Coda Hale: The Other Side of Software Engineering

I'm going to disagree that any sort of centralized decision making and vision is a good thing. I noticed this talk doesn't discuss org size. It's assumed in the tech world that growth is a good thing. It's better to limit orgs to one team in size rather than to have your team's vision dictated to you by someone who thinks they are the inner circle of the company. Teams should grow up, not out. A startup should only grow larger if it has the integrity to have fully fire-walled teams without the power cabal trying to micromanage and dictate vision to every project in the company.

Management types developed the term "siloing" and cast it as a bad thing, which is natural given that a manager's goals are to increase their power and influence over larger and larger groups of people. Silos, small visionary teams, and independent decision making, like regime change or independence, are one of those things that make sense for the innovator stuck in an authoritarian organization but that big org functionaries like architects don't want to hear about. If your software startup is large enough to have an architect, it is too large in my opinion.

As for segregating decision making and "implementation", this is the biggest mistake that can be made. Implementation, or execution, should be done in the context of full decision making authority. Otherwise, the organization is going to get beat by a <5 person team that is making major decisions on the battlefield and not in some big shot architecture meeting to be handed down.