For article #1000, it seems appropriate to end on a site where the web still behaves like a filing cabinet full of actual work. Gordon S. Novak Jr.’s University of Texas at Austin homepage preserves decades of artificial-intelligence research, Lisp code, course notes, papers and experiments without trying to turn any of it into a personal brand.
Visit Gordon Novak’s UT Austin site.
Novak’s research career reaches back into an AI world that looks very different from the current one. His interests include automatic programming, physics problem solving, compilers, software reuse and Lisp. He directed UT Austin’s Artificial Intelligence Laboratory from 1984 to 1999 and developed systems aimed at getting computers to reason about programs and physical problems rather than merely execute hand-written procedures.
One of the recurring ideas is wonderfully ambitious: describe a problem at a higher level and make the machine generate the program. Novak’s work on GLISP explored a Lisp-based language with data abstraction. Other systems worked on specializing generic procedures for particular data structures, generating scientific programs from diagrams and solving physics problems stated in ordinary English.
That last goal is especially fun to encounter in an old academic web directory now that natural-language AI is fashionable again. Decades before today’s chat interfaces, researchers were already asking how a computer could interpret a written physics problem, extract the relevant quantities and relationships, and produce a solution.
The site is not merely a publication list. It keeps teaching material alive. Course pages for algorithms, data structures, symbolic programming and artificial intelligence include syllabi, notes, examples and old software references. One algorithms course casually moves from Big O and trees into Lisp, rewrite rules and MapReduce. Another set of AI notes is still directly readable as plain academic HTML.
Novak also released software through the old mechanisms that made university computing pages so useful: downloadable packages, FTP archives and web demonstrations. His CV records tools for unit conversion, Scheme tutoring, Common Lisp/X Windows interfaces, expert systems and teaching software distributed to other faculty.
There is no clean boundary here between research archive, classroom, software repository and personal homepage. That is precisely why it belongs in this project.
The Ancient Web was never only dancing GIFs and abandoned fan pages. It was also professors putting decades of knowledge directly on public servers because that was the obvious thing to do.
Explore Gordon Novak’s archive. A thousand articles later, this is still the pattern worth saving: one person knows something difficult, puts the evidence online, links the pieces together, and leaves the door unlocked.