[97 Things Every Software Architect Should Know] is much more broad than most people would expect from its title. It's certainly true to its title, but I expected that it would have be 97 things software architects should know about software architecture. Many of the points, while good advice otherwise, aren't special to software or software architecture. They are points any manager, project leader, or executive could apply. It's really 97 Things Every Project Manager Should Know, although there's already a book for that.
The first thing every software architect should know is what is expected from that job title, and I was hoping someone would at least try to define it. In reality, the title is a dumping ground for the tasks you don't give to the programmers but don't trust to the executives, and the job description varies widely.
My notion that nobody really knows what a software architect should do is reinforced by reading the advice from the many contributing experts, each of which briefly write about what they think is important. Some of that advice conflicts with other contributors, is so general so that the it would suitable in any business book, or merely shows that anyone touching a keyboard might be labelled a "software architect".
I was surprised that a lot of the advice tried to actually force the commoditization of "software architect", as if the actual person doing the job was interchangeable. An architect's experience, vision, and artistry should be at the center of the endevour. Architects are not cogs; they create and enforce the philosophy and design concept. In that regard, I actually only know a handful of software architects. Most people who consider themselves an architect, however, are probably merely applying the design and philosophy that somebody else created.
Setting aside the definition of architect, the advice is good for almost any project leader involved with software development no matter their job title. It's much better advice, however, for the journeyman who wants to be a project leader someday.
As with many management books, anecdotes are rife and facile. They are the sort of things you might mention in an elevator, such as changing the Mach 2.5 requirement of the then-future F-16 fighter plane to "escapes combat quickly", but that anecdote doesn't really help anyone, or at least not in the same way as something holistic and fleshed-out like Skunk Works: A Personal Memoir of My Years of Lockheed. I would have appreciated footnotes or references to complete case-studies.
Given the short format of each contribution, this generality is probably unescapeable. That's mitigated somewhat by the accompanying website for the book where longer discussions might take place. There's no overarching concept or guidance since the contributors are advising in different dimensions, coming from different experiences, and using their own idea of what a software architect should do. Very few contributors talk in terms of the complete software life-cycle; much of the advice in the vein is about requirements research, and even then is mostly about proper mindset rather than useful techniques.