WorkingNotes on org design

The Second Product

A successful startup soon has two products: the tech and the organization. You build both the same way.

Software ends up looking like the organization that builds it. So if you engineer the software, you have to engineer the organization too. Engineering is the right word: split the problem into pieces you can solve, solve each one, put them back together. It works for an org the same way it works for a distributed system. One thing follows: you want people who shape the organization, not people who get shaped by it. It’s the same standard you already hold your tech stack to.

Orgs Don’t Solve Problems

Organizations don’t find problems or solve them; people do. The org’s job is the smallest platform everyone needs to operate on.

People find new problems and solve known ones. Organizations implement known solutions and keep them running. That’s the honest division of labor, and it makes the org’s job clear: build the smallest platform everyone needs, then get out of the way.

The main design tension is authority. Judgment makes the difference, and using judgment takes authority: making some people angry and surviving it. Flat orgs spread authority out, so more gets done without waiting on the top. The harder the problem, the less context the center has, and the worse its decisions get. But orgs also slow down from friction, and removing friction sometimes takes very powerful authority.

Titles Are a Line of Credit on Authority

Ownership needs authority, and authority is slow to earn. A title is a line of credit on it: authority borrowed from the future. Like any credit, it’s a problem if it can’t be paid off.

Palantir had no titles, but it was always clear who got listened to and who everyone waited on for a decision. That’s authority. It has to exist; there is no ownership without it. Authority removes friction in a group. But like capital, it’s slow to earn. A founder holds close to absolute authority. How do they hand a piece of it to someone they trust?

So organizations invented credit: the title. It borrows authority from the future, the way credit borrows capital not yet earned. If the authority shows up, the debt is paid and the title was simply early. If it never shows up, the org inherits bad debt.

It’s popular now to say “let’s just not have titles.” That’s my grandmother telling the whole family to never take credit, because she survived the 90s in Yugoslavia. Understandable, and wrong as policy. Titles exist for the same reason companies raise capital: to get somewhere faster.

So before creating a title, ask: what authority is this bringing forward, and can it be paid back?

There is a competing philosophy: titles cost the company nothing and matter a lot to people, so hand them out to get the people you want. But there is a covert contract inside, and covert contracts are how things blow up later. Use carefully.

Don’t Swim Upstream

Orgs are organisms with a will of their own. You can bend one to yours, but your pull works like gravity: the further from the center, the weaker it gets. So don’t swim all the time. Swim when the wave is right, and save energy when it’s wrong.

These notes treat the org as an engineered artifact. Here is the counterweight: past a certain size, however well you designed it, the org develops a will of its own. You can bend it to yours, but your influence fades with distance, like gravity. So the right picture is the sea: pulled by several gravities at once.

The way out of the sea is not to swim the whole time. You can, if you’re strong, but the stronger the waves, the more energy you burn, and you’ll want that energy later. What if you have to run once you’re out? Swim when the wave is right; rest when it’s wrong. Timing over force. It loops back to authority: the stronger your own gravity, the more waves you can create or ignore.