One of the features of Knuth's The Art of Computer Programming is a thorough documentation of the history of ideas. Given that Knuth has been writing for the majority of the period in which we have had digital computers [and nearly all of the time in which we have had high level languages] he has been able to document many things as they have happened. TAoCP has created much of contemporary computing practice and the history of the field.
There is no replacement for primary sources. If you want to understand what makes Lisp the object of love songs, Graham's On Lisp is the book to read. If you want to understand what made Smalltalkers so smug, then grab you a Smalltalk manual and take the time to learn it [and here Knuth's patient approach to "really knowing" is informative...and there's a Norvig short course for just-get-to-the-point Pythonistas].
The pre-internet history of computing lives on the web in PDF's and in boxes at the used book store. Our quilt of knowledge is mostly missing patches and there are a lot of candidate patches sitting unconnected in the box. Even Knuth knows he'll never know it all.
1. Oft cited.
2. Minimally citing.
3. Primarily comprised of original material.
4. Written by an academic.
I don't think that there is only one way of being a primary source. This is pretty common: for example, all the works of Aristotle are believed to be his student's notes, not his original words.
Don't get me wrong, it is worth reading McCarthy's original paper and the Lisp 1.5 manual as well as On Lisp. The bandpass filter should be set to a broad spectrum. Computing is so young that we don't know what will constitute it's canon in 100 years: Today, C. S. Pierce's pragmaticism is philosophical, William James' is considered scientific psychology via historic interpretation. Computing hasn't yet gone through a thorough rewriting as described by Kuhn in Structure of Scientific Revolution. But the interwebs are perhaps the paradigm shift that could cause one.
There is no replacement for primary sources. If you want to understand what makes Lisp the object of love songs, Graham's On Lisp is the book to read. If you want to understand what made Smalltalkers so smug, then grab you a Smalltalk manual and take the time to learn it [and here Knuth's patient approach to "really knowing" is informative...and there's a Norvig short course for just-get-to-the-point Pythonistas].
The pre-internet history of computing lives on the web in PDF's and in boxes at the used book store. Our quilt of knowledge is mostly missing patches and there are a lot of candidate patches sitting unconnected in the box. Even Knuth knows he'll never know it all.
Good luck.