When I got this stack of emails, the one question I wanted answered was, how much money do they make, and how can we, as a community, make that number smaller. I haven't finished this analysis, but here are the preliminaries:
My first clue to MediaDefender's pricing structure is the small "monitoring contract" that nets MD $10,000 a month.
(see
Statement of Work, "FW: Next steps with testing," 6-20-2007; see
Contract Draft, MediaDefender Agreement (Configuration Testing) (v1 6-28-07).doc email on 7-13-2007, and
fully executed contract, MediaDefender Agreement (Configuration Testing) (v1 6-28-07) (Exec)-2.pdf, faxed from UMG on 7-11 and in the email file as of 7-17-2007).
Then the big dog: the 2006 and Randy's proposed changes to 2007 contract with Universal Music Group,
Standard Contract for Anti-Piracy services (turn to page
MediaDefender UMG Agreement DRAFTv1.doc from email on 8-3-2007 in your hymnals).
This is a redline draft, so we can see the 2006 final draft and changes proposed for 2007.
Notes from "Standard Contract for Anti-Piracy Services"
Basic pricing:
$4000 per album-month
$2000 per track-month
$1000 per developing artist album-month (<500,000 units expected monthly volume)
$1000 per month for monitoring single tracks out of previously monitored albums
$500 per month for extended (long-term) monitoring of single tracks out of previously monitored albums
UMG paid
at least $3,500,000 (minumum value to MD) over the course of the 2006 contract (90% of UMI's anti-piracy budget is to be spent with MD).
If someone can extrapolate total tracks and albums monitored, i'm sure you could draw it out of their excel sheets, then apply those figures to these prices and we know about how much they make from the music industry.
We know that record companies and movie houses want to protect things a whole lot more in the initial days/weeks of release. MD's pricing reflects this, as does their "Leak Detection" department, which exists solely to watch for things to pop up on popular p2p sites.
In this contract and probably others,
if MD can't meet its performance metrics, UMI doesn't pay (up to 50% of the contract fees can be waived, with greater penalties, including termination, for greater non-performance).
MD calls it an "SLA Credit" when it's
too easy for someone to find the right files on p2p sites; if you're paying for their anti-piracy expertise, you pay a lot less when your company's songs are easy to download. Randy notes in his email that the SLA credits on BitTorrent is already hurting them:
How would we want to go back on the BT SLA. Since the results have been so bad, we have to position it as us not trying to game the SLA. Maybe focus on the new Uber 2 techniques.
Want to bankrupt MediaDefender? Pay close attention to how they gauge performance,
MD's clients get service credits (i.e. $$$) when authentic files are easy to download. Universal Music's contract stipulates they hire a vendor to run 80 tests per month on the following networks (listed in order of weight, i.e. from maximum possible monthly tests to minimum):
- Clients Not more than __ out of the 80
- Kazaa, Kazaa Lite 30
- Limewire, Morpheus, Shareaza 30
- eMule 30
- Ares 25
- uTorrent, BitTorrent, Azureus 20
- Morpheus, Shareaza 20
- DC++ 15
- MxPie 15
- Piolet 10
- Soulseek 10
- Filetopia 5
In otherwords, they could run as many as 30 tests on Kazaa, 30 on Limewire, and 25 on Ares in a month, but never more than 5 tests on Filetopia or 10 on Soulseek.
A test-run involves searching for "artist track title" or "artist album title," selecting results based on 'best' returns, low/med/hi users offering, and random selection, and let audio files download for 2 hrs, archives for 10.
At the end of 2- or 10- hours, analyze the files: how many complete, pirated tracks did you get? With 45 max attempts per test, this table lists the number of authentic files you have to get to fail the test:
Download Attempts Made Fail if Pirate Count is
- 45 6
- 40 5
- 35 5
- 30 5
- 25 4
- 20 4
- 15 3
- 10 3
These are pretty low hurdles, get four out of 25 downloads to complete with authentic files, MediaDefender fails the test. Every failure makes them look worse.
Bittorrent is handled differently:
Same premise, from any of the torrent sites listed on page 23 of the draft agreement, search for albums / tracks, open up to 45 torrents for download (or as many as you can find) and see how fast you get the files. Fail the test by:
. 50% successful, real files downloaded in 6-hours
. 25% successful, real files downloaded in 4-hours
. This many successful, real files downloaded in 2-hours:
Download Attempts Made (2-hr) Fail if Pirate Count is
- 45 4
- 40 4
- 35 4
- 30 4
- 25 3
- 20 3
- 15 2
- 10 2
So if you find 25 results of a given album on the torrent sites, and 3 or more of those complete with real files in a 2-hr time period, that is a failed test.
This list tells us, among other things, that the
top three networks you should be supporting with clean data and fast downloads are Kazaa, Limewire, and eMule, followed by Ares (apparently BT is already too clean, they're losing ground there). The better those networks perform across the board, the less money MD makes. (see page 18 of the UMG DRAFT contract for schedule of SLA credits per failed tests)
Other notables:
MD spent a -lot- of time developing miivi, and apparently they were going to roll the same concept out on viide; they definitely think there is money to be made in the streaming video market by combining streaming video and whatever's in that client-side code (at one time they draw comparisons to browser-based bittorrent). Unsure how this fits into their plan, if its an extension of the Marketing Intelligence program or something new, but you can bet we haven't seen the last of that miivi code (assuming MD survives this debacle, god forbid).
MD doesn't like anyone new getting into the youtube game, and they have several emails noting design details of video and music sharing sites. I assume they expected some first-to-market advantage with miivi that's either about to disappear or long gone.
.
