Software Voices

In a discussion with the same architect that sent me down the path of investigating Software Factories, I got the first glimpse of why this methodology might be useful -- important even.

Remember when the Java guys trotted out that "anti-bloatware" bandwagon based on the observation that an individual only uses 20% of any given software package? Their solution was to eliminate features from commercial software and have you hunt down the items you needed when you needed it. Given that we didn't find legions of people giving up MS Word for Wordpad, I think it is safe to say this misguided notion is dead.

That said, I think it hits the nail on the head for IT applications. When I was an MIS Director[1] at a medium-sized real estate company, we developed unique applications for each person. The goal was to have the application actively assist people in accomplishing their job, using a workflow that made sense for them. Every feature in the application was needed, so the menus of the programs themselves were helpful in remembering how to do something. The data was presented in a way that matched the user's world view and vocabulary: for example, Marketing and Accounting used different definitions of "sale date" and "sale price."

This was; however, a development intensive route to take. We ended up with a large number of applications that varied considerably even though they all used the same database on the backend and were based on the same concepts. Though this was a signficant amount of work, it had a direct impact on the job satisfaction of those that had to use the system. Because they were generally able to work more effectively, job satisfaction rose and turnover dropped dramatically. This was particularly true of those who spent significant time one the phone helping others -- being able to accomplish a task quickly meant the person on the other end was much less likely to yell at you.

I think the idea of a "product family" in Generative Programming/Software Factories has tremendous promise here. The idea is to shift development effort from a specific application to a language/system that generates that application. Over time, as effort is focused on this higher level of abstraction, you should be able to generate any number of applications in the same domain. Initial development time is the same or longer, but the result is the ability to leverage that work in a way that allows new applications to leverage all of that work and be built much more quickly. The official pitch is that this will reduce development time overall, but I hope that it would make developing a unique application for every employee that exactly fits their needs possible.

In the not too distant past, the personal computer idea of "one person, one CPU" seemed unobtainable but is now commonplace. I hope Software Factories usher in the era of "one person, one application."

[1] MIS is the old name for IT -- Management Information Systems.

At the urging of an architect I really respect, I have been following the evolution of Software Factories (aka, Generative Programming or Intentional Programming). It is one of the most painful things I have ever agreed to do. Not painful in that oh-so-delicious mind expansion that comes from the exploration of new intellectual territory (like I had when first exploring Object Orientation or in moving to C++ from C), but painful like being forced to sit through a presentation on timeshares. If you are going to start a new (programming) religion, you need some spirit at the front of the church. Instead, the "bibles" of the religion are dense prose is virtually impossible to read from cover to cover. One doesn't even start to describe what a Software Factory is until page 155!

Brad Cox and Grady Booch provided much of the fire behind Object Orientation. They used descriptions and analogies that managed to catch the popular imagination: objects as electronics, objects as helpful friends, and objects as legos. All have positive connotations. The Software Factory people use analogies to manufacturing plants: Have they forgotten all of the pain Americans recently experienced with the disappearance of these plants? Are they surrounded by software developers that long to work in a factory?

If this movement is to be a success, I think they need to find some compelling spokespeople with better analogies and an ability to get to the point.