What training a pigeon taught me about bug tickets
I spent a semester at NYU training a pigeon. Here’s what it taught me about troubleshooting bug tickets.
Each of us got one pigeon and a semester to train it to perform a sequence of three actions in a specific order and direction (eg; lift left wing). Training took place in a Skinner box, which is.... a box with a viewing window and a remotely triggered food hopper. My class was right before feeding time.
Day 1: Desensitize the pigeon to the sound of the hopper so it stops panicking and learns that clunk = food. You can see why this took a full semester. After that, you watch your pigeon and reward any behavior that approximates the sequence you want. Over time you raise the bar until exam day when (hopefull) your pigeon performs its routine for your professor.
The warning from my professor: pigeons develop superstitions. If you reward a pigeon right after it turns left, but its right wing happened to be raised at that moment, it might assume the raised right wing was the key to the reward. Forever after, your pigeon struts proudly, right wing up, a victim of false causality.
That’s exactly what can happen in bug reports. “When I set the appointment Medium to Zoom, the Save button is disabled and I can't save the record.” The confident tone makes you think the Medium causes the bug. But often it's a superstition, a coincidental pattern mistaken for causation. The real cause might be something else entirely (say, a data dependency on the selected client).
Users aren’t QA professionals. They notice symptoms, but they may not systematically test the issue, that's our job. And if you too skip that step and go along with their assumed causality, you or someone else may waste a lot of time going down a rabbit hole, looking for the line of code that links the save button to a raised right wing.
I suspect that's a major shortcoming of AI in customer support roles. They're inherently trained to go along with their instructions, so if the person submitting an issue includes false causality, I'm guessing most AI support goes down a very annoying path.
I forget my pigeon’s name, but I’ll never forget what it taught me: we’re wired to see patterns, and it’s our frontal lobe’s job to discern signal from noise.

Jessie Grenfell
Jessie has spent 20+ years architecting and delivering software for organizations where complexity is the norm — financial services, insurance, nonprofits, and regulated industries where getting it wrong isn't an option. She specializes in the hard problems: unique operational workflows, cross-system integrations, and the kind of compliance requirements (HIPAA, CCPA, SOC 2) that most developers treat as someone else's job.
At Codality, she leads every principal engagement from discovery through delivery — because she's seen what happens when that thread gets cut.
