Thursday, January 05, 2017
Scanty? Nope.
Couple days ago I said:
What was the best thing I wrote this year? The least trivial item in a year of trivial shit?
The last few months have been scanty and distracted because of the surgery stuff in Sept. Starting earlier, the Best Thing stands out instantly.....
Scanty? Testable observation. Updating my 'size-by-date' Excel that tracks monthly blog verbosity:
(The numbers are size of monthly archive file in bytes, not wordcounts.)
In fact the last few months show a steady upward trend in verbosity, accelerating from a relative plateau. The windstorms and long power outages in July 2014 and Nov 2015 aren't salient on this graph. The only short month in the last two years is April 2015, with no obvious reason. (April 130k, neighboring months 180k.) Good weather, no memorable events or troubles, last courseware project was finished in Feb 2015. Apparently I just didn't feel like writing. **
Just to be sure, I did a little Python processing of the files to cut out the HTML/CSS/JS overhead. Found the markers that start and end actual posts, and counted only real words between those markers.
Allowing for my total lack of Excel skills, it's pretty much the same graph with numbers/12. So the overhead didn't change through the years, and the overhead wasn't distorting the pattern. Why the factor of 12? Do I use long words? No. Median length of my real words is 6, mean is 5. Agrees with the traditional 5-character word in Morse and typing WPM calculations. The factor rises to 12 because of all the HTML overhead around and inside the posts.
The obvious mode-switch in late 2010, from wild variability to mostly flat, matches two important changes in Aug 2010. Throwing away the TV, and getting blood pressure under control. Which was the driving variable? I'd like to credit the former, but it's more likely the latter.
= = = = =
Later: No, I didn't go silent in April 2015. The archive file as stored on Blogspot is defective. It contains only a few posts from the last two days of April. The rest of April is present in the actual blog and looks like a normal-sized month. It can't be reached from the April archive, but it can be reached by hitting the March archive and then clicking the Newer Posts button.
= = = = =
Later again: Noticed that the bytecount was trending upward in the last two years while the wordcount was exactly flat. Am I using longer words? Yes. In 2014 the mean wordlength was 6.1 characters, and in 2016 the mean wordlength was 6.6 characters.
Mean word lengths by year (excluding 4/2015 defective file):
2008 had the shortest words. 2008 was also the most creative in literary terms, with a long story piece. 2008 was also peak blood pressure, before I started taking steps (lit and fig) to improve health. Which causation is valid? Dunno. Regardless of causation, it's clear that longer words = better health. Seems like a useful research topic for 'telemedicine'.
(The numbers are size of monthly archive file in bytes, not wordcounts.)
In fact the last few months show a steady upward trend in verbosity, accelerating from a relative plateau. The windstorms and long power outages in July 2014 and Nov 2015 aren't salient on this graph. The only short month in the last two years is April 2015, with no obvious reason. (April 130k, neighboring months 180k.) Good weather, no memorable events or troubles, last courseware project was finished in Feb 2015. Apparently I just didn't feel like writing. **
Just to be sure, I did a little Python processing of the files to cut out the HTML/CSS/JS overhead. Found the markers that start and end actual posts, and counted only real words between those markers.
Allowing for my total lack of Excel skills, it's pretty much the same graph with numbers/12. So the overhead didn't change through the years, and the overhead wasn't distorting the pattern. Why the factor of 12? Do I use long words? No. Median length of my real words is 6, mean is 5. Agrees with the traditional 5-character word in Morse and typing WPM calculations. The factor rises to 12 because of all the HTML overhead around and inside the posts.
The obvious mode-switch in late 2010, from wild variability to mostly flat, matches two important changes in Aug 2010. Throwing away the TV, and getting blood pressure under control. Which was the driving variable? I'd like to credit the former, but it's more likely the latter.
= = = = =
Later: No, I didn't go silent in April 2015. The archive file as stored on Blogspot is defective. It contains only a few posts from the last two days of April. The rest of April is present in the actual blog and looks like a normal-sized month. It can't be reached from the April archive, but it can be reached by hitting the March archive and then clicking the Newer Posts button.
= = = = =
Later again: Noticed that the bytecount was trending upward in the last two years while the wordcount was exactly flat. Am I using longer words? Yes. In 2014 the mean wordlength was 6.1 characters, and in 2016 the mean wordlength was 6.6 characters.
Mean word lengths by year (excluding 4/2015 defective file):
2008 had the shortest words. 2008 was also the most creative in literary terms, with a long story piece. 2008 was also peak blood pressure, before I started taking steps (lit and fig) to improve health. Which causation is valid? Dunno. Regardless of causation, it's clear that longer words = better health. Seems like a useful research topic for 'telemedicine'.Labels: coot-proofing, Metrology
¶ 1:07 AM
