I’ve already explained why that principal doesn’t hold up (by referencing tthe very same link you just posted) but to add a bit more detail:

That page lists all of these types of decisions as examples of what should be centralized:

• Standard technology platforms
• Alignment around value streams
• Common ways of working
• International expansion
• Standard development languages
• Standard tooling
• Offshoring
• Product strategy

…in addition to any other decisions that might be made infrequently, last a long time, or benefit from “economies of scale”, which could easily be applied with a broad brush.

When you say “centralize all decisions (except the important ones)” it means you don’t really believe in the power of decentralized decision making.

In the segment you’ve highlighted I was speaking about product strategy decisions baked into Epics that fit your definition of “strategic”. Through your comment you’re basically agreeing that SAFe does not want these decisions to ever be made locally — which was exactly the problem I was pointing out.

Additionally, claiming to value “decentralizing decisions” does not hold weight if you also rigidly enshrine centralized decision making into the structure of your framework (through things like the program level roles, enterprise architect role, solution manager role, portfolio management approvals, standardized cross-team processes and practices, etc).