Statistics on Warner Music Group Takedowns
It is no secret that Warner Music Group is issuing record numbers of takedowns to Youtube but until now I haven’t seen much in the way of statistics. I decided to conduct some quick analysis using simple measurement tools and the Youtomb database. I ran my numbers off the production database but it shouldn’t be any different than our public database. For details on how to access our public database see this post.
To start, I was interested in Warner Music Video takedowns only so I searched the database for all videos where the status was down for copyright reasons and listed WMG.
If you want to run these numbers yourself feel free to submit this query to our public database:
SELECT time_created, timestamp
FROM artifacts
WHERE status = ‘down:copyright:WMG.’
OR status = ‘down:copyright:Warner Music Group.’
I began with a simple test to understand the level of takedowns perpetrated by WMG in the first quarter of 2009 versus last year. To adumbrate the enormity of the increases here are the yearly totals (remember we are only in the 3rd month of 2009)
2008: 1440 takedowns
2009: 3707 takedowns
In just the first three months of 2009 Youtomb has recorded nearly three times as many takedowns than all of 2008. The numbers are even more ridiculous when they are broken down by month:
| month | takedowns |
|---|---|
|
2008-01 |
5 |
|
2008-02 |
4 |
|
2008-03 |
10 |
|
2008-04 |
19 |
|
2008-05 |
33 |
|
2008-06 |
7 |
|
2008-07 |
6 |
|
2008-08 |
4 |
|
2008-09 |
4 |
|
2008-11 |
13 |
|
2008-12 |
1335 |
|
2009-01 |
3241 |
|
2009-02 |
382 |
|
2009-03 |
84 |
Notice the bulk of the 2008 numbers are actually in December and arguably part of the same crackdown effort. I hypothesized that WMG is likely far more aggressive now with regard to video takedowns. To prove this I measured the “lifespan” of a video by subtracting the date of the takedown from the date the video was created. I used both mean and median measurements against a “crackdown” date of 2009-01-30. My assumption was that after this date the official heavy lifting of purging all the known videos would be done and they would start monitoring for anything new.
|
mean(days) |
median(days) |
|
|---|---|---|
|
pre crackdown |
241 |
238 |
|
post crackdown |
11 |
7 |
I rounded up into whole days just to make this easier to read but the trend is pretty clear. Digging a little deeper I broke down the lifespan of videos by month starting with November 2008 (just slightly before the crackdown) and measured the mean and median life spans again:
|
date |
mean(days) |
median(days) |
|
2008-11 |
54 |
46 |
|
2008-12 |
31 |
30 |
|
2009-01 |
20 |
19 |
|
2009-02 |
8 |
5 |
[author's note] I realize the use of tables to present data is a crime against design but its really late and I should get back to my studies. Promise to clean up the presentation at some later date…
Tags: WMG

October 8th, 2009 at 1:56 am
Great post!
I wonder if they stopped completely, now that WMG is back? It would also be great to understand what kind of videos keep getting removed, despite the agreement with Youtube.
October 29th, 2009 at 11:40 pm
First…very interesting, but not surprising. Have you had a chance to further evaluate the numbers for the remainder of 2009? I would be interested in knowing if they continued to follow the trend of Feb and Mar.
Second, is this YouTomb a database that you have populated with data obtained from YouTube? Do they have some type of open api where you could extract the data for research purposes?