It sounds like you don’t have a problem with (for example) propagating errors up to the top level for reporting, but rather are saying that maybe it doesn’t need to be graphically represented that way, correct? Propagating errors up to a higher level in order to facilitate appropriate error handling is a common coding technique not because it’s particularly efficient (which it’s not) but because the person who designs the code needs to specify where the error handling will occur and how it will manifest. Business analysts who are modeling business processes need to think about the same thing: if an error occurs at some level, how will it be handled? BPMN provides a graphical method for them to propagate the errors up so that they’re visible at correct level of the diagram, both for presentation purposes and to put the management of the error in the right place.

The issue (as Tom Baeyens has stated) is that even though business and IT people might both use BPMN to model a process, the business people are modeling the most effective business process, and the IT people are modeling the most efficient code (albeit from an abstracted viewpoint). That doesn’t always produce the same models in BPMN, which means that somehow you have to be able to take the business process that the business person modeled and tweak it so that it will generate an efficient/complete executable without corrupting the original business process.

Thank for keeping us honest. I think Bruce, Jan and I do not disagree as much as the whole debate makes it look like. It boils down to the question of whether you can control the use of a language after it has been released.

On the one hand, a lot of work went into BPMN and it is a very powerful vehicle, if you know how to use it properly.

On the other hand, users may appropriate and extend the language in ways its designers could not foresee. For instance, Garrett Hunter has presented a great case study how Sony used a mashup of BPMN and IDEF0 to save a failing IT project. Should we tell them “you can’t do that because it’s not how BPMN was intended to be used?”

I believe that yes, there is great merit to structured BPMN education that explains the intention behind the language, its vocabulary, and shows how to use it in a well-formed way. But I also believe we need to keep an eye on how the larger BPM community uses BPMN and feed that information back into the development process. If you read Bill Bryson’s entertaining account on the history of the English language you’ll understand where I’m coming from.

Best

Michael

It’s a bit of a chicken-and-egg problem: most people don’t use the extended BPMN constructs because they’ve never been trained in them: they likely don’t even know that they exist, even if they’re using a BPMN tool that shows them in a palette. If they were trained in them, would they use them? Maybe.

The students in your classes are likely there because they have reached the limitations of what they could do with the basic set, and recognize not only that their processes are more complex, but that there has to be a better way to model them. That level of self-awareness about their own shortcomings does make them atypical.