Want to know the secret of lasting organizational success? Go on a diet.

No, losing weight won’t change the organization you lead. But then, going on a diet probably won’t result in you losing weight and keeping it off, either, and for very similar reasons.

Want to lose weight? Eat less, exercise more. This doesn’t require brilliance, expertise, or a book. (But if you want to buy a book that will help you lose weight, read The Moral Hazard of Lime Daiquiris by Bob Lewis and Dave Kaiser. On a treadmill. While snacking on carrots instead of Little Debbies.)

Anyway, losing weight is a poor goal, and when you set a poor goal, even if you reach it you’ll be in a poor place.

A better goal is to live the weight you want, not to lose the weight you don’t want.

It’s entirely parallel to the secret of lasting organizational success.

Choose a problem — the organizational equivalent of weight you want to lose. Perhaps it’s a proliferation of custom-coded point-to-point system interfaces, a common-enough problem. How do you go about fixing it?

I see a hand raised. Yes, Clark?

“Implement a service bus?”

Uh … Clark? What exactly do you mean by “implement”?

“You know. Select a service bus technology, and launch a project to convert all of your interfaces to services that run through the bus.”

“Thssspffft!”

Nice Bill the Cat impression, Kent. Did you have something to contribute?

“Thanks. I feel like Rodney Dangerfield in Back to School saying this, but seriously … in my company the IT Steering Committee would never approve a project like this. They’d compare the new business value we’d get from it to the latest requests from, oh, I don’t know, say, Marketing? And that would be the end of that.”

“Even if it did get approved, while the project team was busily converting interfaces to the new technology, all the other application project teams would be just as busy creating new interfaces using the old technology.”

Thanks, April. But that line of thought points to a solution. Any suggestions, June?

“Well, obviously, in addition to the conversion project you’d have to train all the other developers how to write interfaces using the new technology.”

Why are you making a face, May?

“Maybe I’m just old and cynical. I just remember our old COBOL programmers who got trained in C++. Most of them ended up coding 60,000-line-of-code “objects” that looked almost exactly like COBOL programs. So I’d bet if you trained everyone in the new technology, half of ’em would write the same old custom point-to-point interfaces. The only difference would be the language and transport.”

Thank you! Right on the money.

Okay, class, here’s the deal. If you want to clean up an interface tangle, you want the wrong thing, or at least the not-right-enough thing. I gave you the wrong problem to solve and you fell for it. As leaders, part of your job is to recognize when someone does this.

The wrong problem is fixing an interface tangle. What’s a better problem to solve? Yes, Bruce?

“Well, you told us the answer, didn’t you? The right problem to solve is how to change the IT organization so it only builds well-engineered interfaces. Do that and every project team will replace the old-style interfaces with new ones as part of every project.”

Bingo. So how do you change an IT organization so it works this way?

“Training?”

Necessary, Wayne, but not sufficient. What else?

“Standards?”

Nice try, Newton, and yes, you would need standards. But standards are usually about enforcement, and if you have to enforce you’ll be working way too hard. And besides, once you start to rely on enforcement you’ll need professional enforcers. Sound like the kind of organization you want to run? Yes, Mr. Boaz?

“Vell, I tink ver you’re going mit dis iss to change der culture.”

There you go. There are a lot of definitions of culture out there. From a purely practical perspective, a useful way of thinking about it is that the culture is “how we do things around here.”

Want well-engineered interfaces? Make the interface style you want “how we do things around here,” and the problem will, as Bruce suggested, solve itself. Yes, Isaac?

“Any suggestions on how to change a culture?”

Gee, what a shame … that was the bell. Class dismissed.

* * *

Until next week, at least. See you then!

* * *

New feature: This column marks the beginning of year 19 for Keep the Joint Running and its predecessor, InfoWorld’s IS Survival Guide. To celebrate, each week this year I’m going to pick out a favorite from the corresponding week from one of the past 18 years. This week’s blast from the past has to be the column that started it all, “Welcome to the IS Survival Guide,” (1/8/1996). All in all it holds up pretty well.