Programming language verbosity

When people discuss "language X vs. language Y", one of the key factors under discussion is "verbosity": "... but Java is just so verbose!" Put another way, Java apparently has a very high level of redundancy.

One measure of redundancy in a signal is its compression ratio. This is easily computed by comparing the size of code when uncompressed to its size when compressed, with, say, gzip.

I tried this out on each language's most popular three projects on GitHub. The results of this are charted below.

Disparate projects in the same language bunch closely together. This implies the language is an important factor in the project's redundancy level. See how the languages stack up:

Some observations:

• That old criticism of Java is totally right. Java is over twice as verbose as the most character-efficient languages.
• Whitespace-sensitive languages do well in this test. This is because curly-brace languages also use whitespace indentation, thus encoding the same information twice.
• I have an untested theory that two things contribute to overall redundancy:
  • Syntactic redundancy (e.g., curly braces)
  • Semantic redundancy (lack of support for abstraction, leading to copy-and-paste)