— Paul Graham (@paulg) October 12, 2019 Bel is interesting, not least because it highlights the enduring and improbable longevity of one of the earliest high-level programming languages. That’s no small feat, especially given the hype-driven nature of the software development sphere, where front-end JavaScript frameworks emerge and disappear on a near-weekly basis. That said, before we delve into the technical weeds, it’s probably worth doing a bit of a recap on LISP.
The origins of LISP
LISP started life in 1958, when mathematician and computer scientist John McCarthy published a paper called “Recursive Functions of Symbolic Expressions and Their Computation by Machine, Part I.” His goal was to create a system of mathematical notation for algorithms, rather than a fleshed-out programming language. History, however, had other ideas. McCarthy’s ideas caught traction, and a few years later, researcher Steve Russell wrote the first LISP interpreter on an IBM 704 computer.
For the next thirty or so years, LISP would be an area of focus for computer scientists. The language fragmented into multiple distinct and interesting dialects which pioneered concepts still used today, like running code through REPL (read, eval, print loop) environments. Much of the early work in AI was also written in LISP. Beyond the academic sphere, LISP found favor with a vocal niche of developers, most notably Paul Graham and Eric Raymond. Although it lacks the mainstream success of Python and Java, there are those who insist it’s the greatest thing since sliced bread. And LISP’s advocates make a compelling case. LISP inherently lends itself to the creation of code that’s readable and expressive, as explained by Peter Serbel: And in the same way that learning Latin makes you better at languages, many argue that learning LISP makes you a better programmer. Here’s what Eric Raymond wrote in his essay titled “How to become a hacker:”
Meet Bel
So, let’s go back to Bel. Some readers will already know that Graham already created his own LISP dialect, called Arc. Bel is separate to that. Perhaps the most intriguing facet of Bel is that its development is intended to mirror that of the original LISP spec as much as possible, even if the language diverts from it in some parts. Just like LISP started life as a system of notation and later evolved into a fully-fledged programming language, Graham wants to keep Bel in its incubation period for as long as possible. Here’s what he wrote in the guide to the language: By doing this, Graham hopes to answer the question: “If computers were as powerful as we wanted, what would languages look like?” Bel is an interesting experiment, and I’m interested to see where it leads. Although Graham has published a guide to the language, along with some source code, I’d recommend you hold fire before you add it to your LinkedIn profile. “This is not a language you can use to program computers, just as the Lisp in the 1960 paper wasn’t,” Graham wrote.
— Paul Graham (@paulg) October 13, 2019 Although he has no plans to write a formal implementation that would allow developers to write Bel programs (at least, not just yet), Graham has no qualms with the community making one. “If you’d like to try writing an implementation based on Bel, please do. I’ll be one of your first users,” he said.
