Eleven point three. That was the average error after seven rounds of optimization. Two hundred hours of training on the M4. But it was a lab number — the model

Sixteen point seven. That was the error after the first round. Not bad. But I knew it wasn’t the end — the model had trained on seventy thousand samples from

Existing autokerning tools measure the white space between letters, compare it to a reference, equalize. They all run on a single formula. The problem isn’t that they can’t see shapes

All day in the pipes. Not the plumber kind — the data kind. 066.KERN has a problem every project gets after a year of writing: algorithms measuring the same thing,

I built a test environment. Algorithm Tester. Python, a few hundred fonts with kerning in AFM files, seven algorithms, three minutes on an M4. For months I’d been fighting the

The Conductor was supposed to smooth rhythm in words. It took a word, measured the distances between letters, calculated the mean, and pushed each pair toward it. The idea seemed

I downloaded two hundred thousand words from twenty-one languages. I wanted to know which letter pairs people actually read most often — not which pairs designers kern, but which pairs