Information architecture is not set in stone. Done well, it will live longer than the raw data, which is moving and accumulating through time at the speed of time and the amassing of a quantity of information driven by the quantity of people. But as the data shifts, as our data substrate and understanding grows, as ideologies are finally seen and sloughed: the information architecture needs to change with it.
We tend to look at big infrastructure projects as one-and-dones. We built a glorious roadway system in the middle part of the 20th century that is crumbling around us. We built an energy transmission system that is allowed to keep chugging along until it fails, like in the California Camp Fire (2018) or the Texas cold snap (2021). Old water mains break, and can failed more spectacularly through additional mismanagement like in Flint, Michigan (2014).
Recently I even ran across something that worked so beautifully that, with the pressures of how capitalism works without a societal backstop, the information and understanding to fix it is nearly lost: steam heat. Think about that. Lost skill and understanding about a tool we continue to use today, and continued through from inception to now. How is this not going to happen in all of our information building that is being black-boxed into LLM’s?
LLM’s are worse than losing the understanding that built steam heat. The information to rebuild understanding of steam heat is still deep in archives. We don’t have to figure it out from experimenting with the tool itself, explosions and all. With LLM’s, they are hiding the supporting information and injecting it with false information. When what is built with them in the mix breaks, after the existing knowledge today dissipates through time, that’s the end of it. We are extracting from future generations with this technology.
The behavior around each instance is interesting. But the key bit, that transcends motive?
We spent all this time and money on these things, generations ago. Their function was a trusted stability point. They worked. The perception was they would continue working, just like the sun would continue to rise in the east, because they didn’t break quickly. Maintenance disrupted what needed to function optimally; don’t fix what isn’t broken.
Don’t doubt that information architecture is an infrastructure project. It has no definable ROI by today’s standards. How it adds to the running of business and accumulation of profit doesn’t work by the assumed metrics. It touches everything, it can’t be silo’ed to prove what single line item it should be attributed to. Done well, information architecture decreases stress, makes it easier to shift processes, and trace problems to fix them.
The accountability to fix things even when I first drafted this piece 3 years ago (it’s part of Movements) was meaningful in business. There were pathways and power structures, and if a big problem was traced to one particular person it could lead to firing. Business is in the process of abdicating responsibility for their own information with LLM’s, which is something that deeply disquiets me.
With LLM’s getting deeper and deeper into the infrastructure, business has abdicated responsibility. Do they feel like they can’t fire a machine. It’s easy: unplug it. The clear differences is that they accept the mistakes and move on. Which implies that it was never about the specific mistakes, but about wielding power, enacting retribution, making sure the mistake didn’t stick to them. As long as there is someone/thing else to blame – Todd from accounting, or that pesky LLM – the responsibility is abdicated by the person who was supposed to be watching.
Just like a roadway, people will use it without really thinking about the work that went into the initial building, the maintenance, and how long it will survive. They use it, so frequently it’s a stability point in their world — inconceivable that it’s not already perfect or might break, impossible that they might need to continue paying for it, annoying that so many other people are congesting it’s built pathways.
Use of the infrastructure can degrade it. It can degrade physically, by finding it doesn’t meet needs as expected, by realizing that there were false assumptions built in. With information architecture, as it’s currently built most often, it degrades as we grow our understanding and as our cultures shift.
Repetitive use of infrastructure, like when someone has driven to work the same way a hundred times, can confabulate easy with, “I’ve built all the workarounds I need.” What’s known is familiar, calming, and can be a moment of peace in an otherwise chaotic day, even if it’s held together with duct tape, hope, and recognition.
There will also be a cohort who see the existing form as less risky than all the variables involved in change. Why close half of a busy roadway for anything in less than clear, existing failure?
Information architecture is abstract. While it has a lot of sympathy with how with think about roadways, it has one very clear difference: it’s intangible. We don’t have to set aside millions of acres to work on it as a holistic beast. Nor is it like the busy city blocks that will back up traffic all day if a few feet are closed. It’s great if we have a huge whiteboard and a stack of cards and post-it notes, but reality is all we need is our preferred tools, an active mind that cares about both information and people, and access to people, information, and data to build shared architectures. For our individual ones, we just need a mind with a halfway decent memory.
Most of what I’ve shared on this site isn’t prescriptive. It’s a few symbols and patterns and some knacks to help think about people as people, and not their labels and categories (which are fleeting and predicated on culture). It’s not even getting into the really complex stuff, but just an introduction to what pieces I most tend to pick from as I work on an information architecture project. People are intrinsic because they are the context for information. People are intrinsic because most of the information we’re interested in is supplied by other people. Information has meaning based on context, within itself and for whom it’s intended.
Most information architecture projects are not holistic. They don’t need every piece, but can be pulled together for targeted use cases. What we most need to remember is that people are multivariate. For instance, the book format as a whole was structured as a linear experience, front to back, and a network still existed within. The site version is network nearly tipping over into system as I consider adding to it. Simple and complex, light and dark – not as binaries, but as spectrum nodes. Given enough time, thought, and iteration, information architecture can be as complex as it needs to be.