Erik has finally started blogging! So far there are some interesting posts related to Architecture and Web 2.0. I particularly like the discussion on “perspectives on architecture” (see here) as this theme pops up in my work all the time. Last week, even, I gave a presentation about a project to harvest architectural guidelines which we use for the infamous ProjectStartArchitecture and ended up debating the difference between architecture and architecture description. Erik ends with the statement that “It’s not your definition of architecture that matters, but what you do with it” and frankly I couldn’t agree more 
Sometimes one has to be pragmatic, though.
There is one post that particularly drew my attention. In this post Erik presents (part one of) his perspective on Web2.0. As such he argues that the Web2.0 needs information architects who, among other things, tackle the fundamental problem of the distinction between content and presentation. Since we wrote several papers on Transformations in a Web context I would like to comment on Erik’s post here (See my publications page for some papers on transformations).
I agree with Erik that the Web2.0 could use some fundamental redesign and that one important design principle would be: decouple content from presentation. Fair enough. Simple enough. Once you think about it, things are not as easy as they might seem. From a usability perspective, one could argue that a second principle is relevant here too: (logically) store content only once. I’m referring to the situation where we (Erik and me) have several multi-modal discussions (face to face, blog, E-mail, Jabber) and I would like to store “the content” of our discussions in one logical place. For face to face communication this might be tricky to say the least, but that’s a different discussion.
These -and probably other principles as well- restrict our design freedom with respect to the technical infrastructure of the Web. In more technical terms, I would say that as a corollary from the second principle we would need a lingua franca: a common format (in terms of our search: data resource type) to store content. A second corollary of these design principles is the fact that we need Transformations which transform content from one mode to another. In terms of the theory presented in our research / my dissertation: we would need transformations that transform between the different modes of communication online. Examples (formally):
T1 such that SEM(T1) = blogpost → linguafranca
T2 such that SEM(T2) = linguafranca → blospostT3 such that SEM(T3) = Email → linguafranca
T4 such that SEM(T4) = linguafranca → Email
In terms of software architecture some interesting observations can be made as well. Principles that apply are, for example, the princple of loose coupling. Several patterns apply also, such as the gateway pattern and the translator pattern. Consider the following example to illustrate my point: let’s say Erik and myself use blog postings and E-mail as our communication channels. Then we would need two ways to interface with our “lingua franca content store” (the store). For blog postings we’d have to tap into the Web server to wire it up with the store (and execute a transformation along the way). Similarly, for Email we’d have to tap into the mail server (i.e., postfix) and make sure it is also wired up to the store in a similar fashion. Luckily modern Web/Mail servers have hooks to achieve this more easily.
In short, I agree with Erik. The Web 2.0 could definitely use some re-design and a solid information architecture. Perhaps we should round up some students to dive into this topic?