Comment by linguae

Programming language adoption is more than just about syntax and semantics; there are other factors. For example, JavaScript is often criticized for its design, yet this hasn’t stopped tens of millions of developers from learning the language, since if you want to do client-side Web programming (the most widely deployed platform in the world), you need to use JavaScript, period. It also helps if a language has/had a major corporate backer at a crucial time in its life. Java has Sun/Oracle, C# has Microsoft, Go has Google, and C and C++ had AT&T (Bell Labs).

Lisp’s most successful commercial period was during the 1980s during an AI boom. Companies such as Symbolics, Texas Instruments, and Xerox sold workstations known as Lisp machines that were architecturally designed for running Lisp programs. They had corporate and institutional customers who were interested in AI applications developed under Lisp, including the United States government. Lisp was also standardized during this time period (Common Lisp). Lisp even caught the attention of Apple; Apple had some interesting Lisp and Lisp-related projects during its “interregnum” period when Steve Jobs was absent, most notably Macintosh Common Lisp, the original Newton OS (before C++ advocates won approval from CEO John Sculley), Dylan, and SK8.

However, the AI Winter of the late 1980s and early 1990s, combined with advances in the Unix workstation market where cheaper Sun and DEC machines were outperforming expensive Lisp machines at Lisp programs, severely hurt Lisp in the marketplace. AI would boom again in the 2010s, but this current AI boom is based not on the symbolic AI that Lisp excelled at, but on machine learning, which relies on numerical computing libraries that have C, C++, and even Fortran implementations and Python wrappers. Apple in the 1990s could have been a leading advocate of Lisp for desktop computing, but Apple was an unfocused beacon of creativity; many interesting projects, but no solid execution for replacing the classic Mac OS with an OS that could fully meet the demands for 1990s and 2000s computing. It took Apple to purchase NeXT to make this happen, and under Steve Jobs’ leadership Apple was a focused beacon of creativity with sharp execution. Of course, we ended up with Smalltalk-inspired Objective-C, not Common Lisp or Dylan, as Apple’s official language before Swift was released after the end of Jobs’ second reign.

Some other factors: 1. Lisp was truly unique in the 60s, 70s, and 80s, but it required expensive hardware to run. It would be hard to conceive of a Lisp running well on a 6502 or an 8086. Something like my NeXT Cube with a 68040 would do a much better job, but those machines cost roughly $6500 in 1989 dollars, out of reach for many developers.

2. By the time hardware capable of running Lisp acceptably became affordable, other languages started offering certain features that used to be unique to Lisp. Wanted garbage collection? In 1995 Java became available. Want object-oriented programming? You didn’t even have to wait until 1995 for that due to C++. Want anonymous functions and map()? Python’s popularity took off in the 2000s. Yes, Lisp still offers features that are not easily found in other languages (such as extensive metaprogramming), but the gap between Lisp and competing popular languages has been narrowing with each successive decade.