automating notice and takedown
DESCRIPTION
Presentation given at Progress and Freedom Foundation seminar, "What Goes Up Must Come Down: Copyright and Process in the Age of User-posted Content," March 2007, Washington, DCTRANSCRIPT
Slide 1
GiantStepsMedia Technology Strategies© 2007 1
Automating Notice and Takedown
Bill RosenblattGiantSteps Media Technology Strategies
[email protected]://www.giantstepsmts.com
(212) 956 1045
Slide 2
GiantStepsMedia Technology Strategies© 2007 2
Two Competing Ideas
Notice and Takedown Provided for in copyright law Reactive Observe one’s copyrighted
material on a network without authorization
Send notice to network operator Operator removes work Example: YouTube
Filtering Private-sector technology Proactive Network operator adopts
technology to identify content When conetent is uploaded to
network, acoustic fingerprintingtechnology automatically identifies and blocks it
Technology vendors: Audible Magic, Philips, Gracenote, others
Examples: MySpace, iMesh
Slide 3
GiantStepsMedia Technology Strategies© 2007 3
Whose Responsibility?
Notice and Takedown Copyright owner:
– Monitor networks to identify works it owns
– Send notices– Cost: expensive– Scalability: terrible
Network operator:– Respond (reactively) to notices– Cost: not expensive– Scalability: not very good
Filtering Copyright owner:
– Feed filtering technology vendor content and other info to enable identification
– Cost: cheap*– Scalability: excellent
Network operator:– Buy and maintain technology– Cost: more expensive– Scalability: unknown beyond a
certain point
Technology vendor– Maintain database of fingerprints– Ensure accuracy of IDs
*Assuming small number of vendors.
Slide 4
GiantStepsMedia Technology Strategies© 2007 4
Automated Notice and Takedown(YouTube)
Forms for content owners to fill out to request takedown– E.g., web interface
For content owners: more efficient but not really more scalable
For network operators: highly scalable
Issue: is removal automatic or does it require network operator review?– Should network operator pay to mitigate risk of false positives?
Slide 5
GiantStepsMedia Technology Strategies© 2007 5
Filtering and DRM
When fingerprint is identified, content owner/licensor can choose action to take, e.g.:– Require payment– Check user’s subscription– Offer a free sample– Require user’s email address– Substitute an encrypted version – Any or all of the above
Original vision of Snocap and Mashboxx network
Slide 6
GiantStepsMedia Technology Strategies© 2007 6
Two Points Along the Continuum
Standardizing Notice and Takedown automation
Filtering plus copyright registration
Slide 7
GiantStepsMedia Technology Strategies© 2007 7
Standardizing Takedown Notices
Agree on standard protocol for sending notice– E.g., NTML (Notice and Takedown Markup Language)
Content owner sends standard messages to network operators– Via RSS feeds or similar expedient method
Improves scalability on content owner side slightly– Problem of observing unauthorized content remains– Allows content owners to send messages to multiple network
operators from same tool– And to add support for new network operators easily
Issue: requires agreement on content identification scheme– Artist and title probably not precise enough– 17 U.S.C. § 512 (3) does not specify
Slide 8
GiantStepsMedia Technology Strategies© 2007 8
Filtering Plus Copyright Registration
Neutral entity* – Standardizes on single acoustic
fingerprinting technology– Provides service to network
operators on cost recovery basis
Add fingerprint registration to copyright registration process
For copyright owners, cheap and scalable
For network operators, not too expensive, questionably scalable
Issues: Choosing the technology
– Antitrust concerns– Open standards– Vendors want to make profits– Patent coverage
No technology is 100% accurate – how to adjudicate disputes?
Setting appropriate fees Who pays???
– Content owners don’t pay for DRM…
*E.g., Copyright Office
Slide 9
GiantStepsMedia Technology Strategies© 2007 9
Comparison
1: Manual Notice and Takedown
2: Standardized Notice Automation
3: Fingerprint
Filtering
4: Fingerprints
With ©Registration
Proactive No No Yes Yes
Cost C: expensive
N: not expensive
C: somewhat cheaper than 1
N: same as 1
C: cheap
N: more expensive
C: very cheap
N: same as 3
Scalability C: terrible
N: not good
C: a bit better than 1
N: same as 1
C: very good
N: unknown
C: excellent
N: unknown
C: Copyright Owners
N: Network Operators
Slide 10
GiantStepsMedia Technology Strategies© 2007 10
Issues that Technology Won’t Solve
Who verifies identity of allegedly copyrighted work?
Who gets benefit of doubt in disputes?
Who pays?