It's good to get excited about something dumb now and then.

I won't deny that wc(1) is a pretty handy tool. I've gotten a lot of use out of it in the past year or so, ever since moving from Microsoft Word to a general purpose text editor (currently Visual Studio Code) for my writing. Unfortunately, it's not perfect.

What's annoying about wc?

Here's an example of the output:

$ wc chapters/*
     105    2258   12563 chapters/c01-1.nv
      87    1582    8861 chapters/c01-2.nv
     113    2324   12933 chapters/c01-3.nv
      79    1312    7253 chapters/c01-4.nv
     384    7476   41610 total

In order of my give-a-damn level, that's word count, line count, and byte count. In order of appearance, it's line count, word count, and byte count. Obviously, there's a bit of a mismatch there... First off, I could not care any less about byte count. Second, line count is not, in fact, the statistic I'm looking for.

And, of course, thirdly, wc(1) counts # Chapter 1 as three words. -.-

What's different about ncount?

Here's the same query by ncount:

$ ncount chapters/*
c01-1.nv: 2251, 52 (95)
c01-2.nv: 1575, 43 (80)
c01-3.nv: 2317, 56 (96)
c01-4.nv: 1305, 38 (78)

In order of give-a-damn, that's word count (excluding nonsense), paragraph count (ncount ignores blank lines and headings), and the length of the longest paragraph in words. The columns are sorted by my level of give-a-damn. One thing that is missing is the overall word count, but that would take five seconds to add if I really wanted it. Maybe I will, eventually.

Longest paragraph?

I'm primarily interested in tracking the shape and feel of my writing. I know that sounds stupid to you, but hear me out.

Once upon a time, we all wrote on typewriters. We all wrote on paper. We all developed an instinctive feel for the flow of a document based on how many pages it was, how many inches were in each paragraph, and so forth. In the modern era, with digital tools, it's really, really difficult to pull that off. This is doubly true of tools like Visual Studio Code, which, by default, do not enforce any line wrapping or give you any sort of page numbering.

This isn't a problem when writing in Microsoft Word or another similar product provided that you always stick with the same format. You can get a "feel" for it, and it should be pretty reliable. I've worked to counteract that in Code by setting my line width to eighty columns so that a "five line" paragraph is always the same length--which is to say "approximately 80 words."

These factors--scene length, paragraph length, and sentence length--are critical parts of the shape and feel of a text. I don't have a problem with sentence length, and word counts are no big deal to trach with wc, but paragraph length is another matter. Tracking that is not only important but also difficult without the right tooling.

Is this dumb thing working?

Yes. Check the output above: between the fixed line length and tracking my paragraph lengths, I've brought down the length of the longest paragraphs from 150 words to under 100. In theory, this should translate to a better pace for the reader. I mean, I don't know for sure. I'll ask one.