MetaAnalysis

This page indexes the analysis that has been done using the Infodump. You can do your own work, too, with various tools for working with the Infodump, including Infodumpster.

Major threads

• The original Infodump announcement.
• User matching thread.
• The August 2009 Infodump revival thread.
• A thread for analysis, post-November-2009-Favorites experiment.
• A thread about the quickest series of posts, some Infodump analysis, and a Corpus prototype

Analysis of users

• Burger-Eating Invasion Monkey:
  • The top self-favoriters.
  • graphs number of AskMe posts against userid number - MeTa comment
  • BEIM graphs number of comments by userid.

• Pronoiac posted a list of short usernames.

• Plutor has posted
  • a list of "all of the people who made at least 100 comments total, haven't participated since at least 1 Sept 2007 and whose last comment was in a thread with more than 100 comments" to see if long, contentious threads scared people away.
  • Average FPPs per user, Average questions per user, & Average questions per user with some overwhelming superstars removed, grouped.
  • Average answers per user, in groups.

• MetaMonkey:
  • plots the number of posts on the blue by each user Y-axis: number of posts by a given user - MeTa comment
  • the same for the top 675 posters only - MeTa comment

• signal posted dendrograms - clustering users by the tags they use.

• thrako plots the number of AskMe answers per user - MeTa comment

• FishBike is doing some awesome work.
  • looked for interactions between users, in the user-matching thread, updated in the Infodump 2.0 thread, for many users.
  • listed users with most favorites per comment, then again with over 1000 comments.
  • listed users with most favorites per post.
  • listed users with most comments per post.
  • charted the Genius Index, taking number of favorites & number of contributions into account.
    • using posts
    • using posts & comments.
    • update.
  • charted users with most favorites per contribution, counting both posts & comments.
  • charted users most / least mainstream in their favoriting tastes.
  • ranked users by "average delay between posts being made and their first comment in those posts," then broke it down by subsite.
  • graphed user contribution counts on mefi vs. AskMe, then charted relative percentages, then percentages over time.
  • analysis: do people tend to favorite their contacts more? update.
  • ranking "highest number of comments made by one user in a single thread," then as percentage of comments in the thread.
  • ranks "Special Guests," most favorited for under 5 comments, with commentary by cortex.
  • ranks most-commented deleted threads.
  • ranks those who have commented in the most deleted threads, then sorted by percentage of deleted threads after their signup.
  • runs some stats on time elapsed between new user registration & first posts.

• rakish_yet_centered looked for patterns in mutual favoriting, & provided a spreadsheet for the top users.

• vacapinta charted the users most added as contacts.

• jedicus charted users who most often had answers marked as best answers.

• mathlete:
  • looks into degrees of separation to mathowie, following FishBike's lead.
  • I think the degrees of separation from mathowie followed.
  • charts users by number of mutual contacts - Most Promiscuous Mefites
  • then charted most contacted - Celebrity Mefites - and also Groupie Mefites, who have contacted the most, with mutual contacts subtracted from the total.
    • examines the contact network.
  • lists self-contacters.
    • then provided full files for many of these categories.

Analysis of Threads / Posts

• rakish_yet_centered posted a list of threads with over 1000 comments, & the number of threads per year with over 300 comments.

• fantabulous timewaster graphed comments per posting time, probability of getting certain numbers of comments, & the delay between first & last comments (or unique users?) in a thread.

• thrako plots the number of AskMe answers by post Each point represents a post, and for each point the x-coordinate is log(number of answers on that post), and the y-coordinate is log( (number of posts with more answers)/(number of answers) ). MeTa comment

• Plutor has posted
  • average number of comments to AskMe questions posted at that time (twice)
  • a graph of "Percentage of AskMe questions with no more than 10, 5, and 1 answers over time"

• Cortex's stats on MetaTalk thread closures

• Andy Baio has a bunch of interesting graphs at Waxy:
  • most active threads

• gwint worked out then how many posts each mod deleted.

• chrisamiller charted the top users by number of best answers per day.

• hoppytoad checked out the ratio of favorites to comments to make graphs of percentiles (comments per thread, favorites per thread, average comments per user), & find threads with lots of "light" - many more favorites than comments.

• FishBike worked out:
  • that threads with certain tags got the most favorites & comments, by subsite.
  • which common tags were most often also used in the body, & which ones the least.
  • which tags had "the most comments associated with them, by year," & the most favorites, then by comments & by month.
  • threads that attracted the most first comments by new users.

• flatluigi looked at the longest Metatalk threads to date then broke it down by category (November 2009 - browser-killing long thread)

Analysis of Favorites

• pravit analyzed favorite inequality, focusing on users in the top 1% by highest favorite count and most favorites earned in the past year.
  • How much of the favorites wealth is owned by the top 1% most favorited posters?
  • Over the past year, what proportion of the favorites were taken by the top 1% favorite earners? What was the bar to being in the 1%? Were the top 10 richest favorite owners last year also among the top 10 highest favorite earners?
  • How much did the top 1% favorite earners contribute to the site as measured by post and comment count, compared to the 99%? How many favorites per comment do the top 1% favorite earners make on average, compared to the 99%? Did the top 10 favorite earners reach their status by posting a lot, or were they awarded more favorites per comment than others? Which 1%er had the lowest average favorites per comment? Which 1%er had the highest?

• Burger-Eating Invasion Monkey:
  • BEIM graphs the number of favorites against length of username The y-axis is the total number of favorites a user has received. It appears that there is a statistically significant relationship, with each added character to a username adding 2.8 favorites on average. MeTa comment
  • BEIM lists the top self-favoriters

• Plutor has posted
  • a list of most favorited users (also, on Music)
  • most favorited comments
  • Favorite accumulation during the November 2009 experiment (warning: huge thread)

• rakish_yet_centered posted posts with over 200 favorites.

• Pronoiac posted graphs of favorites vs comments for threads.

• tkolar listed the comments with the most favorites

Analysis of Time of day, or otherwise over time

• Andy Baio:
  • most active days by comments & by posts - (also most active threads, above)
  • number of posts & comments by month
  • number of new users by month

• Burger-Eating Invasion Monkey:
  • graphs number of AskMe question by hour - MeTa comment
  • asks when do you post a question to get the most answers - MeTa comment
  • asks when do you post a question to get the most favorites - MeTa comment
  • graphs the date of final comments on the blue - MeTa comment

• Galvatron posted some awesome & pretty graphs of volume of posting over time.

• hattifattener posted a page of graphs regarding comments & thread ages -
  • a histogram of the post's location on the front page at the time of each user's first comment on that post
  • how well a thread is able to attract new commenters over time
  • how long threads take to receive a comment

• Plutor posted a graph of activity by day & thread in response to a request from cortex

• rakish_yet_centered checks the timing of deleted threads, & also Postroad specifically.

• FishBike graphed "Activity by Site, Type, and Hour/Weekday/Day of Month" for the different subsites.

• Pronoiac analyzed:
  • new signups over time, annotated with significant events
  • the fraction of successful signups
  • incomplete signups - overall & pre-$5-signup
  • AskMe questions asked per month, Dec 2003 - Dec 2012.
  • Music posts per month, 2006-2011.
  • Overall monthly posts to Mefi, AskMe, Metatalk, & Music on the same graph, then on a logarithmic scale for more detail on MeTa & Music, through 2011.

Analysis of Best Answers

This data isn't currently included in the Infodump.

• tkolar listed "Top ten users by number of "best answers" for 2004-2006," pre-dating the Infodump.

Unsorted

• BEIM graphs postid, userid, comments, favorites, & datetime against each other

• The Metafilter Data Playground by null terminated. No longer in service.