Word Count Bug Hunt - help appreciated

[KB]

Another update (in case anybody is listening ). I'm not sure, but I think the bug may have something to do with saving, whereby the text of a document gets saved without the meta-data (including word count). It's still hard to trace, though, because both of these should be happening at the same time.

So: I have made some minor amendments to the way save works, and I have also included a "Fix Statistics" feature for the next beta, which will correct any stats that have gone awry for those people who have run into the bug. (I might even leave this in the finished product just in case anybody edits the RTFD files that are contained within the .scriv wrapper.)

I would still love it if somebody could find a case in which word count went off every time. My theory is that it would happen in circumstances something like:

1) Saving an edited document. At this point, everything would look peachy, until:
2) You close and reopen the project. Suddenly the stats are off.

(Because meta-data such as stats haven't carried from one session to the next.)

The hunt continues...

[dunx]

I can't offer any additional information on the off-by-one errors you've been working on here, but I have to say that the word count Scrivener reported was way off compared to that from the nano word count validator.

nanovel word count in Scrivener: 51,316
reported word count from validator: 52,937

Now, I assume that the Scrivener word count does not include chapter titles and the like which the nano validator must do, but that only accounts for 60-70 words. Where might the others be coming from? I didn't export synopses or notes.

I will feed the same text into other tools to compare, of course, and it may be more an issue with the famously generous nano validator code than with Scrivener, but it is interesting.

[KB]

This sounds pretty normal to me. Although 1,600 words may seem a big difference, it depends on what each word counter really counts as a word. Scrivener is quite conservative in this regard, partly because I decided I would rather write more than enough than less than enough, but mainly because it only counts real words and not punctuation.

Are you a fan of hyphens/em-dashes when writing? If so, that may account for the difference. For instance, consider the following:

slight - pause

Scrivener counts that as two words, because it doesn't count punctuation - even punctuation that stands on its own - as words. Word, however, does, so Word would count the above as three words.

Cheers,
Keith

[janra]

I think that's exactly what it is. My Scrivener wordcount is about 1700 words lower than my official NaNo wordcount, and it's because Scrivener seems to be much pickier about what it counts as a word.

NaNo uses the good old unix utility 'wc' which basically says that any non-space characters between two space (or tab, or newline) characters is a word. So: em-dashes, if you put spaces around them, count as a word; scene delimiters count as a word, titles count, and so on. On the other hand, Scrivener seems to count a hyphenated word as two words, while wc counts it as only one.

How exactly the NaNo wordcounter works seems to be a hot topic over there for some reason ;-)

-janra

[djfrantz]

here is my experience during NaNo...
my project has 14 'chapters' all seperate documents.
during NaNo i went from beta 0.1.1b to 0.1.2b, just replacing the program.
scrivener says that i am at 46,031. exported the draft to plain text, no titles, no synopses.
NaNo validator says 50,869.
unix wc also says 50,869.
copy the text file from Textedit and paste back into scrivener and scrivener says 46,031
Word says 45,996
Ulysses says 46,037
and Pages says 45,987

i know this is no real help, just interesting...

[KB]

Again, that sounds right. You can see that Ulysses and Scrivener use similar methods for their word counts, and that Word and Pages are only slightly different. There is nothing strange in this - it is common for word counts to differ between programs, because there is no real standard... (I know that seems strange, as it is like saying that nobody really agrees on what a "word" is!)

[amberv]

It is certainly a step up from just handing over a thick stack of typewriter written pages and saying 250 per page.