The cursor is blinking, a rhythmic, taunting heartbeat on a screen that has just told me I don’t exist. Or, more accurately, that the version of me encoded in a 2-page PDF doesn’t meet the ‘threshold of viability’ for a role I’ve literally performed for 13 years. It’s a strange sensation, sitting in a swivel chair that squeaks exactly 3 times every time I shift my weight, realizing that a series of weights and biases in a neural network has decided my professional trajectory is a non-sequitur. I had hiccups during a presentation last week-real, bone-shaking hiccups that made 13 senior partners look at me like I was a glitching android-and even that felt more human, more ‘data-rich’ than the cold rejection I’m staring at now.

We’ve convinced ourselves that by removing the human from the first gate of hiring, we’ve removed the rot of prejudice. We call it ‘data-driven recruitment,’ a phrase that carries the same weight as ‘evidence-based medicine’ or ‘structural integrity.’ But as any carnival ride inspector will tell you, the integrity of a structure depends entirely on who’s checking the bolts. Iris K.L., a woman who spends her days climbing the rusted ladders of the ‘Tilt-A-Whirl’ and the ‘Megadrop’ in traveling fairs across the Midwest, once told me that a sensor is the most honest liar you’ll ever meet. She’s seen sensors that say a ride is locked and loaded when the actual steel pin is 3 millimeters away from shearing off. The data says ‘Green,’ but the reality is a catastrophe waiting for a scream.

In the world of hiring, we are building ‘Megadrops’ at scale. We’ve taken the messy, biased, often-drunken logic of human intuition and tried to distill it into a scoring rubric that a machine can execute. But here’s the contradiction I can’t quite shake: we hate the bias of the old-school ‘golf club’ hiring network, yet we’ve built a digital version that is arguably more exclusionary because it lacks the one thing humans occasionally possess-the ability to be moved by a story. We’ve replaced the ‘good ol’ boy’ with a ‘good ol’ algorithm’ that has been trained on the resumes of the people who already hold the jobs. If the current winners are 83% likely to have played lacrosse or attended a specific three-letter university, the machine doesn’t see a correlation; it sees a requirement.

I’m not saying we should go back to the days of smoke-filled rooms and ‘gut feelings’ that usually just meant ‘this guy reminds me of my nephew.’ But the current obsession with objectivity is a ghost story. We want to believe the data is neutral, but data is just a fossilized version of yesterday’s decisions. When an AI screening tool rejects a candidate because they took a 123-day gap to care for a sick parent, the machine isn’t being objective about their skills; it’s being objective about a rubric that someone-likely a 23-year-old developer who has never managed a budget-decided was a proxy for ‘commitment.’

Iris K.L. tells a story about a Ferris wheel in a small town that kept tripping a safety alarm. Every 3 hours, the ride would emergency-brake, leaving 23 people dangling in the humid air. The data logs showed a power surge. The technicians replaced the motors, the wiring, and the control board. $83,003 later, the ride still tripped. Iris finally sat in the operator’s booth for an entire shift. She noticed that every time the town’s fire siren went off, the operator jumped, his foot hitting a poorly placed manual override switch. The data was ‘correct’-there was a surge of input-but the data couldn’t tell the story of a nervous man and a loud noise. We are currently firing the candidate because the fire siren went off, and we’re calling it a ‘data-driven optimization.’

This is why I find the modern interview landscape so bafflingly disconnected. We prepare for these algorithmic gatekeepers by trying to become more like them. We keyword-stuff our histories and sanitize our failures. We treat our career paths as a series of 43-point bullet lists. But the real problem isn’t the candidate’s lack of data; it’s the company’s lack of context. When the algorithm fails, it fails at a scale that should terrify us. If a human recruiter is biased, they might ruin 33 opportunities a month. If an algorithm is biased, it can systematically erase an entire demographic from an industry in 3 seconds.

I remember talking to a software engineer who was part of a team building a ‘productivity tracker’ for a major warehouse. They used 233 different data points to determine who was a high performer. One of those points was the speed at which a person walked between stations. The data suggested that fast walkers were more productive. What the data didn’t show was that the fastest walkers were also the ones most likely to quit within 63 days due to burnout or joint pain. The ‘objective’ metric was optimizing for a workforce that would eventually cease to exist. They were building a machine that was eating its own gears.

Navigating this requires a level of tactical awareness that most people simply aren’t taught in school. You have to understand the ‘Leadership Principles’ or the ‘Core Values’ not as abstract virtues, but as the literal code the machine is looking for. Firms like Day One Careers have built an entire methodology around this realization-that the interview is no longer a conversation, but a data-transmission event. If you don’t speak the language of the machine, your brilliance is just background noise. It’s a cynical view, perhaps, but it’s the only one that acknowledges the reality of the 503-error world we live in.

I’ve spent the last 3 hours looking at my own ‘data-driven’ profile on a major recruiting platform. Apparently, I am a ‘73% match’ for a role as a logistics manager in a town I’ve never visited. Why? Because I once wrote an article about Iris K.L. and used the word ‘shipping’ three times. The machine doesn’t know the difference between shipping a cargo container and shipping a narrative. It just sees the string of characters. This is the ‘objectivity’ we are betting our careers on. It’s a hollowed-out version of truth that values the frequency of words over the depth of experience. It’s the digital equivalent of Iris’s rusted bolt-it looks fine on the readout, but if you put any real weight on it, the whole thing might come crashing down.

We need to stop pretending that adding more data points makes a decision better. Sometimes, it just makes the bias harder to find. It buries the prejudice under a pile of 1,203 spreadsheets until nobody remembers why the ‘Ideal Candidate’ profile was created in the first place. Was it based on the person who actually did the job well, or was it based on a dream of a person who doesn’t exist? Most of the time, it’s the latter. We are hiring for ghosts, using tools built by people who have never met the spirits they are trying to conjure.

I think back to my hiccup-filled presentation. In the moment, it felt like a disaster. I was 43 minutes into a pitch for a new project, and my body decided to betray me. I could see the data-driven minds in the room calculating my ‘professionalism’ score in real-time. But then, something strange happened. One of the partners, a man who usually looks like he was carved out of a block of ice, started laughing. Not at me, but with a sort of weary recognition. ‘I had those for three days straight during my wedding,’ he said. The tension broke. We stopped being data points and started being people. We closed the deal, not because my metrics were 13% higher than the competition, but because we shared a moment of fallibility.

An algorithm would have just marked me down for ‘vocal instability.’

There is a specific kind of arrogance in thinking we can code our way out of the human condition. Iris K.L. knows this. She carries a physical hammer, a small one with a wooden handle that has been smoothed by 23 years of use. She taps the steel. She doesn’t just look at the sensor; she listens to the ring of the metal. If it sounds ‘thuddy’ instead of ‘bright,’ she shuts the ride down. No data log can capture that sound. No AI can feel the vibration of a failing weld through a wooden handle. We are losing the ‘ring’ in our hiring process. We are replacing the hammer with a screen, and we’re wondering why the rides keep breaking.

The myth of the data-driven hire is that it protects the company. In reality, it protects the people who make the decisions from the consequences of being wrong. If you hire someone who fails, but they were a ‘93% match’ according to the software, you can blame the software. If you hire someone because you liked their story and they fail, you have to blame yourself. We have traded accountability for the illusion of accuracy. We are building a world where no one is responsible for anything, because ‘the system’ said it was the right move.

I’m going to go get a glass of water, held upside down, to make sure these hiccups are gone for good. Then, I’m going to rewrite my resume. Not because I’ve changed, but because the ghost in the machine needs to be fed its specific keywords. I’ll make sure to mention ‘structural integrity’ at least 3 times. Maybe I’ll even mention Iris K.L., though the algorithm will probably think I’m talking about a new type of infrared sensor. It’s a game we all have to play now, a carnival of sorts where the prizes are jobs and the rides are built on data that no one quite understands. Just hope that the person inspecting your ‘bolts’ still knows how to use a hammer, and isn’t just looking for a green light on a dashboard that doesn’t know the fire siren is about to go off.