Isn't this some evidence that LLM swarms are a bad idea relative to more structured approaches?

If YC has that 2 person teams tend to be more nimble than 50 person founding teams, why would agent swarms be different?