Paper: https://www.cs.virginia.edu/~evans/cs655/readings/steele.pdf
If you find the wording weird, wait until (or skip to) 8:44 for the explanation.
TL;DR generated with claude.ai:
The main theme is that programming language design has changed. Previously languages were designed fully upfront by a small team. Now languages need to be designed for growth, starting small but with patterns enabling users to extend the language. This allows rapid evolution driven by user needs.
The speaker advocates designing programming languages that empower users to define new types, operators, etc. He argues this is better than trying to design a huge language upfront, which will fail, or using a small language long-term, which is too limiting.
Overall the message is that language designers should create an initial framework to allow organic collaborative growth. The speaker relates this philosophy to the “bazaar” model of open source development.
From the video description:
Guy Steele’s keynote at the 1998 ACM OOPSLA conference on “Growing a Language” discusses the importance of and issues associated with designing a programming language that can be grown by its users.
The logical conclusion of this is to let the syntax and parser of the language itself be modified by users in a modular fashion :D
So python then. On the first order level, there’s operator overriding and duck typing, which achieves this highly modular syntax. But deeper down, there’s the ast module that lets you rewrite your own interpreter at runtime and other wacky shit.
And the logical conclusion of this is there is a non-zero chance you get to enjoy the opportunity, no, the priviledge to maintain that 100k loc codebase written in someones personal PHP dialect. Lucky you.
Which (if I understand correctly) is almost what Forth does. I’ve heard it described as “the only syntax in Forth are the blank characters” :).