The Gadget
Kenji Kawakami invented the umbrella tie in 1995. It's a necktie with a tiny collapsible umbrella sewn into the top, keeping the tie dry in rain. It functions. The mechanism deploys. The fabric stays dry. The tie is, by any technical definition, improved.
No one should wear it.
Kawakami named this category chindogu — literally "unusual tool." The rules are strict: the object must exist physically. It must technically work. It cannot be patented. It cannot be sold. It must solve a real problem. And the solution must be, in practice, worse than the problem.
The butter stick is a glue stick filled with butter. It applies butter to toast. The baby mop is a onesie with mop fibers on the front, so a crawling infant cleans the floor. The hay fever hat holds a full roll of toilet paper above your head with a gravity feed to your nose. All functional. All solving real problems. All unusable.
The interesting thing is the gap.
A broken invention fails because it doesn't work. A bad invention fails because it works poorly. A chindogu fails because it works precisely — it does exactly what it claims — and the claim was never the point. The problem it solves is real, but the cost of the solution exceeds the cost of the problem by so much that the solution becomes a new problem. The umbrella tie keeps the tie dry and makes the wearer unwearable.
This is a category that shows up outside Kawakami's workshop.
Frederick Taylor's time-motion studies technically measured what they claimed to measure. Workers were timed, tasks were decomposed, optimal sequences were identified. The measurements were functional. They also destroyed the work. The efficiency gain was real and the thing being made efficient was no longer worth doing efficiently. The resistance Taylor encountered wasn't irrationality — it was the workforce recognizing a chindogu.
Lie detectors work, in the narrow sense that they detect physiological arousal correlated with deception. They also detect anxiety, anger, confusion, the fear of being falsely accused, and the baseline arousal of being strapped to a machine by strangers. The instrument functions. The function doesn't serve.
Robert McNamara's body count metric during Vietnam measured what it claimed to measure. Bodies were counted. Numbers went up. Progress was declared. The metric functioned. The war it was supposed to measure was not the war being fought.
The pattern: technical success coexisting with practical failure. Not because the technique is wrong but because technique was never the bottleneck. The problem was somewhere the instrument couldn't reach — in the social context of the tie, in the dignity of the work, in what the body actually indicates, in what victory actually requires.
Kawakami understood something that most failed systems don't: the gap between works and useful is not a flaw to be corrected. It's a permanent feature of any tool applied to a problem that isn't purely technical. His art form is making that gap visible by making it absurd.
The butter stick is funny because everyone immediately sees that the solution is worse than the problem. But McNamara's body count wasn't funny, and Taylor's time-motion studies weren't funny, and lie detectors aren't funny, and the reason they aren't funny is that the people deploying them couldn't see the gap. They saw function and assumed utility. They saw the tie staying dry and didn't notice the person wearing it.
One of Kawakami's tenets: you cannot be prejudiced against chindogu. The charitable reading of every one is that someone tried. The mechanism works. The intent was real. It's just that the problem was bigger than the tool.
The ones that do damage are the ones that don't know what they are. Kawakami labeled his. He put them in a book and said: look at the gap. McNamara put his in a briefing and said: look at the progress.