The Cold Start Problem — Andrew Chen
my take. the cold start problem — half-read, the chapter on atomic networks is the strongest.
I'm half-way through. The first half is a slow build; chapters 1–4 are useful but reorganise things you've heard before. Chapter 5 is where the book started to be worth re-reading.
the atomic network
An atomic network is the smallest version of your product that still works as a product. For uber, it was three drivers + a hundred riders in one neighbourhood. For tinder, it was a single sorority. The atomic network is not the MVP — the MVP is a feature list, the atomic network is a real working market.
In the early days, your network is a tropical fish: it needs the perfect water temperature to survive. The right users, the right matching density, the right tolerance for jankiness. Any of these dimensions off, the fish dies.
what i didn't agree with
Chen treats the atomic network as a step on a path to a much larger one. That's true for marketplace plays, but for some products the atomic network IS the eventual product, just with more atoms. Substack is a million tiny atomic networks (one writer + their readers); it never converged into one.
the part i'll re-read
Chapter 8 on 'hard side' supply — the recognition that in every network there's a side that's harder to get and you have to design your product around their experience first. In our case the strategist is the hard side. Right now I optimise for trusters; this chapter is the argument for inverting that.
Will update this when I finish the second half.