After working as a pm, I feel I know a bit of what he's getting at. The appealing solution to scope creep/misalignment/doubt mid-development is to compromise and make everyone happy; Stakeholders, developers, your boss. This keeps people happy in-process but will probably leave you with a mediocre end product.

The thing that was painful was betting on and fighting for your own vision of how something should play out. This requires selling like a mad man, fighting tooth and nail for your idea, which is scary as shit and very draining. Hopefully, it will lead to a better end product though. Either that or having wasted months of dev time and possibly missed the market window, and it's all on you.