Not crazy!
Workflow languages (and BPEL in particular) are something we frequently discuss at MELCOE in relation to LAMS. Yoichi is the expert on this topic, and we are hoping he will lead a major study of workflow in 2006 if one of our current grant applications is successful.
Two other first rate thinkers on workflow are Scott Wilson from CETIS - who has written extensively on education workflows (including the role of LAMS) - eg http://www.cetis.ac.uk/members/scott/blogview?entry=20051108235707 and John Evdemon from Microsoft (and WS-BPEL Technical Committee Co-Chair) - eg http://blogs.msdn.com/jevdemon/archive/2005/09/12/464222.aspx
Although we didn't fully understand this when we first started the LAMS project 4 years ago, any "digital lesson planning" system (or if you prefer, Learning Design system), that actually runs lesson plans/LDs (not just models them) is likely to be something akin to workflow system "under the hood".
(As it happens, it was John Evdemon who helped me realise this - about 2 years ago I visited Microsoft to discuss LAMS, and John was brought in from outside the education area to have a look at it (given his OASIS role). I demonstrated LAMS, and he seemed pretty impressed, and then I remember he said something like - "What would be really nice is if a teacher could track what students were doing" (up to this point I hadn't shown LAMS Monitoring), and so I said "We've got something like that, let me show you..." and I showed him LAMS Monitoring. Once I'd finished demonstrating this, he then looked me in the eye and said "Who are you guys?", and in that moment I realised that what we had achieved in software was not trivial, and I felt alot better about how hard it had been to build LAMS! But I digress 
)
There is one huge area of (current) difference between LAMS/LD and traditional big IT vendor workflow systems - the role of multiple human actors at multiple steps of a workflow.
Most big IT vendor workflow systems focus on "system to system" workflow. If they include any human elements, these tend to be individual steps that only require a single human to be involved (eg, approving a process).
In educational workflows (ie, lesson plans/LD) you need complex co-ordination of multiple humans (teachers and learners; potentially across different roles, eg Chat & Scribe) across multiple steps of the workflow. This makes the software development task for educational workflow systems much more complex than for traditional workflow systems, which are already very complex!
In terms of standards, IMS LD is weak on detail about the workflow-like implications of running lesson plans/LDs. BPEL, by constrast, is pretty strong on detail for running a (business-style) workflow, but its scope is too narrow to deal with all the multi-person, multi-activity requirements of a typical lesson plan/LD. Part of our grant proposal involves some foundational abstract modeling of all of these issues to work out where the gaps are across current standards, and how this might be taken forward under a unified model. At a highly abstract level, WS-Choreography looks like the most promising framework, but our feeling is that it is too abstract, too early, and lacks the detailed implementation lessons that have informed BPEL on the one hand, and LAMS/LD on the other.
The importance of humans in workflows has recently become more recognised in the BPEL community with the launch by IBM and SAP of some BPEL extensions called BPEL4People (see http://xml.coverpages.org/ni2005-08-26-a.html for an overview). While this could grow into something that starts to meet the needs of lesson plans/LDs, it is still very early, and its main focus is not the multi-actor, multi-step processes needed in education workflows.
It's also worth noting, in passing, that some existing education specifications have some workflow-like bits hidden within them, such as IMS Simple Sequencing and IMS QTI.
So..... to answer the question in your post:
(1) Yes, we've looked at this, and hope to look at it more - and would welcome thoughts/collaboration with others in this space;
(2) While there is some prospect for using BPEL tools inside a LAMS sequence (see 3 below), wider compatibility will remain a challenge for some time, as BPEL tools tend not the understand multi-actor, multi-stage requirements that I believe are central to educational workflows/lesson plans/LDs;
(3) In the same way that LAMS V1.1 will provide a SCORM Run-time tool to allow you to plug-in SCORM objects into LAMS sequences, we may be able to build a BPEL-like run-time tool that helps run BPEL-style activities inside a LAMS sequence - but this needs some investigation to understand what is and isn't reasonably possible today.
One day (5+ years) I would expect that big IT vendor workflow systems may have advanced to the stage that they include rich functionality for multi-person, multi-role, multi-stage workflows. Once this happens, it may be possible to run a lesson plan/LD using these systems. It may also then be possible to present special "education-specific" authoring interfaces to these systems that hide all the complex IT workflow stuff, and show only the features that a teacher wants to use for educational workflows. But don't expect this to happen any time soon (as far as I know), because the corporate IT focus on workflow is still quite different to the requirements of education workflows/lesson plans/LDs, and building software with this level of flexibility is hard. So until these systems catch up, I think LAMS has alot to offer you won't find anywhere else.
Posted by James Dalziel
