Adrian:

From a Data Flow Diagrammer's perspective, a Swimlane is an example of a forced, artifical partitioning.   Ya ain't going to be able to do large scale systems integration by "dividing up the pie"  that way!

When I do a decomposed set of data flow diagrams for a larger project, any given organization can logically result on any number of different diagrams.  If I were to, for example, place Purchasing processes all together, the result would often be a real spider's web of data flows.  So much as to make the diagrams worthless.

Tony

Hi Adrian and all,

No, I don't think there's analytical value to a dedicated "swimlane diagram."  The concept of swimlanes is key, but the value of using them for analysis comes through in pairing them with other modeling methods.

As Kimbo mentions above, swimlanes are great for showing interactions between actors.   If not paired with a more analytic technique, however, I'd argue that they function more as illustration than model.  

For example: I can easily throw a couple ad-hoc swimlanes into a PPT or on a whiteboard to describe a flow of interactions.  This gets the concept across very easily to business partners and non-analytical stakeholders... it *illustrates* an example.  I wouldn't necessarily consider it an accurate (or actionable) model.

If someone is asked to produce a "swimlane diagram," I suspect that the requestor is asking for a more structured process model (take your pick of methodologies) but may not be aware of what is available / what is appropriate.

-- Adam