The Caffeinated Firefly

I am currently staring at a pixelated mess on a seventy-six-inch monitor while my boss, a man who considers a pivot table a religious text, explains why a flatline is actually an upward trajectory waiting to happen. The laser pointer dances across the screen like a caffeinated firefly, resting eventually on a bar chart where the Y-axis has been so violently truncated it looks like it was cut with a rusty hacksaw. He is showing us a two percent increase in ‘engagement’ as if it were the parting of the Red Sea. Everyone in the room is nodding. I am nodding too, though mostly because the air conditioning is set to a sub-arctic temperature and my neck muscles are starting to seize.

Quantitative Fetishism and the Glass Conservator

We live in an era of quantitative fetishism. We believe that if we can measure it, we can control it, and if we can control it, we can’t be blamed when it fails. But the measurement itself is often a performance. It reminds me of my friend Noah Y., a stained glass conservator I spent a weekend shadowing in a drafty cathedral basement. Noah is fifty-six, with hands that look like they’ve been scrubbed with pumice for three decades. He doesn’t care about ‘optimization.’ He cares about the structural integrity of lead cames and the way seventeenth-century glass ripples when the sun hits it at exactly four-forty-six in the afternoon.

Noah Y. once told me that the most dangerous thing you can do to a window is try to ‘fix’ it based on a photograph. “A photo only captures the light from one angle,” he said, scraping a bit of ancient putty from a groove. “It lies to you about the weight. You have to touch the glass. You have to feel where it’s thinning out.” Our corporate dashboards are the photographs. They give us a flat, two-dimensional representation of a three-dimensional human mess. We look at a drop in conversion and see a technical glitch or a pricing error, ignoring the fact that maybe our customers just felt insulted by the latest marketing copy.

We spent ninety-six thousand dollars on a data attribution model last year. It was supposed to tell us exactly which dollar of spend resulted in which dollar of profit. It was a beautiful, complex machine of logic. But when the results came back and suggested we should stop spending money on the CEO’s favorite vanity podcast, the model was suddenly ‘under review.’ We tweaked the parameters. We adjusted the weights. We massaged the inputs until the data confessed to exactly what the leadership team wanted to hear. This is the ‘yes-and’ of corporate aikido. We don’t say the data is wrong; we say the data needs more ‘context,’ which is just a polite way of saying we need to find a different set of numbers.

The Soul of the Window

Noah Y. doesn’t have a dashboard. When he’s restoring a panel, he uses a small hammer and a lot of patience. He told me about a job where he had to replace six tiny shards of red glass in a depiction of a saint’s robe. The original glass was made with copper oxides that don’t exist in the same way anymore. He could have used a modern substitute that looked ‘close enough’ on a spectrometer. Instead, he spent six weeks hunting through salvaged crates of Victorian glass to find a match that felt right in the hand. “The numbers might say the refractive index is the same,” he told me, “but the soul of the window knows the difference.”

We’ve lost the soul of the window. We’re so busy measuring the transparency of the glass that we’ve forgotten what we’re supposed to be looking at. This is where the frustration boils over. We claim to be ‘data-driven,’ but we use that data the same way a drunk uses a lamppost: for support rather than illumination. We find the data that justifies our existing biases and discard the rest as ‘noise.’ It’s a closed loop of self-congratulation.

Arithmetic as a Blindfold

I remember one specific campaign where the data was screaming at us to stop. The churn rate was up sixteen percent. The customer support tickets were overflowing with people who felt cheated. But the ‘North Star Metric’-which some genius had decided was ‘Total New Accounts’-was still green. So we kept going. We kept pouring money into a leaking bucket because the one metric the board cared about was looking healthy. We destroyed the long-term trust of our base for a short-term bump in a slide deck. We called it a ‘calculated risk.’ In reality, it was just arithmetic used as a blindfold.

I think about the 1926 window Noah was working on. It survived two world wars and a structural collapse, not because it was ‘optimized,’ but because it was built with an understanding of tension and grace. It wasn’t trying to ‘convert’ anyone. It was just being a window. There is a lesson there for those of us drowning in data. The things that truly matter-trust, craft, the way a person feels when they interact with your work-are often the things that are hardest to put into a column.

Admitting the Mistake

I’m going to go home and delete my cookies. I’m going to let the algorithms forget me for a night. I want to see something that hasn’t been A/B tested to death. I want to make a decision based on the sixty-six percent of my brain that doesn’t care about a dashboard. I’ll probably be wrong. I’ll probably make a mistake that costs us a fraction of a percent in ‘conversion.’ But at least it will be my mistake, and not a lie told by a chart with a broken axis.

In the end, we are the ones who have to live with the consequences of our ‘data-driven’ delusions. The numbers won’t be there to hold our hands when the customers leave. The pivot tables won’t apologize for the bridge we burned. We are humans using tools, not tools being used by numbers. It’s time we started acting like it, even if it means admitting that the Y-axis doesn’t always have to point up.