I have always considered his work on Lisp machines back in the early eighties heroic. When Symbolics stopped sharing their developments, he endeavoured to match them feature for feature, single-handedly, and give away the reimplementation for free. To quote Steven Levy, "RMS had single-handedly attempted to match the work of over a dozen world-class hackers [1], and managed to keep doing it during most of 1982 and almost all of 1983". He saw what Symbolics created, and then reimplemented it and gave it away.
Greenblatt noted that he was out-hacking the whole bunch of them. Gosper called it incredible.
When Stallman finally couldn't keep up, he set a new goal that he hoped would solve the problem permanently. GNU.
There is, of course, the other side of the story, although I find the dismissal of the complexity of the features RMS was matching a little disingenuous - how complex they were seems irrelevant compared to the fact that he was doing it alone and Symbolics was doing it with a bunch of world-class hackers: https://web.archive.org/web/20080112153822/http://dlweinreb....
[1] Some people on HN define "hacker" as "person who can code". The definition in use here is older.
The "dismissal", as you call it, is hardly unfair (note, I was an eyewitness to all this, the players were largely in my social circle and I worked for Lisp Machines Inc during that period, heck, we both gave up on Lisp Machines in general at the same time). As DLW relates, RMS didn't do the designs of the new features, he copied some of them. In part I believe this is because he's much more an improver than an innovator when it comes to code and systems. This can have massive qualitative results, like after he took over ITS EMACS, but it deprives him of the cachet of "inventor", for what that's worth.
Greenblatt noted that he was out-hacking the whole bunch of them. Gosper called it incredible.
When Stallman finally couldn't keep up, he set a new goal that he hoped would solve the problem permanently. GNU.
There is, of course, the other side of the story, although I find the dismissal of the complexity of the features RMS was matching a little disingenuous - how complex they were seems irrelevant compared to the fact that he was doing it alone and Symbolics was doing it with a bunch of world-class hackers: https://web.archive.org/web/20080112153822/http://dlweinreb....
[1] Some people on HN define "hacker" as "person who can code". The definition in use here is older.