Courses on one topic each, and papers for the library, all in the same animated format. The most voted get made first.
Sign in to vote and suggest. It’s free, and you can change your votes anytime.
Sign in to voteJim Gray, 1985
Gray reads through Tandem's outage reports, finds that operators and software cause far more failures than hardware, and shows how process pairs and transactions keep systems up.
Patrick Lewis et al., 2020
Lewis and colleagues pair a language model with a search index over Wikipedia so it pulls in passages before answering, the design that gave RAG its name.
Peter Naur, 1985
Naur argues that a program lives in its programmers' heads as a theory, and that the code and docs alone cannot hand it to a new team.
Richard P. Gabriel, 1991
Gabriel explains why Unix and C, simple and a little wrong, spread faster than the more correct designs coming out of the Lisp world.
Edsger W. Dijkstra, 1968
Dijkstra's one-page letter argues that unrestricted jumps make it hard to tell where a running program is, which pushed languages toward structured loops and branches.
C. A. R. Hoare, 1978
Hoare builds programs from processes that share no memory and talk only through synchronous messages, the model behind Go's channels.
Guy L. Steele Jr., 1998
Steele gives a talk using only one-syllable words until he defines longer ones, to show why a language has to let its users grow it.
Jared Kaplan et al., 2020
Kaplan and colleagues show that language model loss falls as a smooth power law in parameters, data and compute, and use that to decide how to spend a training budget.