the twitter effect

Rereading Alan’s post on his blog hiatus, where he takes a month off of posting on his blog to comment elsewhere, I was struck (as always) by the patterns in activity he described. I decided to take a closer peek at the activity on my own blog – I’ve been thinking a lot about discourse analysis lately, so it’s at least partially non-navel-gazing.

Here’s the graph for the first few years of life for my blog. It started out as a private, personal outboard brain, then kind of took off with a life of its own.

a pretty graph, about nothing

Interesting. This blog’s heyday was 2005-2006. A lifetime ago, in intartube years. Then twitter happened in January 2007. It would be _really_ interesting to run some latent content analysis on both posts and comments, to see if they’re different BT vs. AT. Are the activity patterns different? Is the content different? Linking patterns? etc… It’d be completely nonscientific, but fascinating nonetheless…

WPMU Post and Comment Growth

The group of WPMU rockstars at UBC’s OLT just whipped up a fantastic new plugin for administrators of a WPMU site to get a feel for the growth of the community. It generates a graph to display growth in numbers of blog posts and comments over time, and uses the Google Data Visualization API to let you interactively define data ranges to be graphed.

Here’s the growth of graphed for the last 2 semesters:


Another fantastic job by the OLT blogging platform crew. Now, to just add users and pages, and it’ll be perfect… 😉

wpmu activity reports using the blog_activity plugin

Jim Groom linked to a post by Patrick Murray-John with an interesting summary of the activity on – and I was curious about what activity patterns are on – so I fired up Sequel Pro and dug around in the raw data stored by the blog_activity plugin in the wp_post_activity and wp_comment_activity tables. The tables include aggregate and anonymous activity data for the last month.

There is a relatively new Reports plugin that could do much of this in an automated way, but it only supports generating activity reports for individual users or blogs, not aggregate reports.

Following is the MySQL code I ran to crunch the tables into usable data, which I then (cringingly) copied and pasted into (wincingly) MS Excel to generate tables and visuals.

Posts per Hour of Day

To get the number of posts published by hour of day, I ran this:

select distinct from_unixtime(stamp, "%H") as hour, count(*) as numberOfPosts from wp_post_activity group by hour order by hour;


Posts per Day of Week

select distinct from_unixtime(stamp, "%a") as day, count(*) as numberOfPosts from wp_post_activity group by day;


Comments per Hour of Day

select distinct from_unixtime(stamp, "%H") as hour, count(*) as numberOfComments from wp_comment_activity group by hour order by hour;


Comments per Day of Week

select distinct from_unixtime(stamp, "%a") as day, count(*) as numberOfComments from wp_comment_activity group by day;


Combining some of the data

Now that I’ve got the data out, it’s easy to combine sets to see what’s going on. Comments and Posts per Hour of Day:


and combined posts and comments per day of week:


What’s it mean?

I don’t know what it means. Mostly, I just like shiny graphs with lines that loosely correspond to something. Am I going to read anything into it? Nope. But if nothing else, it’s interesting to see that activity isn’t tightly synchronized with in-class time

Now, it’s clear that we’re nowhere NEAR the activity level of UMWBlogs, nor do we have the sustained activity (we don’t have The Reverend, after all), but I was surprised and impressed that the aggregate activity was much higher in “off” hours/days than I’d have guessed. Actual activity, outside of classroom hours. Who’d have guessed?