I'm currently reading Don Norman's book and I'd like to add that to me it seems to be a very worthwhile read for everyone who designs things that are used by humans. And by that, I explicitly also mean GUIs, APIs, IDEs, programming languages & DSLs...
The book is very focused on what makes good design good, and only provides examples as illustrations. It's really more about applied psychology than design per se (in fact, Norman states in the preface that the original title was "The psychology of everyday things").
Edit: As an example in Software Engineering: His formalization of how humans interact with things (form a plan to reach a goal -> interact with the thing to advance toward said goal -> evaluate the outcome, rinse repeat) and the potential error sources on the way (distractions, unclear consequences, unclear affordances...) make it obvious why in many functional languages, the REPL is such a great tool.
I couldn't agree more re: Norman's book. I wish it had been recommended to me as a linguistics undergrad. He provides such a clear explanation as to how design is a way of solving communication problems over making things look good.
I also run into "Norman doors" at least once a week and always smile when I remember the term.
Get you a man who can do both. Architects in particular seem rather high-falutin' until you read the understanding of their work. Aesthetics often means making something so beautiful that it harmonizes with the rest of how you live. Ugly things stand out and cause greater disunion in our lives.
The book is very focused on what makes good design good, and only provides examples as illustrations. It's really more about applied psychology than design per se (in fact, Norman states in the preface that the original title was "The psychology of everyday things").
Edit: As an example in Software Engineering: His formalization of how humans interact with things (form a plan to reach a goal -> interact with the thing to advance toward said goal -> evaluate the outcome, rinse repeat) and the potential error sources on the way (distractions, unclear consequences, unclear affordances...) make it obvious why in many functional languages, the REPL is such a great tool.