[nullflux]

Not quite sure this applies too much in startupland. There does appear to be a shift away from traditional IA practices with most startup designers, but largely because it takes too damn long and really isn't that necessary.

Proper IA/wireframing/signing-off-on-steps is good for companies where a clueless management is going to make myriad changes to information hierarchy or UI bits along the way, because you have a "we talked about this already and you agreed" defense when some clueless marketing guy comes over and says things need to be bigger, this link needs to be here because it's his friend's company, etc.

A lot of what this article is proposing is missing (organizational hierarchy, an understanding of the order of information) is something that I don't really think is missing in a good tech-company designer's head. He or she is often one of few in control of that domain, and the product roadmap is often known quite a bit in advance. A lot of the polar bear book's 1000+ pages ends up getting turned into heuristics, when quick decisions and fast iteration under multivariate testing is often better off for a young company than sitting back and doing some type of holistic, IDEO-class set of mental exercises.

[Joeri]

In startup land the products usually are simpler. Below a certain order of complexity most of the "best practice" is unnecessary or even harmful.

For projects measured in manyears instead of manweeks or months, a solid IA is essential.

[lishubert]

I think proper Information Architecture is extremely important especially in the startup land. Proper not necessarily meaning formal, structured, long winded process, but more along the lines of though of how this virtual place called the startup product should be structured, as much as one can in the crazy land of every changing, fast paced stuff.

I would agree that a lot of the stuff that is missing can be in a good tech-company designer's head! Totally with that.

[nullflux]

Sure. On the flip side of my argument there is also something to say: quite a few startup designers could learn a lot from following such a process at some point. There's a lot of "copy this guy's design, it converted well for his product!" that ends up leading people down roads that aren't exceptional for that product.

But can we say there's really a "devolution"? Is there not progress in not following Morville and Garrett to the letter? I find a lot of dogma in "UX" practices that make designers feel comfortable, and I just as often find many UX designers to overstate their impact on things because it is not as easily quantifiable as an engineer's impact.