Real Software Engineering is still in the future. There is nothing in current SE that is like the construction of the Empire State building in less than a year by less than 3000 people: they used powerful ideas and power tools that we don’t yet have in software development. If software does “engineering” at all, it is too often at the same level as the ancient Egyptians before the invention of the arch (literally before the making of arches: architecture), who made large structures with hundreds of thousands of slaves toiling for decades to pile stone upon stone: they used weak ideas and weak tools, pretty much like most software development today.

The real question is whether there exists a practice in between the two—stronger than just piling up messes—that can eventually lead us to real modern engineering processes for software.

One of the ways to characterize the current dilemma is that every project we do, even those with seemingly similar goals has a much larger learning curve than it should. This is partly because we don’t yet know what we really need to know about software. But as Butler Lampson has pointed out, this is also partly because Moore’s Law gives us a qualitatively different environment with new and larger requirements every few years, so that projects with similar goals are quite different.

Until real software engineering is developed, the next best practice is to develop with a dynamic system that has extreme late binding in all aspects. The first system to really do this in an important way was LISP, and many of its great ideas were used in the invention of Squeak’s ancestor Smalltalk—the first dynamic completely object-oriented development and operating environment—in the early 70s at Xerox PARC.