Tuesday, June 30, 2009

Innovation 101 for the Perfectionist

To innovate is to introduce something new to the public that is usable and meets a perceived need. This article will dispel some common misconceptions that the perfectionist may have with his approach to design. Since women are perfect anyway and have this area covered, I will only use masculine pronouns. Actually it's for the sake of brevity. As someone who IS a perfectionist, this article is written for other perfectionists just getting his or her feet wet with this stuff. I suppose you could say that a perfect design goes hand-in-hand with proven innovation. But as the perfectionist I cannot say I've always shared this perspective and it's difficult to get over it. Hopefully this article can dispel some of the myths holding the typical perfectionist back from greater success.

MYTH: Perfection in design is necessary.

The fundamental reason perfectionism fails as a good approach to software design is because perfectionism does not factor human error into code design. Human error is inevitable and "nobody's perfect". To expect consistent and perpetual perfection from yourself, your team, or even your customer's requirements is unrealistic. Therefore, the perfectionist must accept human error as a valid and acceptable factor in the design process. It is haphazard to turn a blind-eye to this variable. Expect loose-ends to come from anywhere. Probably the most dangerous assumption to make is that the customer has a good understanding of his or her business challenge. In the end trial and error is inevitable to evolve and refine a product.

MYTH: Perfectly designed software leads to a quality product

How do you measure the quality of a product? Simply by how accurately it accomplishes its original purpose, which is defined by the customer's requirements. But how often are those requirements written in stone? How often can we say that they everyone understands those requirements perfectly? The perfectionist operates with these assumptions. Furthermore, it takes too long to develop a product that is perfect. Since software requirements change quite frequently, the result can be a lot of wasted time on a polished but irrelevant design. If you can't get it out the door, if it's incomplete, then it is far from perfect. One key characteristic shared by all of the most successful products out there is that they each do exactly one thing very very well and that has been attained through multiple iterations of trial and error.

MYTH: A perfectly designed product is highly maintainable.

Since details of implementation are often up to one individual programmer, his perception of perfection may be way-off the mark to someone else. Furthermore, code that conforms to standard OO design patterns religiously is not maintainable because it proves to be an overwhelming task to trace program flow through a deluge of, however highly cohesive and decoupled, unfamiliar classes. An experienced developer knows that any algorithm that can fit conveniently into various design patterns is common-place and more than likely already in a reusable framework somewhere anyhow. For the rest of your intellectual property, you yourself know that it doesn't fit conveniently into standard design patterns. Moreover, you know that your product isn't innovative because it is common-place. And so it becomes necessary to bend the rules while ironically keeping in mind other developers who might have to look at it later.

No comments:

Post a Comment