Elias spends his mornings in a workshop that smells like burnt ozone and stale 1950s oil. He restores vintage espresso machines-the heavy, chrome-clad monsters from an era when “planned obsolescence” wasn’t a phrase anyone had invented yet.

If a gasket fails on a 1954 Gaggia, Elias doesn’t open a browser window. He doesn’t categorize the failure as “Level 2 Hardware Malfunction” or wait for a routing algorithm to assign him a specialist. He picks up a rotary phone and calls a man named Marco in a basement outside of Milan.

Marco is eighty-one. He doesn’t have an email address. He has a lathe and a memory that functions like a library of every brass fitting ever cast in Northern Italy. This relationship is what a modern management consultant would call a “single-point-of-failure risk.” It is unstructured, unmeasured, and entirely outside the purview of an ERP system. But it is also the only reason the espresso machine works.

This is the fundamental friction of the modern office. We have mistaken the map for the territory, and in our desperate quest to make every human interaction “scalable,” we are accidentally lobotomizing our organizations.

I realized this recently when I was staring at a support dashboard that was glowing a serene, deceptive green. For three years, I had been calling the “epitome” of something the “epi-tome,” like it was a very large book. I said it in meetings. I said it to my boss.

It was a small, humiliating realization that I had been operating with a fundamental misunderstanding of the mechanics of the word, despite using it every day. Our ticketing systems are the “epi-tomes” of this kind of error. They look like the right word, they occupy the right space, but the internal logic is fundamentally broken.

The Living Ecosystem of the Support Floor

Before the “Transformation,” our support floor was a mess of noise. It was a living, breathing ecosystem. If an agent at a desk encountered a complex payments issue-something that didn’t fit the usual script-they didn’t send it into a digital void. They stood up. They looked around.

They knew that Lek, two rows over, had been with the company since the servers were just a stack of towers in a ventilated closet. They walked over to Lek.

“Hey, the transaction 402 is hanging. Is it the gateway or the user’s bank?”

“It’s the gateway. They’re running a maintenance patch until 2 PM. Just tell them to retry after lunch.”

– Lek, Expert Analyst

Total time elapsed: 45 seconds. Resolution: Complete. Human connection: Reinforced.

Then came the new ticketing system. The “Chaos” was deemed unacceptable. Management wanted data. They wanted to know the “Average Handle Time” and the “Resolution Rate per FTE.” They wanted a clean floor where everyone stayed in their seats and spoke through the software.

Now, when that same payments issue happens, the agent opens a ticket. They have to select a category from a dropdown menu that doesn’t actually include the problem they are facing. They pick “General Technical Error.” The ticket goes into a queue.

The queue is managed by a routing engine that sees Lek is currently “at capacity” with thirty other tickets, so it routes the payments issue to a junior agent in a different time zone who has never heard of the gateway maintenance.

The junior agent asks the customer for a screenshot. The customer gets frustrated. The ticket bounces back and forth for three days. Eventually, it reaches Lek, but by then, the “Resolution Rate” metric is already bleeding red, and Lek is too burnt out by the sheer volume of “General Technical Error” tickets to provide the nuance he used to offer over a shoulder.

The org chart looks beautiful on a slide. The silos are perfectly defined. But the invisible web-the actual structure that kept the company alive-has been severed.

The Invisible Web of Transactive Memory

This invisible web is composed of what sociologists call “transactive memory.” It’s the collective knowledge of who knows what. When you systematize a human network, you often delete the “who” and replace it with a “what.” But “what” doesn’t have intuition. “What” doesn’t remember the maintenance schedule from three years ago that always causes a glitch in November.

Management sees the unstructured floor as a liability because it’s hard to measure. You can’t put “Lek’s brain” on a spreadsheet. You can’t “scale” the fact that Sarah in accounting knows exactly why the third-party API fails on Tuesdays. So, we replace Sarah and Lek with a workflow.

But a workflow is a static thing. It is a series of “If/Then” statements designed by someone who likely hasn’t handled a customer call in five years. It assumes that problems are predictable. But the real world-the world of payments, and hardware, and human frustration-is an “If/Maybe” world. It requires the ability to navigate the gray areas.

I’m sitting here now, looking at my own workflow, and I realize I’ve been doing it too. I’ve been trying to “systematize” my thinking to the point where I don’t even trust my own instincts anymore. I wait for the data to tell me what I already know. It’s like the “epi-tome” thing. I was so focused on the structure of the word that I missed the sound of it.

We are building digital cathedrals of process while the actual faith is leaking out of the building. The more we “route” our problems, the less we solve them. We’ve created a world where everyone is busy, but nothing is happening. We are all moving tickets from one side of the screen to the other, while the espresso machines of our industry are sitting cold and broken because no one knows how to call Marco anymore.

Restoring the Order of the Solution

The irony is that the “chaos” was actually the highest form of order. It was a self-organizing system that prioritized the solution over the record of the solution. Now, we prioritize the record. We have the best data in the world about why our customers are unhappy, and we are too busy filing that data to actually make them happy.

If you want to know how a system actually functions, don’t look at the manual. Don’t look at the Jira board. Look at the paths worn into the carpet. Look at the desks where people congregate. Look at the people who don’t use the ticketing system because they’re too busy actually doing the work.

We need to stop trying to “fix” the noise. The noise is the sound of people thinking. It is the sound of Lek telling someone to wait until after lunch. It is the sound of expertise finding its way to the problem without needing a permission slip from a piece of software.

If we keep “optimizing” the humanity out of our systems, we will eventually find ourselves with a perfectly efficient organization that doesn’t actually do anything. We will have 100% visibility into a vacuum.

I think I’ll go find a Marco. Or a Lek. I’ll walk over to their desk, I’ll ignore the “In-Office Messaging” app, and I’ll ask a question. I might even mispronounce a word or two. But at least the machine will start running again. And in the end, that’s the only metric that actually matters. The rest is just a very expensive way to stay quiet.