The 1988 paper
Smalltalk and Exploratory Programming, based on the experience with Smalltalk-80 at Tektronix, among other things, compares some similarities and differences between Smalltalk-80 and Interlisp-D, a couple of which I'm not sure are entirely accurate. On page 4 of the PDF:
A user must be able to find the abstractions relevant to the current task. This has been a problem in Interlisp-D, Unix, and Smalltalk-80, but has not been seriously addressed in any of these environments. Unix does offer some tools for finding things. The manuals include a permuted index and the man command can search by keyword.
Isn't APROPOS equivalent to Unix man's search by keyword? Interlisp-D could also look up keywords in the IRM via HELPSYS and DInfo.
On page 5:
Code in the Smalltalk-80 environment is easier to understand than in the Interlisp-D environment. This is supported by the fact that programmers share code at the source text level in Smalltalk-80, whereas this is rare in Interlisp-D.
Weren't symbolic files and the File Manager widely used on Interlisp?
--