The biggest cost to creating a prototype is coordination among a collection of experts.
- The biggest cost to creating a prototype is coordination among a collection of experts.
- For example, a PM, designer, engineer, etc, all with an individual picture of the overall problem.
- The PM doesn't have enough time to do it themselves, so they have to communicate it with enough clarity to a team of people to execute.
- The coordination cost dominates the cost to create.
- But if you could have it one head, you could get away with significantly less coordination for a given magnitude of output.
- LLMs provide that leverage.
- This implies the power of generalist PMs should go up.
- What are some implications for product organizations?[nv]
- It should be easier to swarm and find good ideas cheaply.
- Organizations should design themselves to be more resilient to experiments failing and reducing the downside cost of them.
- They should also make sure the employees that get the upside are also exposed to the downside.
- This helps align the individual and collective incentives.
- The worst is the person who gets the promotion (upside) is only exposed to indirect / diffuse downside, so it won't hurt them directly, or by the time it does they'll be gone.
- If individuals can be more agentic and productive, then it's more important the swarm of individuals is aligned with the collective's goal.