YouTube’s content ID updates

Plagiarism Today has worthwhile commentary on the YouTube content ID update. Read it there.

What strikes me is the depth of the bureaucracy under construction at YouTube. That’s not to complain, because it seems reasonable – the complexity of the solution flows from the complexity of the problem.

There are three layers in their stack:

  1. There is a layer related to copyright infringement. This is about the safe harbor provisions of the DMCA and related law like the Grokster inducement decision. If Google is successful with this layer, it avoids getting sued.
  2. There is a layer for content ID. This identifies which works are embodied in uploaded videos, and maps them to works in recordings uploaded by copyright holder. If Google is successful with this layer it minimizes false positives (reporting matches that aren’t correct) and false negatives (failing to report matches that are correct). This is basically a science problem.
  3. There is a layer for content licensing. This strikes deals with copyright holders, sells ads, inserts ads, and pays the copyright holders.

Where is the user in this stack? Product. If users/uploaders are treated fairly and respectfully, it’s a win on the product management side.

The stack is a unified whole. Infringement and licensing are two facets of the same problem, content ID links them.

As Janko Roettgers explains the YouTube update:

Adding the DMCA to this set of tools could have two significant effects: It could help uploaders defend the use of third-party material when it is covered by fair-use exemptions – and it could help YouTube to convince more content owners to go down the path of monetization. After all, why chose the hard path when the easier one actually helps you to make money?

Coping with copyright law – the first layer – is a universal problem. I would be surprised if there is a single big Internet company that doesn’t get DMCA-based requests to take down content. But it’s joined tightly with monetization and licensing. YouTube is blazing the trail but won’t be the only one to go down it. You can assume that Yahoo, Microsoft, Apple, eBay, etc are all growing their own versions of the stack, each one with its own complexity.

With something like YouTube’s content ID program, licensing and takedown requests are sides of the same coin. Infringement is unmet demand. YouTube’s licensing program enables demand to be met.

Listening to Hollywood

The Internet Society goes to Hollywood:

As you are likely aware Internet Society North America Bureau Director Paul Brigner’s previous job was CTO to the MPAA. Since “switching sides” in March 2012 he has strongly pursued a program of increasing communication and engagement between the Internet community and big content, including recruiting the MPAA, the RIAA, and Disney as organizational members. This week he has led a foray by ISOC staff to visit major Hollywood studios – a listening tour. Accompanying him were Leslie Daigle (ISOC’s CTO), Markus Kummer (VP of Public Policy), and Konstantinos Komaitis (Policy Advisor). Today, September 27 2012, the group reported to the North America Chapter leaders via Google Hangout. Video is here.

Working remote

Some people can work remotely and some can’t. The difference is staying in sync.

Have an excellent internet connection. Have a working phone. Be on IM. Answer emails with low latency. Be on Skype. Have a working video camera for video calls.

Have a business-quality computer. An eccentric setup puts demands on your coworkers that create inefficiency – they aren’t going to use IRC in a console window just because you object to the Heminahaw License of the Foobar. Homebrewed linux setups are only ok if they have all the latest trimmings. Old Windows and OS X machines are never ok. Keep up with recent OSs and common tools.

When you have a question, ask it. When you have a casual stupid comment, make it. Call just as casually as you would say hello in person. Interact constantly.

Have a job-oriented working environment. Be at a desk. Be working during working hours. Get a coworking desk instead of going to a cafe.

If you don’t do all these things you aren’t ready to work remotely. Don’t kid yourself about your productivity. Stop working remote.

Stripe

The payment processing company Stripe is forwarding cease and desists to Chilling Effects.

We’re partnering with Chilling Effects. Developed at the Berkman Center for Internet & Society, Chilling Effects is run by the EFF and law clinics at Harvard, Stanford, Berkeley, and other universities across the country. Chilling Effects tracks the use of cease-and-desist notices to gauge the impact of legal threats to online expression. This database of notices provides important empirical data concerning the real-world effects of laws and policies.

From now on, whenever a third party sends a legal request for us to stop working with a user, we will pass a copy of the request to Chilling Effects. (We will, however, first give the user the chance to opt out of this disclosure.) Reporting take-down requests is standard practice at Google, Twitter, and GitHub today.

Chilling Effects is about takedown requests related to copyright infringement, but Stripe doesn’t say takedown requests, they say cease and desists. Which makes sense because Stripe doesn’t do copyrighted stuff, they do financial transactions. The trick is that they’re applying the process defined in the DMCA for non-DMCA stuff. Because it’s a pretty good rulebook for handling conflicts of all kinds.

Dated URIs

Sometimes you need globally unique identifiers, such as for ISRCs. One way to approach this is by using URIs – the uniqueness of the domain part of URIs is administered by ICANN, the uniqueness of the rest of the URI is administered by the owner of the domain. There’s a good balance between centralization and decentralization.

But ownership of domains changes, and when it does the identifiers associated with one ownership become invalid. Many types of identifiers, including ISRCs, need to be persistent. So you couldn’t use URLs for these.

This could be corrected by adding a date to the identifier. The new identifier would be a URL/date pair, like

[“January 1, 2000”,”http://example.com/1234”]

The date specifies a point in time when the identifier was valid. So if there is a valid identifier http://example.com/1234 on January 1, 2000, and then example.com comes under new management which doesn’t support this identifier, the dated URL continues to be valid.

This only works for URIs used as identifiers and not for URIs used as locators. Adding the date would not restore an old URI that was out of service.