I'm in Brussels this week at EuroOSCON, and yesterday I was fortunate enough to give a tutorial session on "Building Better Ajax Applications". The slides are online, but don't try to view them with anything but Firefox 1.5 or later.
For the last year or so I've been occasionally giving a talk like this and this time I decided to overhaul the whole thing. Every time I've done the talk I admonish folks not to use Ajax just because they can, and if at all possible to work with interaction designers. Last week at the "Future of Web Applications" conference in SF, Jeff Veen boiled the key points of the ID problem down into the most coherent set of objectives I've heard. He outlined them as:
Having worked with a wonderful set of interaction designers (and now friends) at Informatica, I've got nothing but the deepest respect and admiration for the questioning eye that good user-centric design brings to the process of building applications. Jeff's description of the problem instantly made sense, and I've tried to put the example code in section 2 of the tutorial into the framework of those points. The distillation of the problem into easily digestible terminology is one of the things that was new with "Ajax". Hopefully Jeff's list of the 4 goals of ID can help developers similarly start to frame the problem better.