Key Takeaways

  • Nick Kokonas found chef Grant Achatz at Alinea at 6:00 a.m. testing 20 variations of a single sauce labeled on blue tape to isolate the winning preparation.
  • Shaan Puri cited Chris Rock to show that public performance is merely the byproduct of morning writing sessions where most jokes fail.
  • Kokonas generated over $100M trading options by exploiting Chicago-New York latency before bringing that same empirical volume to Alinea and Tock.
  • First-time founders waste time on vanity assets like logos, while top performers grind through iterations and reject 99% of their output.
  • Teams can implement this systematic approach by applying The 25-Sauces Rule of Iterative Quality.

The 25-Sauces Rule of Iterative Quality

When Kokonas walked into Alinea before dawn, he found Achatz testing 20 variations of a single sauce, adjusting salt and acidity across small pots labeled with blue tape. Kokonas explained the standard: “If you want to be a great chef, you show up at 6:00 a.m. and you make 25 sauces.” Amateurs obsess over outward appearances instead: “Everyone wants to make the logo, right? You know, logo first, name for I got a great name for a restaurant. Awesome. That's terrific. Make the 50 sauces, you know?”

Puri pointed out that comedy runs on the same principle. Comedians like Chris Rock spend mornings writing bad jokes to isolate one good line. “The advice is you write jokes every morning. The money is made in the day. We just happen to collect it at night,” Puri said. “The stuff you're doing on the stage is the byproduct of the daily writing of bad jokes. If you're not doing the daily writing of bad jokes, like that's the whole racket.”

Kokonas applied this approach before selling Tock for hundreds of millions of dollars. He made over $100M trading options by exploiting latency between Chicago and New York. He treated decisions as empirical distributions: “You don't simply go like, well, here's the recipe and it should taste like this. You have to know what great is. And knowing what great is is making mostly 99% of the time.”

Here is the exact method:

  • Establish the Baseline: Create a foundational recipe, draft, or concept in the middle of the spectrum.
  • Produce 20 to 25 Variations Across the Spectrum: Deliberately skew the parameters in every direction (sweeter, saltier, more acidic, longer, shorter) and label each variation.
  • Embrace 99% Failure to Isolate the Exceptional: Acknowledge that greatness is discovered through high volume; reject 98-99% of the attempts until you locate the single variation that stands out.

When This Works (and When It Doesn't)

This method applies to creative and entrepreneurial disciplines where the standard recipe cannot be mathematically pre-determined and requires exhaustive empirical iteration. It succeeds when iteration cycles are cheap, feedback is rapid, and the creator has the taste to judge output. Puri cited the classic pottery class experiment: students graded on the total weight of pots produced better work than students trying to sculpt one flawless vase.

The approach breaks down when iteration costs are prohibitive or feedback takes years. You cannot deploy 25 experimental aircraft engines or medical devices to see which one functions. It also fails if the builder lacks discernment. Making 25 variations of a pitch accomplishes nothing without the taste to recognize what lands.

What to Do With This

Apply this rule to your cold outbound messaging this week. Most founders write one email, tweak two words, and send it to 500 prospects. Instead, run the sequence tomorrow morning:

1. Establish the Baseline: Draft a three-sentence email with your core value proposition, a customer metric, and a short call to action.

2. Produce 20 to 25 Variations Across the Spectrum: Write 25 distinct hooks in one sitting. Skew the dials deliberately. Make one variation 15 words long. Lead another with a counterintuitive data point. Write one with self-deprecating humor, and another asking for a referral. Label each draft from 1 to 25.

3. Embrace 99% Failure to Isolate the Exceptional: Review all 25 drafts and delete 23. Test the top two variants across batches of 50 prospects, and let real reply data choose the winner.