Good point about simplicity and the Twitter example.

In the past component have tended to be complex and grow to try and do everything – interfacing needed (needs) specialists – a gravy train for the IT service/support industry.

If user “components” are simple and have simple connectivity then this will help plugging them together – standard interfaces etc.

Thanks for the blog – you have given me some ideas to feed into something I may try to write soon on this type of subject.

Look at the success of Twitter: this lies not in having one system that does everything, but having a simple system with a stable and well defined interface to enable other bits to be bolted on at will.

Technologies like OpenID, Shibboleth and OAuth should make this easier in the future as they become more widely adopted.

There is a question of ownership of one’s identity. I’d love, for example, for the university to make use of OpenID. But this begs the question: what do we do when we move on? If the university is my OpenID provider, how do I take my identity (which is theoretically inseparable from me) with me? And if I use a third-party identity provider, how can I convince the university that it is really me?

The problem with a single product is with that it can be slow and difficult to change and there is reduced choice – “one size fits all”.

The problem with coupled separate products is with integration.

I’m increasingly a fan of loosely coupled approaches – keep the core minimal and enough to hook up whatever teachers and students wish to use.

I see signs that loosely coupled approaches will become easier with developments in on-line identity and authentication.

I’m quite a fan of people using their own resources and identity wherever possible rather different on-line identities in different roles – work, study, self – especially if you study and work in different roles.