A good manager is able to act as the interface between multiple teams, for example collecting requirements and resolving impedence mismatches. A developer who has advanced along the technical path does need some social and people management skills, but the role is very different. For example, they can mentor junior developers, or they can work with their managers to ensure that technical problems at said interface are spotted before they become problematic.
(Conway's Law is huge for a high-level architect... in a lot of ways you do not so much "create" the architecture as discover how to correctly translate the pre-existing corporate structures into code in the most business-efficient manner. "Architects" in this position who think they're working with a blank slate crash and burn, and probably are the source of the disdain for the term. A real architect is in fact working in some of the most constrained design space there is. In some ways that actually sorta kinda makes the job easier; you can sort through and dismiss a lot of bad choices very quickly with Conway's Law. The biggest challenge is when you get constrained down to zero choices for some problem and have to work Conway's Law in the other direction, and figure out how to fix the corresponding structural failures in the organization itself. It's a lot easier to change code than people structures, but sometimes you have no choice....)
Why ever not? You can study and discuss existing architectures, their advantages and disadvantages, discuss case studies of existing code and new requirements and what architecture to implement them in etc.
Of course, you also need some decent programming experience to come up with architectures that work in practice, but that can be also acquired.
The Conway's Law factors are also harder to treat "fairly" in a theoretical sense relative to some of the real team-related difficulties that exist in any large enough organization.