Gianluca Gimini, Velocipedia (2016)
In 2009 an Italian designer named Gianluca Gimini started asking people to draw a bicycle from memory. He kept doing it for years, collecting hundreds of drawings. Most of them were wrong. People would put the chain on the front wheel, or connect the handlebars directly to the rear axle, or draw frames that were physically impossible to ride. These were people who had seen and ridden bicycles their entire lives. Some of them biked to work. They could recognize a bicycle instantly, pick their own bike out of a rack of fifty, tell you the difference between a road bike and a mountain bike. They just couldn't reconstruct one from the parts up.
Gimini built realistic 3D renders of the drawings, and the results look like something out of a fever dream. Frames that would collapse under any weight. Pedals attached to nothing. Wheels that couldn't turn. Each drawing was made with total confidence by someone who would have told you, if you'd asked, that they knew what a bicycle looked like.
There's a term in psychology for this: the illusion of explanatory depth. People consistently overestimate how well they understand things they interact with daily. Toilets, zippers, helicopters, government policy. When asked to explain them in detail, confidence collapses. The knowledge was never there. What was there was a feeling of knowing, which turns out to be a different thing entirely and much cheaper to produce.
Early in my career, I shadowed interviews at startups and I kept seeing the same thing happen. A candidate walks in with clearly decades of experience on the interviewer. Their resume is heavy. They've built real systems at real scale. And then they slip up on something a new grad could answer, some elementary data structures question or a basic algorithmic problem, and you can see them lose the room. The interviewer, who might be 24 and two years out of school, marks it down. The candidate knows what happened. They've been riding the bicycle for 20 years and they just got asked to draw one.
What separates a good interviewer from a bad one is whether they reserve empathy for precisely this moment. A good interviewer has seen it enough times to understand that the slip isn't signal. They know what it's like to operate at a level of abstraction where the elementary components have been packaged away for years, because they've either been there themselves or they've watched enough people they respect go through it. A bad interviewer takes the miss at face value and moves on to the next candidate.
The situation has gotten stranger in the last few years. First-time founders who are essentially junior engineers, people who haven't held a real job but managed to raise capital and now need to build a team, are sitting across the table from candidates with literally decades of experience. This is a relatively new phenomenon and nobody seems to have fully absorbed how weird it is. A 17-year-old interviewing someone who was writing production code before the interviewer was born. A 23-year-old CEO of a company valued in the millions deciding whether a 45-year-old principal engineer is good enough. Many such cases.
Good founders figure this out fast. They learn to evaluate differently, or they hire someone who knows how, because they burn through enough qualified candidates to realize the bicycle drawing test is destroying their pipeline. The senior engineer and the daily cyclist have the same problem. They know too much about riding to remember how the chain connects to the wheels. That doesn't mean they can't ride.