---
title: "Focus"
description: "The next hour on something already working may be worth more than the first hour on something new."
canonical: "https://www.rickblalock.dev/scratch-notes/focus/"
date: "2026-09-07"
updated: "2026-09-07"
author: "Rick Blalock"
---

# Focus

> The next hour on something already working may be worth more than the first hour on something new.

Published: 2026-09-07

Catalog: SN-010

[Joe Lonsdale wrote in 2010](https://www.8vc.com/resources/lessons-from-peter-thiel):

> Don’t divide your attention: focusing on one thing yields increasing returns for each unit of effort.

We tend to price a new idea by its visible cost. A few hours each week. One person for a month. A side project that will not interfere with the main thing.

I suppose that assumes every hour has equal value.

Existing work carries accumulated context. The team knows what it has tried, where customers hesitate, which details matter, and what looked promising but failed. Feedback arrives faster because people know what to look for. The next hour can be worth more than the first simply because of that accumulation.

A new project begins lower on the curve. It needs context, vocabulary, trust, and its own feedback loop.

[After ChatGPT launched](https://x.com/davidsenra/status/2091538100623151216), it was growing quickly, but the growth felt unstable and perhaps low-value. People inside OpenAI were discussing five or six other things they could pursue instead.

Peter Thiel told Sam Altman:

> “It’s an obvious mistake to do anything but this.”

Thiel compared ChatGPT’s open text box to Google’s. People could type almost anything into it and get something useful back. It was growing, unusually flexible, and did not fit the prevailing Silicon Valley wisdom.

He was looking at the evidence already in front of them. Something rare was working before OpenAI had a clean explanation for why. Dividing attention among five or six other directions would have ignored the clearest signal they had.

Altman said they went “super hard on ChatGPT.”

“This could be good too” is a low bar. When something is working, the new thing has to compete with the next return on the current work, not its first return.

Before adding another project, ask what its first hour will replace.

---

[All Scratch Notes](https://www.rickblalock.dev/scratch-notes/)