Pages

Wednesday, May 2, 2012

Rigid Engineers, Obsessive MBAs

I feel a good, old-fashioned rant coming on. Just for balance, I will combine two rants: Rigid engineers and obsessive MBAs.

RIGID ENGINEERS

Engineering is a great discipline. I like it. I try to adhere to its basic tenets. I feel that engineering has the following strengths: it is rigorous, it is regular and it focuses on the relevant.

Bridges tend to stay up. Planes tend to fly. Ships tend to float. If you have a house or a car or a smart phone, you owe the engineering discipline thanks.

Software engineering is a little bit different: our building materials are abstract and our rules are a bit fuzzy and our adherence to our rules is not guaranteed.

Weinberg's Second Law: If builders built buildings the way programmers wrote programs, then the first woodpecker that came along would destroy civilization.

However, my rant today is not about loosey-goosey software engineering, but rather its opposite: inflexible software engineering. The discipline of engineering is not good when there is no good rule and some software usage requirements create a very sloppy execution path. Instead of a measured, graduated response to the complex set of requirement, instead of a long chain of events from messy reality to complex model of the messy reality, the rigid engineer gives us a clean abstraction. Even if the clean abstraction does not, in fact, support the actual business process to be automated.

I have had to deal with engineers who actually raised the following objections to proposed designs or changes to their beautiful cathedrals of structured code:

  • "That would break my parser"
  • "That would require a goto"
  • "That would importing a large, messy library"
  • "My structure can't do that"
The subtext is always the same: "I want reality to be changed to suit my model." This often expressed as "the user can do {long complex sequence of instructions} to get the same result. There is no need to change my code."

To me, this is the tail wagging the dog: software should do what it needs to do, first and foremost. It is the job of the engineer to make that happen. It is not the job of the engineer to change the scope of the task to make the software simpler, unless the scope is beyond the ability of the implementor. In that case, the implementation folks should say that and try to work with the users to find a compromise. Decreeing that some functions are a bad idea is not what I mean by "compromise."

This belief that if your implementation cannot support it, then the request is bad and the implementation is good is related to the "great code is not great software" idea I touched on before, here.

OBSESSIVE MBAs

A colleague drew my attention to this great article:

http://www.forbes.com/sites/adamhartung/2012/04/20/sayonara-sony-how-industrial-mba-style-leadership-killed-once-great-company/

This article is a study in how classical MBA management has hurt Sony. The article describes a situation that I find increasing common not just in consumer electronics, but also in IT. (I have ranted about other manifestations of this, notable here.)

This article describes the typical MBA focus on operating efficiency, a la Demming. This focus is great, short-term: assuming that what you are doing today is what you need to be doing tomorrow, then finding better ways to do that is a great idea.

But in technology, figuring out what you should be doing tomorrow is often very important. Operating efficiencies gained with obsolete technology are often smaller than operating efficiencies to be gained by using current or even leading-edge technology.

Taking chances and taking risks as you figure out what it all means to you pays real rewards.

But classical MBA management teaches managers to be wary of the new and to concentrate on the current. In an industry where 18 months is a typical development cycle, a three year plan to increase efficiency will be two cycles out of date by the time it is finished.

I understand that operating efficiency makes sense. I realize that most organizations cannot handle constant flux. I appreciate the need to plan and to budget. But I am horrified at the ever-increasing tendency of the managers we meet to assume that all technology is pretty much the same and that three-to-five year roll-out plans make sense for software, with "freezes" while the implementation is done and evaluated. No wonder so many companies are out of date and out of touch with IT.

No comments:

Post a Comment