usually happens is that, as the project comes close to the budgeted cost ceiling and the delivery day draws near, the decision is typically made to compromise on the zero-defects goal rather than exceed the budget and deliver late (for additional references on this trade-off, recall the early versions of the 11i E-Business Suite). I anticipate that the same trade-off will be made here, probably sometime in early 2008.

A zero-defects policy is a wonderful goal early in any development project. And I'll tip my hat to anyone delivering a perfect initial release within budget and on schedule. However, experience indicates that software quality usually loses out to cost and schedule. Do I expect the first release of Fusion Applications to be of higher quality than the early releases of the 11i E-Business Suite? Reading John Wookey's words from that E-Week article, he sounds pretty serious about quality, so I'd say "yes". However, any customer expecting zero defects should prepare for disappointment.

No comments: