I used to think that running a software business required a small village.

You needed people to manage products engineers to build things writers to make documents and sales people to talk to customers. If you were building alone without a lot of money, you simply had to accept that you would move very slowly.

Then things changed with the tools we use.

Today as a person who runs a company by myself I can see what my competitors are doing, pull customer complaints from Reddit and Twitter, understand what people want, and have an updated version running before the day is over. I do not need a team. I have computer programs that help me monitor feedback write code and update documentation.

On paper it looks like the ultimate dream. The whole company running by itself.

We tell ourselves that by connecting customer feedback directly to automated code, we have solved the business. But when you actually look at the math, you realize that moving fast on autopilot is a trap.

Because feedback loops on autopilot do not make you unique. They make you average.

When every person who runs a company uses the same automated tools. Watching what competitors do, ingesting the same online complaints, and building the top five requested features. Every single product begins to look identical. Startup A copies Startup B’s sidebar. Startup B copies Startup A’s pricing model.

When code costs zero dollars and everyone listens to the exact same aggregate data, software becomes a utility. Like electricity or water. Profit margins collapse to nothing because nobody is building anything distinct.

The machine can only optimize for the consensus.

An algorithm looks at customer feedback and says, Eighty percent of users want more buttons more integrations and more settings. An algorithm will never make an opinionated product. It will never build something minimalist, and it will never say no to customer data.

The only thing an algorithm cannot do is make an irrational bet.

The machine is brilliant at keeping up with the weather. It helps you stay alive and keeps your costs near zero. But true differentiation is never found in aggregate data. It comes from a human founder having the courage to ignore what the data says and build something the algorithms think is a bad idea.

An algorithm optimizes for the average. True differentiation is always an irrational human bet.