We had a meeting today to discuss the architecture of the OU VLE. I was the project director during phase 1, and we made the recommendation that we should adopt a service oriented architecture. This was partly a pragmatic decision because although we didn't have a VLE as such, we had over the years developed, or bought in, a number of the components, including conferencing, assignment handling, authentication, etc. It was also partly a recognition that this was where the world was moving to. I made the analogy today that it is like the claim that many sociologists and economists make (I think Castells is amongst them), that many developing countries can now skip the industrial revolution and go straight to a knowledge economy (not sure how true this is, but for the sake of the analogy we'll go with it). The OU could effectively skip the 'monolith' stage of VLE deployment and go straight to SOA.
However, we did end up opting for Moodle. I think this was an entirely reasonable compromise - it gained us time in that it already had some functionality we needed, it gained us a technological method (instead of arguing how to do things, we knew we had to do them the moodle way - this 'hard target' for integration is important) and it gained us some kudos (not to be underestimated). And at the same time we still had access to, and some control over the code. However, it did represent a slight compromise on the pure SOA vision. So today's meeting was to discuss the extent to which that has happened, and where we want to head in the longer term.
Overall I came away happy that we were moving in the right direction, and that Moodle remained the sensible option. As long as we develop with this in mind, and don't take too many shortcuts, the overall service approach is not compromised too much. It is also an issue that lots of people are grappling with I think. Joel Greenberg and Jason Cole, who were at the meeting today, will be discussing these issues at Alt-i-lab this year.