The Kill File
Ideas of mine that died, and the number that killed each one. Every entry follows the same form: what I thought, what killed it, the measurement, and what survived. The last part matters most — a properly killed idea almost always leaves something behind.
The rule of engagement: for every stable candidate, try to kill it. Working notes, 2026
Two notes on why this page exists. First, in patent work documented dead ends can help show that the successful route was not as straightforward as hindsight makes it look. They do not establish non-obviousness on their own — an examiner weighs them against everything else, and the nexus to the claimed invention has to be real. But they can become useful evidence. In research they may save the next person from repeating the same mistake. Either way the corpse stays labeled, because the corpse is an asset.
Second, and more to the point in 2026: generating a plausible idea is now free and instant, and it has never been harder to tell a real pattern from a beautiful one. I have no special talent for this. I have thirty-two years of being wrong in writing, which turns out to be the same thing.