October 2015 in review

This is the thirteenth of my month-in-review posts. I started month-in-review posts in October 2014. You can see my first post here and my previous month-in-review post here.

Some overall background

This month was quite busy for me work-wise, so my non-work output was very limited.

Wikipedia editing

For background information, see my site page about Wikipedia

I created the following Wikipedia pages this month:

  • Neil Barsky: Barsky is a filmmaker and journalist, best known for making the documentary Koch and for co-founding The Marshall Project (disclosure: I created that Wikipedia article).
  • Chae Chan Ping v. United States, better known as the Chinese Exclusion Case. The creation of the page was an outgrowth of my efforts to better understand the events surrounding the passage of the Chinese Exclusion Act in the United States.
  • Angell Treaty of 1880, also known as the Treaty Regulating Immigration from China. Like the previous page, this was part of my exploration into the Chinese Exclusion Act.

I also made a bunch of edits to timeline pages, including the timeline of Uber, timeline of Snapchat, timeline of Facebook, timeline of Twitter, timeline of Pinterest.

Content plans for November: I intend to make a few more pages on topics related to the Chinese Exclusion Act, and also finish my page on consular nonreviewability. I might also make pages on some court cases or BIA decisions related to migration, particularly those that have bearing on the idea of consular nonreviewability or particular social group. I don’t have clear expectations regarding how many pages I will create. I am also interested in editing existing pages, and also in creating new pages, on content related to charities and charity evaluation, in preparation for the end-of-year giving season.

Pageview counts and impact: The numbers below are for pageviews in October 2015, and are based on data collated by stats.grok.se and further summarized at Wikipedia Views. As with all numbers from stats.grok.se, note that these numbers exclude mobile pageviews but include bots. One day was missing; adjusting for the missing day would make all counts go up by an extra 1/30 (or 3.33%).

Comparison with forecast: The number of pageviews (201,317) was a little above the 90th percentile in probability distribution for the number of pageviews. Adjusting for the missing day continues to keep it above the 90th percentile but still well below the 97.5th percentile estimate of 225,000.

November forecast: After poorly centered forecasts for two consecutive months, with one of them having the actual value outside the 95% range, it’s time to rethink my methodology. It seems that there are a few pages where the number of pageviews could vary dramatically by month, and these drive most of the variation in the estimates. Two pages that appear to be most responsible for month-to-month variation are Internet.org and Park Yeon-mi. In addition, in the month of November, another page that I expect will get a lot of traffic is the page I created on Giving Tuesday.

Therefore, I will make a forecast for the distribution of total pageview counts for pages I created excluding these three pages.

  • 2.5th percentile: 95,000 pageviews.
  • 10th percentile: 110,000 pageviews.
  • 25th percentile: 125,000 pageviews.
  • 50th percentile: 141,000 pageviews.
  • 75th percentile: 166,000 pageviews.
  • 90th percentile: 190,000 pageviews.
  • 97.5th percentile: 215,000 pageviews.

Here are my percentile estimates including the three pages:

  • 2.5th percentile: 110,000 pageviews.
  • 10th percentile: 135,000 pageviews.
  • 25th percentile: 148,000 pageviews.
  • 50th percentile: 176,000 pageviews.
  • 75th percentile: 205,000 pageviews.
  • 90th percentile: 230,000 pageviews.
  • 97.5th percentile: 280,000 pageviews.

Note the huge room on the right, reflecting uncertainty about the number of pageviews of those three pages.

Subject wikis

General background information: My site page about the subject wikis

I did not spend any time on the subject wikis this month. Year-on-year traffic growth on Groupprops was quite small, at about 6%. It’s not clear what’s driving the variation by month in year-on-year traffic growth.

Metric Oct 2015 Sep 2015 Aug 2015 Oct 2014 Sep 2014 Aug 2014
Pageviews 111,203 73,510 41,813 104,663 60,637 36,844
Sessions 55,802 37,313 21,775 49,971 31,475 17,778
Pages/session 1.99 1.97 1.92 2.09 1.93 2.07
Pageviews in sessions with at least 5 pageviews 35,440 23,066 13,046 36,456 18,157 13,205

Here are pageview counts for the other subject wikis:

  • Market: 22,697 pageviews, versus 24,028 for October 2014.
  • Calculus: 24,430 pageviews, versus 22,170 for October 2014.

Comparison with forecast: The value of 111,206 pageviews was close to the median estimate of 113,000 pageviews, and within my 50% probability interval estimate (95,000-130,000 pageviews) and 80% probability interval estimate (80,000-150,000 pageviews).

November forecast: The number of pageviews in November 2014 was 99,940. Assuming a similar 6% year-on-year growth, my median estimate for November 2015 is 105,500 pageviews. My 50% probability interval estimate is 90,000-120,000 pageviews and my 80% probability interval estimate is 80,000-135,000 pageviews. The intervals have contracted somewhat owing to greater confidence about the slowdown in year-on-year growth.

Other things

I expect the extended weekend of Thanksgiving to be mostly work-free and therefore I hope to get some content creation work done at the time. I intend to have published, by the end of the month, one blog post for the Effective Altruism Forum and one blog post for Open Borders: The Case. However, I’m not very sure I’ll accomplish either goal.


I enjoyed listening to and bought the MP3 for Love Me like You Do, sung by Ellie Goulding and used for the movie Fifty Shades of Grey. I also listened to Gangnam Style and the female version featuring Hyuna and Psy, which I enjoyed but not enough to buy either.