Back to Blog
WritingsMar 8, 2025Source article

How to catch a thief

Finding one culprit among billions is a filtering problem: apply known facts one by one until the search space narrows to the smallest possible set of suspects.

A murder occurred last night in the Shibuya station. Many many people were there, and some even witnessed the horrific event. His weapon of choice was a Samurai sword, one that was later retrieved nearby in a common dumpster… The man was 5’4, and knew how to wield the sword…

There are 8 billion people on the planet. Who did it?

Death Note artwork featuring Light Yagami and L

Well, firstly, the murderer was in Japan that night. There are around 125 Japanese people, and another 20 million tourists plus immigrants. So that means everybody not in Japan that night is not part of the murder (directly).

The other prefectures we can cross out, and they were obviously in Tokyo.

So who was in Tokyo that night? That leaves around 30 million people.

«funny side note. once we have autonomous ai agents, this will be so much more difficult to track. who programmed it? what was the make? the possibilities could become endless, and therefore far harder to find the thief»

Death Note artwork featuring Light Yagami and Ryuk

The Sword was made in Japan, and the wielder knew how to use it. So now of the 30 million people that night, 15 million are men, and of those men, which subset had a hand made Japanese samurai sword, was between the ages of 20 and 50, and knew how to wield it?

To catch a thief, we use a Filter function, slowly bringing the noose closer and closer until the right profile is found, with as few people as possible. We then take those suspects, and whittle it down ideally to one.

Share this article
Explore what I'm building.