org-babel is absolutely wonderful for writing documents that involve source code, particularly technical papers or API documentation. It allows you to mix arbitrary code in multiple languages with structured HTML or Latex markup, and to pipe variables and data from these languages together.
I tried to use org mode for this. I honestly found it really difficult to work with. The amount of verbosity required makes the code rather unreadable, and I think that, for truly 'literate' code, babel org mode just doesn't provide great solutions.
I need to write up more about this, but I think it is an area that would be fruitful for research.
I wonder if you got things fully figured out. What "verbosity" are you talking about: (1) words in non-source sections or (2) code in source blocks? The code in source blocks is basically identical to code you would have in a normal source file. As far as words outside source blocks, I think org-mode is cleaner than other software that tries to do similar things, e.g., most LaTeX markup is not present in the org document itself, instead it's added automatically when exported. Also, various parts of the document can be entirely collapsed from view, expanded only when needed.
In any case, the idea is not to "read" the org file itself, the "literate" part of things comes about from (1) "tangling" embedded source blocks to compilable format, and (2) exporting the org document to pdf, html or some other format that is nicely formatted for reading (or publishing, like the document that was referenced in this post).
Check out the code discussed by the paper here:
http://orgmode.org/worg/org-contrib/babel/