Virtual World DMCA Takedowns Are Complicated

Mon 16 November 2009 | -- (permalink)

Pixels and Policy criticizes Linden Labs' claim that developing the capability to "allow IP owners to request that Linden Lab search for and remove all copies of an identified item created by a particular resident . . . is one of the most complicated tasks they have ever undertaken":

The most complicated task Linden Lab has ever undertaken? Isn't that a bit dramatic, even for a developer trying to shake a persistent and well-funded lawsuit? Major content sharing websites like YouTube and even the raunchy humor website eBaumsWorld have already integrated simple online forms for reporting IP theft, and use e-mail as a valid means for reporting DMCA violations. Is working in three dimensions so much more complicated?

Well, maybe yes, it is that complicated. First, the web forms on sites like YouTube aren't a panacea. Content owners still grouse about having to track down and report every infringing link, and argue that they should be able to just send an email to YouTube saying "all your SNL clips are infringing, so take them all down, even if we don't know where they all are." As far as I know, you can't do that even on YouTube (much to Viacom's dismay).

Technical Complexity

In Second Life, there are a variety of ways for one person to copy another's content. The method of copying will affect the ease with which Linden Lab can create a list of all the inworld items to disable.

  • Textures: If you know the exact UUID of the infringing texture, tracking down all the copies and disabling them is probably not too complicated. Slightly harder: Also disabling every skin, build, or article of clothing that incorporates that texture. Should they all be reverted to that old Missing Image texture? Very hard: Tracking down every copy of a texture that was ripped and re-uploaded, possibly with slight modifications.
  • Scripts: Matching every exact copy of a script that hasn't been recompiled (so the scripts still share the same bytecode UUID) is probably not too hard. A bit harder: tracking down every script that has identical text to a given script, regardless of the script's creator (ie, when a script has been copied and pasted into an item created by someone else). Very hard: Tracking down every script that is 70% (or 80 or 90%) identical to the infringed one.
  • Objects: Seems hard to me, but maybe LL has better tools than I'm picturing. The main problem with objects is that a box prim that I make today could be copied and turned into part of someone's skirt or hair tomorrow. Sometimes people build using prims created by others. This happens a lot when people purchase build tools that rez prims that are packed inside. While most give you the option of repacking the tool with prims you made yourself, many don't bother. So if Joe says that Sally is selling a copybotted version of a set of hair he made, but the creator of the prims in Sally's hair is Bob, then LL would need to run a filter like "all objects created by Bob whose last owner was Sally." Whether they have that ability I don't know.

Social and Legal Complexity

Suppose the Lab puts up an easy to use web form for reporting DMCA violations, and promptly disables content based on those reports. An unscrupulous businessman (or woman) might think "Hmmm I can damage my competitors' reputation and freak out his customers by filing a DMCA notice with some made-up claim that his content is stolen. When the customers' skins and clothes show up with 'Missing Image' the word will get out on the blogs and I'll have more of the market left to myself!"

To prevent this gaming of the DMCA system, the Lab will have to do some bare minimum of filtering and investigation of DMCA takedown notices. But that very act of investigating comes with its own risks. The longer the Lab waits to take down content, the angrier content owners will be about the Lab's inactivity. Even more importantly, if the Lab investigates a DMCA report and decides that there isn't infringement going on, but gets it wrong, then they will have lost the DMCA's safe harbor protection from claims of vicarious and/or contributory copyright infringement. The Lab has to walk a very fine line to get this right.

But I hope they do. Not all cases are borderline. Lots of them will fall into the category of "easy to track down technically, and easy to categorize legally." In those cases, I wouldn't mind seeing every copy of the infringing content removed, grid-wide, both those copies rezzed inworld and those lurking in residents' inventories. The cultures of "rip and share" or "rip and resell" would become far less prevalent if wearing a ripped skin, for example, meant the risk of one day seeing the words "Texture Removed for Copyright Infringement" written across your virtual body.