Only tangentially related but I loved reading this interview with him from Computer History Museum https://archive.computerhistory.org/resources/access/text/20...
It’s not the same impact as the video interview in Designing for Interactions but covers a lot of ground from Larry Tesler’s perspective.
Eric Schmidt was, as I recall (which was when I was there), at PARC proper, doing his PhD thesis research and writing his thesis. But as Larry Tesler's interaction showed, there were fluid interactions between at least some people at PARC and the XBS and Xerox Star teams.
Which side of the debate was he on personally?
He was very much in favor of modeless design, which emacs is much closer to than vi is. My argument was that emacs still has modes, operationally. You can make it a debugger, a mail reader, or a text editor by invoking code that puts it in that "mode." So the discussion quickly becomes what is meant by 'mode' and how are operational modes different than semantic modes which are different than presentation modes. If you haven't guessed it was pretty memorable for me, it helped me see some insights into the difference between design and engineering.
I'm not sure that debate was really important since it's clear that his real answer was 'none of the above'.
I first met him while I was visiting my wife at her office in Xerox Business Systems (XBS). He came over to discuss some suggestions to improve the protocol she was working on. I thought he was one of her co-workers because the discussion was very peer to peer as opposed to top down. She corrected me to point out he was one of the movers and shakers at PARC. That left a very positive impression on me.
He was also "the other Larry" at Xerox. Larry Garlick, who was also "Larry" to most people, was also at XBS (as was Eric Schmidt) and later followed Eric over to Sun.