The trouble with defining Information Architecture

ADDED NOTE: I have a more recent post than this where I finally manage to sort some of it out. See it here: A Layer Model for the IA Profession.

I’ve been letting these thoughts roll around in my head for a while, but haven’t jotted them down. I figure I’ll jot them here. Lacking the energy to explain them much, this really is just a jot … but hey, it’s my blog.

There’s been an ongoing issue among “information architects” (or at least those who call stuff they do information architecture) regarding “defining the d**n thing.” It rears its convoluted head nearly in sync with phases of the moon, especially on the IA-related mailing lists.

This is in no way an argument for any particular definition. Just thoughts on why we can’t seem to get there.

Let me start by saying that I do think Information Architecture is an appropriate term — that what we do is much like traditional architecture, at least conceptually. We design shared environments; they just happen to be made of bits, not atoms. Frankly, for me, that’s enough … but if you want to start making official curricula, licensing, job descriptions, etc, it of course gets stickier. So …

1. Distinguishing between the “title,” the practice, and activities within and related to the practice is essential before we can even discuss the issue. In one frame of reference, one definition makes some sense, whereas in another it just doesn’t. (Saying that card sorting is an IA-related activity is clearly true, but saying that IA as a practice or discipline is defined by card sorting just going the wrong way up the hill.)

2. Unlike traditional, physical architecture, which is essentially about walls, information architecture is essentially about links. IA is an essentially interstitial concern. Whereas physical architecture’s final form can be noticed because you end up with a visible container, the “spaces” we make are defined by stuff that, when we do it well, is invisible. (Yes this is a simplification, but I warrant it may have merit.) True the best architecture concerns how a building is used, not just what its walls look like. But even the most usable buildings are discussed in architecture criticism (and noticed by people as something with identity and value) as physical objects with shape. If the inside is made so they don’t have to think about how it’s made, excellent … but that’s not the part people talk about. Information architects, however, don’t get to skin their buildings with post-ironic waves of aluminum, or in sheets of shimmering glass. (I’m not criticizing all architecture here, ok? Much of it’s quite beautiful and/or usable.) So, whenever we try pointing to something concrete as “that’s IA” some other discipline can easily say “hey, um, no, that’s what *we* do.” Why? Because they probably have a point. IA done well makes sure that A and B are designed in a way that makes semantic relevance live on their surfaces, and that getting from A to B and back again (or to whatever else) is also relevant, functional and valuable. This of course influences the visible design, but I think we mainly act as consultants to the craft of interface design here — not as the primary actors (at least as concerns the role of practice of IA itself — of course any individual may be doing both the interface design and the info-architecture, but that’s more a question or roles and responsibilities, not the definition of one practice in contrast to another).

3. Information architecture is also basically about making things out of language — but not only language as a concrete (if it can be said to ever be concrete) but out of the relationships between bits of language. That is, we make things out of semantic relevance. It’s inherently ethereal, and places us in a weird tautological relationship to our raw material — how does one define it, if it is made of the very stuff we use to define things? (Unlike, say, linguistics, which is a discipline *about* language. Ok, this distinction doesn’t hold up written down, but it does in my head, so I need to figure out the diff.) We end up in a strange Heisenbergian conundrum: every time we try manipulating a bit of this semantic goo in order to shape a definition, we change its relationship to us for that moment, and the relevance seems to sort of slip through our fingers.

I’m not sure if any of that made sense, but it’s been itching at my frontal lobe. And my frontal lobe has enough trouble keeping up with regular daily life.

Author: Andrew Hinton

I use information to architect better places for humans. More at