Now that the year is over, I wanted to share a bit of fun trivia from our Google Analytics account: the ten posts with the most views in 2020.
- Growth and the case against randomista development (10,666 views)
- Are we living at the most influential time in history? (7,710 views)
- How long does it take to research and develop a new vaccine? (7,665 views)
- COVID-19 brief for friends and family (6,472 views)
- Concerning the Recent 2019-Novel Coronavirus Outbreak (5,772 views)
- Problem areas beyond 80,000 Hours' current priorities (5,538 views)
- Will Three Gorges Dam collapse and kill millions? (5,527 views)
- Reducing long-term risks from malevolent actors (5,421 views)
- Should you donate to the Wikimedia foundation? (4,540 views)
- UK Income Tax & Donations (4,389 views)
Honorable mention: The Drowning Child and the Expanding Circle (4,283 views)
I chose an honorable mention because it was nice to see how many people we've helped to reach Singer's work.
Notes on this list:
- People were very interested in COVID this year. Go figure.
- For context, all Forum posts combined got 745,614 total views in 2020.
- The Forum's content has a surprisingly long tail. Here are some graphs showing daily views for the posts not published in 2020:
Out of curiosity, how many views does the average post get? What about the 90th-percentile post?
Before I look anything up, what do you mean by "the average post"? The Forum used to be much less popular than it is now, with posts getting much less attention. Maybe something like "how many views did the average post from 2020 get in 2020?" And ditto for 90th-percentile posts?
I'd be interested in the total number of pageviews and unique pageviews per year for the whole forum, plus yearly growth (unless there already is a post with that info).
Rather than chasing down a bunch of stats to put into a comment, the best step for us seems to be putting up a "statistics" page that we occasionally update. I'm planning to add something like this in the next ~2 months.
Sounds great, and like the right call.
Good question. Yeah, how about views of the average post from 2020 in 2020? And ditto for 90th percentile.
Some rough estimates based on a bit of Analytics work (emphasis on "rough"):
How long does it take to research and develop a new vaccine? was probably viewed so much because it was cited by The New York Times article in 2018 to back up a claim that "it takes 10 years and more than $1 billion to develop a vaccine" (which obviously wasn't talking about the kind of situation were are in now). And I imagine that the New York Times article got some reads in 2020 as it became very relevant.
It performed well in some google searches, probably partially because of that link. I would guess that most users game from google, but I haven't checked, I just remember it doing well when I looked in ~May.
These are total pageviews, not unique views, right? So if I view the same article five times, it counts as five views, not one view, right?
Yes, these are total views. But using unique views changes very little; a few posts change position, and "should you donate to Wikimedia" drops to 12th, but the other nine posts in this list are still in the top 10.
I'd still be curious how many unique views there are - I'm pretty surprised at the high view counts above. I had expected the discrepancy between unique views and upvotes to be smaller.
Are there just a lot of silent readers who never upvote or do the same readers who already upvoted click on the post again and again (to read the comments)?
Yep, 90% of readers on LW and the EA Forum never vote. And 90% of voters never comment. This holds empirically for lots of forums.
Many of the posts here are either highly ranked for certain Google searches or were featured in a major news story, so I'm not surprised they brought in lots of silent readers who never considered making an account. Even the posts that weren't in those categories were shared widely in the broader EA community, and many of those people also don't have Forum accounts.
Unique views were pretty close to total views. Here's a quick screenshot of the two figures for the top ten posts:
I imagine that most repeat views are people reading comments, though it surprises me that the Wikimedia post has the highest fraction of repeat views (since I assume it has many non-EA readers who wouldn't care much about following the discussion). Maybe a lot of people who get ten Wikipedia ads over the course of a year wound up reading it a few times?
Generally for most engagement there is a vast discrepancy between viewers, people who interact and people who comment/post.
1% rule - link with more details.