You Can’t Promote a Programming Language By Its Features Alone

It is deceptively easy to make the mistake of attributing a language’s popularity solely to the syntax and functional features of the language. In reviewing why languages became successful, there is often a retroactive feature-driven mythology wherein we over-attribute the success of a language to its unique syntax and functionality. Simultaneously, we under-attribute the environment the language was in at the time of its early maturity to its eventual uptake.

Ultimately, no programming language has become highly adopted on function and syntax alone, and so it does not make sense to retroactively attribute its success solely to function and syntax. Its ascent must be contextualized with the era of its uptake to be well understood.

The adoption of a language toolchain takes significant effort for a team or an org to achieve, and some threshold must be crossed to make the integration last. While language features can lower this threshold, it is the environment in which the language flourishes that justifies the overcoming of the threshold in the first place. This forward momentum — the why — is essential for adoption.

In practice, the most popular languages we use are the result of the intersection of trends in computing with the function of the language. It is informative to look at the environment in which each of these languages flourished.

• C C was the best language for hardware-agnostic Unix programming at the time when Unix was taking off. Low-overhead Unix syscall compatibility and direct memory access were a fit for the everything-is-a-file Unix trend, whereas Pascal was weaker in these crucial areas.

• C++ By the 1990s, programs were increasingly created by larger teams. C++ offered syntactic boundaries that allowed programs to map to team structure. Additionally, the 1990s were a golden age for selling compilers in boxes at retail, enabling unparalleled awareness of the language. GUIs became commonplace, and the OOP additions to C++ were perceived as useful for composing graphical interfaces.

• Perl CGI needed something more practical for extraction and reporting than Bash. JavaScript was effectively unreliable in web browsers of the era, so sophisticated full-page server-side rendering was the order of the day. Late ‘90s cheap shared hosting democratized server access. The availability of MySQL and Postgres clinched the deal, and the LAMP stack was born.

• Python 2 By the 2000s, scripts were increasingly created by larger teams. Python 2 offered syntactic constructs that made it easy to compose long-running scripts in a more readable and maintainable way than Perl.

• Python 3 PyData/ML stack became institutional. As data science and ML hiring surged in the 2010s, Python 3 became the default research-to-production path. Many new developers may even think of NumPy/SciPy as being intrinsic to the toolchain.

• Ruby (on Rails) Developers needed to iterate quickly in the problem space of web service business. This happened at a time when VCs infused cash into a cultural movement that emphasized iterating quickly to find product-market fit.

• JavaScript The web browser is the open, universal runtime we got, and JavaScript had a twenty-year head start on any real alternative. (Sorry, Java Applets).

• Java The onboarding of businesses to the Internet happened at a time when a memory-safe language was needed for network-aware programming by large numbers of programmers. The influx of many freshly graduated compsci majors who found direct memory manipulation challenging could be put to work to deliver real corporate value on large teams. Java’s timing was serendipitous for the enterprise build-out.

• C# Microsoft, at a time when they were the most dominant software vendor in the world, massively backed developers during “dot com” layoffs and early 2000s corporate downsizing. They emphatically showed the Windows platform-supported way of doing memory-safe programming. Using COM or Java to build Windows was now off-book. Out-of-work developers took them seriously and reskilled with C#. Unity later bet the farm on C# which was a second shot in the arm for the language.

• Rust Modern networked devices have to deal with increasingly diverse untrusted data processing (such as bluetooth packets, fonts, audio streaming). Major industry players such as Microsoft, Google and AWS have sought to reduce memory-safety-related CVEs in “zero-click” attack surfaces without the use of garbage collection.

A lesson we can take from this is that you can’t market a language a position of popularity by simply espousing its features and qualities in public. You can, however, build a language and the foundations of its toolchain and help it hitch onto a greater trend. Changing needs are what prompt teams to overcome the friction involved in upgrading their software development tooling.

As always, timing is key.