Model driven architecture could set the standard

In my time, I've seen too many standards bodies get subverted or run aground to expect much from public standards processes, but I'm always open to the possibility that -- this time -- things might actually work out for them. And the latest group to raise that hope for me is the Object Management Group, with its efforts to bring harmony to software development.

Object Management Group is a hoary veteran of the early 1990's vogue for groups of vendors agreeing specifications and agitating for adoption of particular technologies. While most of the other groups from that time died the death (the Open Group still exists in some limbo, but the work its component parts -- X/Open and OSF -- set out to do is pretty much lost).

OMG, meanwhile, still goes on. CORBA, its architecture for distributed processing is still reasonably well regarded -- the current fashion for Web services is based on very similar ideas. But the group has also played a role -- or at least cheered on -- as one of the industry's long-standing feuds got sorted out.

I'm talking about software engineering. Back in the 1980s, there were several gurus, each with their own method for object-oriented analysis and design. They were usually called "methodologies" rather than methods -- for no reason I could understand, except that "methodology" sounded more impressive. Coders working within one method would produce designs that were incomprehensible within another method - examples included Booch and Jacobsen.

Each method was a "cult" with "zealots", according to Andrew Watson, the technical director of OMG. But during the 1990s, commercial pressures began to tell. Users were fed up with training and retraining developers for no reason, so IS shops mostly refused to join any of the cults. Since none of the cults could take over the world, all the methods gradually folded into one, as their originators were absorbed by one company -- Rational Software. "Rational Software became the regional superpower," says Watson.

At the same time, OMG became the World Bank or the United Nations -- the body charged with presenting decisions acceptable to all. The group started work on a Unified Modelling Language (UML) in 1994, and published it in late 1997. As Watson tells it, now there was one object-oriented modelling standard, OO modelling became acceptable to users, leading to an explosion of use: "70 percent of IT shops use UML somewhere now, and there are more than 60 books." A second version is well underway.