[mattzito]

> A good engineer can PM his own work, and doesn't need someone else to tell him what color to paint the bike shed

A good product manager, having done their job, has provided the good engineer with enough information about goals, features, requirements, user profiles and so on that the good engineer can run with that information and build something kickass. In many cases, the good engineer has enough information that they can make informed decisions about things that were not covered by the PM, such as what color to paint the bike shed.

[jjoonathan]

Sure, but the process can become wildly unbalanced. If the engineering is enough of a limiting factor that 90% of the planned ideas never happen and the 10% remainder consists entirely of mission-critical features, then a PM doesn't have much value proposition and it makes sense to combine the engineer and PM roles (supplemented with a culture of "why are you doing that" at standups to prevent tunnel-vision).

The opposite can happen too, of course, when the engineers focus on features that nobody actually wants. I think grandparent's point was that PMs aren't always a good cure for this problem.

[guyzero]

"If the engineering is enough of a limiting factor that 90% of the planned ideas never happen"

in other words...

If the team is wildly dysfunctional, then bad things are going to happen.