1. 20

A while back I watched a talk about “meta” on Youtube but I cannot find it now (searching “meta” on Google just gives a lot of useless results about “Meta” the company). That talk is from early 2000s, but I believe it’s very famous, so chances are that you guys have watched it as well. Most notably, it’s in “simple English”, i.e. every single word in that talk consists of ~3 syllables or fewer. Also, the presenter defined “meta” as something like “treating verbs as nouns”.

That’s everything I can remember now. Do you know the name of that talk?

  1. 18

    “Growing a Language” by Guy L. Steele Jr.; here is a PDF.

    1. 7

      Yup, I actually watched this talk a long time ago before I knew anything about programming languages. I didn’t really get it, and somehow missed that he is specifically advocating adding three things to Java:

      1. Operator overloading
      2. Value Types
      3. “Generics” / Parameterized types

      As far as I know only the 3rd happened in Java (more than 10 years ago I think).

      What I find interesting looking back 20 years is that Python actually did what he was advocating for! At least with operator overloading. Because NumPy was developed around that time on top of those hooks in the Python language, and a huge numeric ecosystem was built on NumPy (including Pandas!), and then a huge machine learning ecosystem.

      That is, I’m pretty sure PyTorch and TensorFlow are now >90% of machine learning code. (Although to be fair this is narrowly defined, because I think there’s a big overemphasis on deep learning today)

      Ask HN: How did Python become the lingua franca of ML/AI?

      My response: https://news.ycombinator.com/item?id=29171519, which links to a good recent interview of NumPy creator Travis Oliphant: https://www.youtube.com/watch?v=gFEE3w7F0ww

      Recall that Guy Steele worked on the Fortress language, which was probably a purer version of Python/Julia. But Python is the one that “evolved” and “grew” to be popular and widely used, sort of as he predicted in the talk.

      1. 5

        A recording on YouTube: https://www.youtube.com/watch?v=_ahvzDzKdB0

        There seem to be occasional audio glitches, causing a second or so to be cut out; it might be a good idea to have the transcript handy.

        1. 15

          This is my favorite programming talk of all time. I’ve had to suffer through the glitches in this recording many times, but luckily a year and a half ago, the Computer History Museum put up a higher quality recording with no skips: https://www.youtube.com/watch?v=lw6TaiXzHAE

        2. 3

          Nice! Reading this talk feels like reading the source code of a FORTH interpreter, as exemplified by JonesForth.

          1. 1

            This reads like an acid trip. I wish I could have attended the lecture.

            Edit: Found a link!