I think that’s what GP means, that if you don’t include the “context” (interpreter, cpu architecture, power supply), you’re only burying the most malleable tip of the iceberg.
It reminds me of the Zork source code that was published here some months ago. It was written in a language for which the compiler appears to have been lost. People have tried to develop compilers given the source code provided, but we can’t get it to work just right, because some crucial bit of context appears to be missing.
There will probably be people who maintain LISP and C for decades more, but imagine trying to write a Java or Haskell compiler with only the source code for Kafka or Pandoc as your guide?
Not impossible of course, but hey, maybe you should check the source code for a JVM or ghc version (and maybe add a Linux distro and gcc for good measure) in with your source in time for the cut off date. :)
Point is, once you enumerate all of the dependencies, you realize the only solution is what GP said. Constant maintenance at all levels is the only thing that keeps the ship running.