I attended the keynote by Dennis Wisnosky, CTO of Dept Of Defense, today at the Architecture & Process conference. He is currently on a campaign to get vendors to make truly interoperable implementations of BPM technology. He has been testing implementations of BPMN, and found disparity.
A notation is…
He feels that the most important thing is a standard notation for business processes. What do you get with a standard notation? You can have a document that displays something, and everyone should be able to read it, and agree on what it means. He gave two examples of notations.
First, he showed an electronics schematic. He pointed out the resistor symbol, the capacitor symbol, and the ground symbol, included in a simple filter circuit. You go to any electrical engineer, in any country of the world, and they can read this diagram. It is a standard notation that works.
Second, he showed musical notation. Again, this is a world-wide standard which consists of notes on the staff, as well as a collection of other notation. Again, you can take a piece of music to any musician in the world, and have agreement on what the notation means. It is clear what notes are to be played, when they should be played, and for how long.
He is absolutely right. You should be able to take a BPMN digram from any product, and show it to a person who is trained in BPMN, and there should be an unambiguous interpretation of what the diagram means. That is the goal of BPMN.
But a notation is not…
The analogy with music is a very good one, because it exposes some other properties of a notation that people find non-intuitive. For example, you can write a jumble of notes on a staff, but that does not mean that any given instrument can play it. A tuba player can read the piccolo part, and agree what the notes mean, but that does not mean that the part can be played on a tuba.
A standard notation does not imply that all legal notation can be played. Each instrument has allowable parameters around what you can and can not play, limited certainly by range, but also other factors: a piano can play chords while a wind instrument can not. (Please, no smart comments about bag pipes, OK?) A composer does not write a piece of music as an “abstract” composition of notes. A composer considers each instrument, and writes music that that instrument can play. This is expressed as standard notation, which anyone can read, but there is a lot more composition than simply creating the notation.
It is really important to keep this in mind for BPMN. When a process analyst creates a process diagram, that analyst must keep in mind the capabilities of the process engine that will be executing the process. The process should be designed specifically to run on that engine. But then the process can be expressed in a standard notation that we all can read and agree on the meaning of.
Conversely, this means that when an engine imports an XPDL file containing a BPMN diagram, the BPMN and XPDL standards do not, by themselves, guarantee that the engine will be able to execute it. The best we can hope for is that the engine tells you when there is something “out of its range” and can not perform.
Don’t confuse notation with universal language…
Imagine a universal instrument that could play any combination of notes for any duration, but there are other considerations beyond simply being able to throw any notation at an instrument. A trumpet and an oboe may play the same notes, but they are qualitatively different instruments in a way that transcends the notation. The notation communicates certain important things, but it is not the full embodiment of the music.
Some people jump to the conclusion that all process engines should have a “complete” implementation of BPMN — whatever that means. Having all process engines with exactly the same capability would mean stopping all innovation on support for business process, and limiting all engines to a least-common-denominator approach. It has been my experience that each engine offers new and unique difference that are good for supporting particular kinds of processes. Stifling innovation for the sake of universality is a short sighted goal. The fact that the engines are different is not a bad thing. Yet we still can expect, and must demand, that the notation be consistent.
Dennis Wisnosky represents a very large and powerful customer of process technology, with influence across the industry. Watch what he does, because I think he is in a position to have a profound effect on the BPM marketplace this coming year — and he is on the right track.