Comparing BPM to JBoss downloads is rather disingenuous, Keith. JBoss is free to download, and free is a great motivator. A fairer comparison would be to look at revenue between the industries, though as to why one would compare a BPMS to an application server, I'm not quite sure.

Good BPM, like any software that users like to use, requires good business analysis. I agree, it requires *modeling*. But object modeling wasn't the first nor will be the last approach to constraining concepts into some kind of notation or terminology. SOA is trying an approach that was heavily influenced by objects, but no one really agrees on what is/is not SOA, so it's hard to tell what principles they're keeping or jettisoning. I do know that most that I talk to don't believe SOA = "web services" which is the biggest culprit at forgetting many of the lessons of object orientation.

I think it remains to be seen what the industry will ultimately find comfortable. There may be a wide variety of ways to tackle that, as we seem to have lost any real unifying force -- everyone's going their own way these days.

Speaking of which, if you think BPM and SOA are over-hyped, how about the application suite wars? Oracle didn't buy all of those companies up because it thinks "build it yourself with JBoss" is going to triumph, it still believes packages are going to rein , maybe with BPM making it a bit more flexible. And in some cases, they're right. I see SAP, or Oracle, or [one of the 5 companies Oracle bought] nearly everywhere, and their presence growing. New corporate regimes have continued incentive to bring massive capital investments to bear to change the political structure of organizations by overhauling its application software. This game isn't about "what's good for users", it's about crafting a new power base. And the big SI / consultant ecosystem, which guides many CIOs, has little incentive to promote "analyze, model & build it yourself" -- they want to build a practice around large numbers of people maintaining a packaged software appliaction for 10 years.

A CIO recently put it to me this way -- either you're a CIO that wants your team to think their way through a problem -- that is, model it (with components, services, objects, relations, resources, message exchanges, whatever -- pick your poison) and work through your problems -- or sign a cheque to an outsourcer or packaged sutie vendor, and have your team manage the timelines & budgets. Many aren't willing to deal with the political pain that's involved with the former, so they're much more disposed to the latter. With CIO tenures being fairly low, they'll be gone before the maintenance bill arrives, anyway.

Is BPM and SOA "sweeping the world"? Not really. But any useful technology that has a *hint* of freedom from the lock-in from the outsourcers, SIs and package suites of the world is going to have legs. And anything with legs tends to have a perception that always vastly outstrips the relaity. Sometimes reality catches up (look at the iPod), but other times we wake up wondering what the fuss was about (remembering "push" technology, for example).

I would say the comparison between BPM/SOA and ERP suites is itself disingenuous. After all, a main claim for BPM/SOA is that you can use them to leverage ERP suites - extract more from them via "process projection". So BPM/SOA and ERP are (supposedly) compatible technologies.

The reason that I compare BPM/SOA and application servers is that people generally take one or the other route to building custom functionality for automating business processes. In particular, I would stick by the comparison of "all BPM/workflow installations ever" with "JBoss application server downloads per month" as illustrative, especially as there are now various open source and free BPM products.