I think that it's probably qed, but I can't find any sources to confirm or deny this.
Also, ed has existed for the entire duration of Unix, and sam was written in the early 1980s. Is it known what most people used at Bell Labs from the Unix era to Plan9?
I'm thinking that the timeline looks like this:
qed -> ed -> sam -> acme
Best Answer
I think your advised time line is correct.
At the very beginning before Unix (called Unics at 60s, standing for UNiplexed Information and Computing Service) , Bell Labs had focused on the Multics Project which failed (Multiplexed Information and Computer Services). Multics was written in QED.
Actually QED had been developed even before Multics; it was first written by Butler Lampson and Peter Deutsch for the Berkeley time-sharing system on the SDS 940; see their paper in C. ACM 10 #12 (December, 1967).
QED had been brought to the project team of Multics by Ken Thompson which he upgraded from character oriented to line oriented (QED CTSS).
Later, when Multics project failed, Ken and Ritchie "forked" Multics to Unics and QED had been continuously further improved reaching finally the later ED and VI.
All these and much more are well explained in this Bell Labs paper about QED History, written by Ritchie himself (!).
Also in this Bell Labs link you can download the QED Manual (dated 1970) as had been enhanced by Ken and Ritchie for the Unics Project (!).
Among other concepts, in this QED manual basic regular expressions are explained in the same format we use them up today.
PS: I admit that I'm excited about finding all this information!