Too sweet to eat

In 1996 French chemists made a sweetener 250,000 times stronger than sugar. Thirty years later it has sweetened exactly nothing.

Lugduname is perfect. A teaspoon of the stuff is sweeter than a METRIC TON of sugar. It’s impossible to use because we can’t figure out how to filter out micrograms of it. So it sits in a freezer in Lyon while the rest of us drink the cheap stuff.

Every sweetener that did reach a shelf solved this at the plant. A Splenda packet is about 1% sucralose and 99% filler, so nobody in a diner ever touches the concentrate. Dosing is the factory’s job.

And that’s the secret to GOOD workslop.

Which brings us to curl, the open-source tool moving data inside cars, phones and medical devices. Billions of installs. A handful of volunteers. They’ve run a bug bounty since 2019 to pay people who find flaws in their code.

Then the AI flood came: Bounty filings with real findings fell from about 15% of submissions to under 5%. Only 1 in 20 had anything useful to say. Daniel Stenberg shut the bounty down on January 31.

But the models are getting better, so this one fixes itself, right?

He reopened it in March and it got worse.

The reports are real now. Reproducible. But they’re also identical. Everyone prompts the same model, everyone gets the same bug, everyone mails it to a stranger. Finding the flaw is free. Ranking it and writing the fix takes a tired volunteer at a kitchen table.

Stenberg's fix wasn't a ban. He killed the payout that made submitting free and demanded a working proof of the bug before anyone reads a word. No one at curl touches concentrated slop. Dosing became the sender's job. 

Barbara Minto did the same thing at McKinsey in the 1960s. Back when partners were drowning in analyst decks, she made a new rule, everyone had to write “bottom line up front.” It was more than a writing tip, it was a workslop dosage limit.

Drowning in GOOD slop?

Make it the slop factory’s problem.