MetaFilter activity stats
May 22, 2024 4:59 AM Subscribe
MetaFilter activity stats
Tracking monthly active users, posts, and comments on MetaFilter and subsites. Automatically updated whenever the Infodump updates.
There have been various activity stats produced over the years, including by Rumple (2020), tkolar (2018-2021), and this excellent thread by wesleyac (2022).
I thought it would be useful to have charts which always show the latest Infodump data. Comments and suggestions welcome.
Tracking monthly active users, posts, and comments on MetaFilter and subsites. Automatically updated whenever the Infodump updates.
There have been various activity stats produced over the years, including by Rumple (2020), tkolar (2018-2021), and this excellent thread by wesleyac (2022).
I thought it would be useful to have charts which always show the latest Infodump data. Comments and suggestions welcome.
Role: coder
Thanks.
What I thought was interesting was that activity trends are not just about having fewer active users — it's also that the average active user now makes 6-8 comments a month (across all sites), compared to a fairly steady 12 a month over 2006 to 2016. It may be that the same users are commenting less, and/or high-activity users have left and low-activity users joined.
The 2003-2004 MetaTalk "spike" is because early AskMes were stored in the MetaTalk database table, before Matt copied AskMe to a dedicated table. I've pushed an update which strips them out.
posted by Klipspringer at 12:07 AM on May 23 [1 favorite]
What I thought was interesting was that activity trends are not just about having fewer active users — it's also that the average active user now makes 6-8 comments a month (across all sites), compared to a fairly steady 12 a month over 2006 to 2016. It may be that the same users are commenting less, and/or high-activity users have left and low-activity users joined.
The 2003-2004 MetaTalk "spike" is because early AskMes were stored in the MetaTalk database table, before Matt copied AskMe to a dedicated table. I've pushed an update which strips them out.
posted by Klipspringer at 12:07 AM on May 23 [1 favorite]
Klipspringer: "What I thought was interesting was that activity trends are not just about having fewer active users — it's also that the average active user now makes 6-8 comments a month (across all sites), compared to a fairly steady 12 a month over 2006 to 2016. It may be that the same users are commenting less, and/or high-activity users have left and low-activity users joined."
First thing I thought was maybe the end of the politics megathreads, but that happened in July 2019 -- towards the end of a ~4-year downswing for MeFi -- so that was at best neutral and may have even helped. That MeFi metric has also been trending up the last three years, which roughly correlates with the increase in traffic, though it coincides with a decline in Ask.
There is a related Slack group that took on a lot of the political chat, but according to the analytics on weekly logins there it only added about 100 extra users (~70 of them posting) after the megathreads ended, and plateaued at ~140 total active in the years following the 2020 election. Activity is an apples-and-oranges comparison given it's basically live chat (and Slack hides a lot of relevant stats on the free tier), but even if every one of those was a former megathread patron who left, that hardly accounts for the difference given MeFi's orders-of-magnitude larger membership. (The Slack's also been slowly eroding in membership itself -- out of a total pool of ~700 who ever signed up, it's gone from a max of ~180 active users in 2020 to ~140 in 2022 to ~130 now. I could talk about a few insular/unhealthy aspects specific to it that might be discouraging some from sticking around, but overall it's a similar downward pressure that's been sapping a lot of other social spaces on the web outside of the major platforms.)
Anyway, if a site change as big as the megathreads ending barely changes the average comments/user trend, the driver is likely something more secular -- general depression/exhaustion/malaise would be my first guess, seeing as how the downtrend starts immediately after the 2016 election while the opposite happens post-2020. (And speaking of visible events, I'm really surprised that both COVID and the death of Google Reader in 2013 didn't have a larger effect.)
posted by Rhaomi at 7:31 AM on May 23 [1 favorite]
First thing I thought was maybe the end of the politics megathreads, but that happened in July 2019 -- towards the end of a ~4-year downswing for MeFi -- so that was at best neutral and may have even helped. That MeFi metric has also been trending up the last three years, which roughly correlates with the increase in traffic, though it coincides with a decline in Ask.
There is a related Slack group that took on a lot of the political chat, but according to the analytics on weekly logins there it only added about 100 extra users (~70 of them posting) after the megathreads ended, and plateaued at ~140 total active in the years following the 2020 election. Activity is an apples-and-oranges comparison given it's basically live chat (and Slack hides a lot of relevant stats on the free tier), but even if every one of those was a former megathread patron who left, that hardly accounts for the difference given MeFi's orders-of-magnitude larger membership. (The Slack's also been slowly eroding in membership itself -- out of a total pool of ~700 who ever signed up, it's gone from a max of ~180 active users in 2020 to ~140 in 2022 to ~130 now. I could talk about a few insular/unhealthy aspects specific to it that might be discouraging some from sticking around, but overall it's a similar downward pressure that's been sapping a lot of other social spaces on the web outside of the major platforms.)
Anyway, if a site change as big as the megathreads ending barely changes the average comments/user trend, the driver is likely something more secular -- general depression/exhaustion/malaise would be my first guess, seeing as how the downtrend starts immediately after the 2016 election while the opposite happens post-2020. (And speaking of visible events, I'm really surprised that both COVID and the death of Google Reader in 2013 didn't have a larger effect.)
posted by Rhaomi at 7:31 AM on May 23 [1 favorite]
Added day of week and hour charts. On the whole, people use MetaFilter to slack off during working weekdays — apart from Fanfare which sees an evening and Sunday-Monday surge.
posted by Klipspringer at 2:30 PM on May 25
posted by Klipspringer at 2:30 PM on May 25
Added charts:
- active users broken down by year joined - currently, about 75% of site activity is from users who joined more than 10 years ago, and 90% is from users who joined more than 5 years ago
- new active users - the site currently gains about 20-30 new active users per month. but more users leave
- cumulative active users - 47,500 users have ever been active
posted by Klipspringer at 6:08 PM on May 25
- active users broken down by year joined - currently, about 75% of site activity is from users who joined more than 10 years ago, and 90% is from users who joined more than 5 years ago
- new active users - the site currently gains about 20-30 new active users per month. but more users leave
- cumulative active users - 47,500 users have ever been active
posted by Klipspringer at 6:08 PM on May 25
Added ability to filter charts by year to make recent trends more visible.
posted by Klipspringer at 11:02 AM on May 26
posted by Klipspringer at 11:02 AM on May 26
What I thought was interesting was that activity trends are not just about having fewer active users — it's also that the average active user now makes 6-8 comments a month (across all sites), compared to a fairly steady 12 a month over 2006 to 2016.
This could still be related - if there's less active users commenting and posting then there's fewer posts and comments for the active users who are here to respond to.
posted by aubilenon at 7:48 PM on May 27 [1 favorite]
This could still be related - if there's less active users commenting and posting then there's fewer posts and comments for the active users who are here to respond to.
posted by aubilenon at 7:48 PM on May 27 [1 favorite]
Added chart of posts and comments by age of user account.
posted by Klipspringer at 3:37 PM on May 28
posted by Klipspringer at 3:37 PM on May 28
This is brilliant and something i've wanted to do myself for literally years, but never got around to actually doing it. Glad you were able to get it done; it looks fantastic!
I "knew" (unscientific gut feel knowledge) that a large percentage of the content was by old-timers here, but seeing that ~50% of the posts and comments were made by user accounts 15 years and older was still surprising. Looks like that cohort just crossed that 50% threshold recently...
posted by cgg at 10:03 AM on May 30 [1 favorite]
I "knew" (unscientific gut feel knowledge) that a large percentage of the content was by old-timers here, but seeing that ~50% of the posts and comments were made by user accounts 15 years and older was still surprising. Looks like that cohort just crossed that 50% threshold recently...
posted by cgg at 10:03 AM on May 30 [1 favorite]
In a continuing effort to procrastinate real work:
- added chart comparing registered with active users
- added chart of deleted posts - we're very well-behaved these days
- rewrote to use polars for data aggregation and sveltekit for frontend
- github keeps telling me the DNS is improperly configured, but it works fine for me. pls let me know if it's broken
posted by Klipspringer at 2:07 PM on June 7
- added chart comparing registered with active users
- added chart of deleted posts - we're very well-behaved these days
- rewrote to use polars for data aggregation and sveltekit for frontend
- github keeps telling me the DNS is improperly configured, but it works fine for me. pls let me know if it's broken
posted by Klipspringer at 2:07 PM on June 7
Added charts on what proportion of posts and comments are made by the most active posters and commenters, i.e. how skewed the distibutions are.
On MetaFilter, the top 1% posters make 10% of the posts, and the top 10% make 50%. This dipped in 2021-22, before reaching all-time highs in 2023-24 - I think driven by the the excellent daily posting of chariot pulled by cassowaries, chavenet, and others. Comments are more skewed: at the height of the 2016-17 megathreads, the top 1% commenters made 25% of comments, and top 10% commenters made 65%. Skewness of commenting declined sharply post-2017, before stabilising.
On AskMe, the top 1% of users ask 10% of questions. This trend is gently declining, which didn't match my priors. Beyond the top 1%, activity is still skewed but less so than on MeFi, with the top 10% askers and answerers responsible for 25% of questions and 45% of answers respectively. You can see the bumps in July 2017 and March 2022 when weekly question limits were liberalised.
posted by Klipspringer at 5:29 AM on June 9 [1 favorite]
On MetaFilter, the top 1% posters make 10% of the posts, and the top 10% make 50%. This dipped in 2021-22, before reaching all-time highs in 2023-24 - I think driven by the the excellent daily posting of chariot pulled by cassowaries, chavenet, and others. Comments are more skewed: at the height of the 2016-17 megathreads, the top 1% commenters made 25% of comments, and top 10% commenters made 65%. Skewness of commenting declined sharply post-2017, before stabilising.
On AskMe, the top 1% of users ask 10% of questions. This trend is gently declining, which didn't match my priors. Beyond the top 1%, activity is still skewed but less so than on MeFi, with the top 10% askers and answerers responsible for 25% of questions and 45% of answers respectively. You can see the bumps in July 2017 and March 2022 when weekly question limits were liberalised.
posted by Klipspringer at 5:29 AM on June 9 [1 favorite]
new 'dump just dropped
posted by Klipspringer at 4:41 AM on June 16
posted by Klipspringer at 4:41 AM on June 16
« Older Blog/Column: Humans and Technology... | A Servalan playlist... Newer »
- The overall graph seems to inversely track the growth of mobile social media, particularly the rise of the Facebook app in tandem with the growth of iOS and especially Android circa 2010-2011. MeFi has always been a desktop/laptop-tailored experience so the overall shift towards smartphones has not been kind, though I imagine it would have been a lot worse without the responsive redesign in 2014. There are certainly more opportunities to make signup and participation more mobile-friendly.
- The peak activity for MetaFilter is in October 2011 -- precisely when Google suddenly cut traffic by 40% overnight. New signups appear to crater soon after. Based on data I've seen, there has been some reversal on the search front in the last few years, causing MeFi to rank consistently better and pull in more traffic, which might account for the late stabilization; the increasing enshittification of Facebook, Quora, Twitter, and Reddit likely also plays a role here.
- Some noticeable peaks in activity: 9/11, the opening of $5 signups in 2004, the 10th anniversary in July '09, spring 2011 (Fukushima/Bin Laden), fall 2012 (Obama-Romney), assorted contest and theme months, the 2016 election bump, the 2020 election through January 6th, the introduction of free threads in January '22
- There's a weird glitch in the MetaTalk numbers for 2004 -- several months show dozens of posts per day and north of 500 per month, but that's not reflected in the Infodump data or the site archives.
posted by Rhaomi at 4:37 PM on May 22 [2 favorites]