I’m a Transformers fan. For the last few years I’ve been active on the Transformers Wiki. And the TFWiki is currently debating whether or not to migrate our site-wide licensing from GFDL to CC-BY-SA3. Essentially a platform change in our site’s underlying “legalese” from Linux to Mac.

Sounds boring, right?

The complete meltdown of Western Civilization, after the cut.

The New GNU

On June 15, Wikipedia changed from a GFDL license to Creative Commons license, that’s the legalese governing what can and cannot be done with their content. The Creative Commons license is a sort of “human readable” legal boilerplate created in 2002.

Jimbo Wales (founder of Wikipedia) is on-record as saying that if Creative Commons had been available when Wikipedia launched in 2001, it would have been the license Wikipedia used. But since it didn’t exist yet, they were forced to use the GNU Free Documentation License, a license designed for “manuals, textbooks, other reference and instructional materials, and documentation which often accompanies GNU software.”

The Creative Commons license has many ‘flavors,’ one of which is “CC-BY-SA3,” or “Creative Commons — Attribution-Share Alike 3.0,” whose general terms are more-or-less equivalent to that of the GFDL;

1. Anyone can make a copy of or take and alter this work.
2. The resulting work must also be freely available for others to copy or alter under the same terms.
3. The original authors must be credited.

GFDL has a bunch of clauses and conditions relating to invariant sections and cover texts such as you might encounter with software manuals. Wikipedia simplifies this by not having any of those, allowing it to pretty much ignore half the license as not relevant to its content.

While GFDL was workable, it wasn’t designed with the type of content-generation / user-contribution-model used by Wikipedia in mind; so there was understandable pressure to switch to a new option.

What does “Reliscensing” mean?

To relicense does not mean “change the license,” as you might think… ‘re’ here means ‘again’; the old license is not discarded.
Content offered under GFDL has the legal requirement that all future versions of that content also be available under GFDL. No matter what is done to it, the content must remain available for anyone else to re-use under the terms of the GFDL.

Simply switching your wiki from GFDL to a Creative Commons license would not normally remove the GFDL requirement… it would mean that the content was available under both GFDL and CC, leaving it up to the person re-using the content to decide which license’s terms they prefer to be bound by.

This is very silly; GFDL and CC-BY-SA3 both have essentially the same rights and requirements. There’s little practical difference between them. Offering the content under both licenses is very much like McDonalds listing a “large soda” and a “32 ounce soda” and allowing customers to choose which they’d prefer. A large soda is 32 ounces. You’d get two of the same item, but with a receipt treating them like they were completely different things.

An amendment was added to GFDL 1.3 in 2007 that released wikis and similar “massively multi-user collaborations” from the requirement that all future content be GFDL, provided that they switch to the nigh-identical CC-BY-SA3 before August 1, 2009. GFDL is a forward-compatible license, so they can do that, just like a board or directors can conven a meeting to change the company charter. Such changes apply to all GFDL content, even stuff createdlicensed under older versions of the License. (You can opt-out of the forward-computability clause that allows the FSF to revise the license on you, but no one bothers to.)

Under this arrangement, Wikipedia’s article on “Gold Farming” as it existed on June 14 remains available for use under GFDL or Creative Commons. The version of the text prior to the changeover could be moved to any GFDL project (such as a private wiki) or a Creative Commons project with no legal problems. But the Gold Farming article has (as of this writing) been edited 6 times since the July 16th legal changeover. And the version of the article which includes those changes is covered by CC-BY-SA3 only. So it cannot be incorporated into a project whose content is GFDL– the switch was one-way and one-time, (before August 1.) Going forward, all future versions of the article will be Creative Commons only.

This is how GNU puts it. (Note: the following excerpt from GFDL 1.3 has been edited for clarity. Certian clauses and legal terms have been replaced in order to present the relevant text specifically as is applies to wikis. The struck-out section does not apply to wikis, but was left in because it looked funny without it.)

GFDL vs. CC-BY-SA3

Back when the Transformers Wiki was decamping from Wikia, I had the singular joy of reviewing the text of Wikia’s Terms of Service and the GFDL in order to figure out just what had to be done for the move. (And specifically to figure out if we could dump Wikia’s annoying link-export spam. Answer: Yes.)

Because anything worth doing is worth doing right (and slowly) I took the time to review the complete text of the GFDL (all 3 versions) and CC-BY-SA3, with a specific eye toward how they apply to wikis. This is what I’ve learned;

1. The common assertion that the Creative Commons License is easier to understand is a lie

   Creative Commons has an easy-to-understand graphical front-end that tells you what you can do with it. But as they note at the end… This is a human-readable summary of the Legal Code.
   This front-end carries no legal weight, it’s just a summary. And the legal status it describes doesn’t even match up with the legalese in places! (More on that later.)
   GFDL has a non-binding “Section 0″ that it uses for a mission statement, but it’s clearly labeled as such. The Creative Commons GUI presents a real danger of confusing users into believing the simplified GUI itself is legally binding.
   The actual legalese of CC-BY-SA3 is denser and harder to read than GFDL 1.3… but it’s also a bit shorter. On balance… GFDL is still the easier to understand document, even with all the stuff that’s irrelevant to Wikis… but not by much.

2. See Attached License

   The GFDL requires that any copy or derivative of the work include the full-text of the GFDL license itself. Not a big deal with electronic copies– you just provide a link. But if you wanted to excerpt any significant chunk of an article’s text for a book, you’d also have to publish a 3-page legal license.

   Creative Commons allows re-users to simply note that the work is licensed under the the Creative Commons CC-BY-SA3 license. A significant plus as Wikipedia becomes more and more cited and excerpted. CC also dumps on all the provisions for cover-sheets and invariant sections that don’t apply to wikis anyway.

3. ‘Attribution’ ain’t just a river in Egypt

   CC-BY-SA3 requires (this is heavily edited for readability) If You Distribute, or Publicly Perform the Work or any Adaptations or Collections, You must keep intact the name of the Original Author (or username, if that’s what was used) for attribution.

   That means that (for wikis,) every author who’s ever edited an article must remain credited forever. On a major popular Wikipedia page that might mean thousands of editors are legally required to be credited if that article is sourced, adapted or excerpted in any way. In short, an attached list of authors even longer than the license text GFDL requires.
   This is about the point at which you should be realizing that CC-BY-SA3 is no more ‘intended for wikis’ than GFDL was. It was written for individual artists remixing one another’s work.

   GFDL, by contrast, requires that re-users: List on the Title Page, as authors, one or more persons or entities responsible for authorship of the modifications in the Modified Version, together with at least five of the principal authors of the Document (all of its principal authors, if it has fewer than five), unless they release you from this requirement.

   Much saner. …until you realize that wikis don’t have Title Pages, (Title Pages are one of those ‘invariant sections’ Wikis lack.) Instead author credit is handled under a sort of sideways interpretation of sub-section I. (edited for clarity) Preserve the section “History” and add to it an item stating at least the title, year, new authors, and publisher of the Modified Version. If there is no section Entitled “History” in the Document, create one stating the title, year, authors, and publisher …

   This is actually what TfWiki.net ran into when leaving Wikia… Wikia was inserting a link back to itself in the footer of every document, claiming that this was a legally required ‘author credit.’ That’s foolish on its face– it’s a publisher credit. TFWiki.net ran a full history import– which included a record of every single edit (with usernames) to meet this requirement instead.

   This also means that the standard mediaWiki ‘article history export’ (which IIRC only exports the last… 200 edits?) exporter is technically in violation of GFDL, which requires that such credits be made easily available in a machine-readable form. Hrm.
   This is about the point at which you should be realizing that neither GFDL nor CC-BY-SA3 was written with Wikis in mind.

   GFDL tries to play nice with it’s Title Section allowing for primary author credits… but wikis don’t use ‘em, so everything defaults to the “history” section– which was never really designed for the purpose.

   In March 2009, TFWiki.net suffered a database crash that wiped out about 7 months of edits. The articles themselves were recovered, but the edit history (who did what) for those months were lost… which technically places us in violation of section I– the history has not been preserved, though there has been a good-faith attempt to do so.

   Here’s the thing– Wikipedia has billed itself as ‘the encyclopedia anyone can edit’ that anyone can use… but for the last 8 years, the GFDL terms-of-reuse technically required that anyone porting a copy of the article ‘preserve the history,’ of that article, which means a full-history export. Wikipedia has never offered full-history exports. This means that Wikipedia should technically have lost the right to use its own content under GFDL… but the reality is that everyone has simply been ignoring the issue as hard as possible.

   The credit-preservation requirements under CC-BY-SA3 equally absolute, but dump the ‘preserve the history’ aspect that’s such a nightmare for wikis. (Full-history exports are about 20 times larger than current-state exports.) Ultimately, the Creative Commons license section 8, sub-section C includes the magic phrase that really applies here; If any provision of this License is invalid or unenforceable under applicable law, it shall not affect the validity or enforceability of the remainder of the terms of this License.

   Author credit really needs to be treated differently on Massive Multiauthor Collaboration sites like wikis. CC-BY-SA3 doesn’t do that… but it does contain the magic language that allows you to ignore the problem legally while making a good-faith attempt to preserve the spirit.

4. In Perpetuity, throughout the Universe…

   Both GFDL and CC-BY-SA3 grant the licensee ‘world-wide’ (or worldwide) rights to re-use the content on a royalty-free basis. That language really needs to be updated; right now copyright ends at orbit, but as the commercial exploitation of space continues to grow, the “legal wild west” enjoyed by the space program will eventually settle down. It’d be nice if the next version of GFDL and Creative Commons got ahead-of-the-curve and extended their language to the Solar System to prevent legal shenanigans.

What is a Derived Work?

So, aside from a few piddling specifics about whether or not a license is attached and just how authors are credited… GFDL and CC-BY-SA3 are substantially identical. They both allow you to make copies or modify content as long as the new work is available to the public under the same terms.

GFDL phrases this as a terse permitting of “modification or translation” of the original material. It was written for Software Documentation after all…. It’s not like someone’s gonna stage an interpretative dance of it. Creative Commons, written for artists, is more verbose;

This is standard legal boilerplate. You can’t turn a book into a motion picture without paying the author.
But what about writing a sequel to a book? That’s not a re-casting of the original story… it’s an entirely new story told in a pre-existing setting. That’s something else entirely– surely it’s a derivative work, which the section goes out of its way to include?

That’s odd, US law seems to draw the line at abridged versions. A sequel to a movie would only be considered a derived work if it was nothing more than a re-arrangement of the events of the original movie with little or no original story added! (Insert your own joke.)

Under this logic, a new story told with someone else’s characters (an unauthorized sequel to Sleepless in Seattle, for example) would not be a copyright violation! It would be fair use!

Everything is a Derived Work

A combination of trademark and an expansive interpretation of copyright have combined to quash almost all most all such ‘unauthorized sequels’ told with someone else’s characters. The operative logic basically goes something like this:
It is impossible to separate a character from the story in which they first appeared.

There’s a certain validity to this argument. While every Spider-man story told since 1962 hasn’t involved him beating up muggers, virtually all of them have hung on re-treading his original guilt over his uncle’s death, his obligation to his Aunt May, or “With Great Power Comes Great Responsibility.” The character is the story, even if few characters are as clear-cut examples of this as Spider-man.

When you get to visual works, a similarly expansive standard has been applied. A copyrighted photograph reproduced as a background element on somone’s wall in a movie, barely glimpsed, is a copyright violation. Or, notoriously, a few seconds of the Simpsons visible on a TV in the background of an interview.

Fair Use is an unproven assertion

Fair use is: “not an infringement of copyright” under U.S. law. This is legally distinction from the way the rest of the world handles it; ‘fair dealing’ is a violation of copyright, but the manner in which it was committed provides a defense which justifies the violation.

On paper that sounds like Fair use is much better justified in the U.S.– no crime occurs here, wheres in the rest of the world a crime does occur, you just have a defense for it.
In practice, U.S. law has been distorted into a fair-dealing type system because it is quite common to have fair use challenged in court (or threatened to be challenged) as a copyright violation, even for uses that clearly fall under the doctrine. Fair Use is actually an assertion– no usage can actually be assured to be safe under fair use unless/until found ot be so in court– after being challenged.

The reality is that ‘fair use’ which is supposed to protect certain use of copyrighted material has been eroded into virtual non-existence. Any use of copyrighted material, must be legally accounted for.

Downstream Fruit of a Poisoned Tree

That’s all nice, but how does it affect Wikipedia?

In 2008, US presidential candidate John McCain gave a speech about the nation of Georgia which lifted several paragraphs, almost verbatim, from Wikipedia’s article on the country. He did not follow this with a 7 minute reading of the GFDL license. (Hilarious as that might have been.)

So was this Fair Use, or was the section cited enough to make the entire speech GFDL? Well Fair Use is a legal justification for using someone else’s work, but the uncited use of blocks of text from another work, even with the minimal paraphrasing seen here, is plagiarism. With no sources cited, let’s just call this ‘theft’ (it’d be regarded so in academia) and say that Fair Use doesn’t enter into it.

So the senator has mixed chunks of a GFDL source into his speech. If this were a proper relicensing then the speech (and his performance of it) would become GFDL. But the senator didn’t include a GFDL notice. Lets look at GFDL section 9, shall we? (edited for clarity)
You may not copy, modify, sublicense, or distribute [the Wikipedia article on Georgia] except as expressly provided under this License.* Any attempt otherwise to copy, modify, sublicense, or distribute it is void, and will automatically terminate your rights under this License, but does not terminate the licenses of parties who have received copies or rights from you under this License.

See that asterisk in there? That’s my addition. ‘As expressly provided’ here means in accordance with all the requirements of the GFDL… such as the requirement that a modified work also be GFDL, and include the full text of the GFDL license. The real critical section is that second part– McCain’s failure to cite and provide a license does not in any way prevent the work from being GFDL, nor does it stop anyone from re-licensing it.

Not a big deal, right? So the speech is GFDL, it’s a speech… he wants to get it out there!

So what happens when the speech is re-used? We’ll ignore news programs for the moment (since there are broad protections for fair use when documenting current events) and say that, for example… the speech is included in a later documentary about the 2008 election. That documentary becomes GFDL, even if the film-makers were unaware of it.
An unrelated interview from that documentary which is later re-cut for a Motion Picture to show the senator reacting to news that space aliens are invading… causes that movie to become GFDL. Anyone can make a copy, legally, or remix this Hollywood Blockbuster to their heart’s content.

That all of this stems from a documentary failure on the part of some speechwriter is irrelevant. If it was a normal copyright violation, you’d call this ‘downstream liability.’ In 1995 a piece of public art which had been donated to the city of Los Angeles, but whose copyright was still owned by the author made it into the production of Batman Forever. And by ‘made it into’ we mean “had a giant set reproduction made, and was featured in the trailer, poster, mugs, t-shirts…” Oops! The fact that the source of this screw-up was the City of Los Angeles not realizing that they couldn’t give away the rights to this sculpture in no way prevented the artist from seeking damages directly from the t-shirt manufacturers for every single violation.

But under GFDL, instead of cash damages, the ‘downstream liability’ is everything downstream becoming GFDL, something anyone can make a copy of.

CC-BY-SA3 works in an identical manner.

Giving Notice

Of course, all this could be avoided if McCain’s re-use of the GFDL material had been fair use– a citation of the source it came from and a statement that ‘this is fair use.’ TFWiki.net, which obviously contains a lot of images from Transformers, has to do this for every image on our site.

It’s not that claiming Fair Use necessarily prevents a re-usage from being a copyright violation– too large or too ‘central’ a chunk of the original material and it still might be illegal especially depending how it’s used. But because fair use is an unproven assertion until it’s challenged in court– the ‘downstream liability’ would not occur until and unless it’s found not to be fair use in a court of law.

A proper citation of fair use is like a legal dam; liability cannot flow into the derived works while the dam still stands.

The Hell You SA3

Remember back when I said that part of the Creative Commons GUI don’t actually match up with the underlying legalese, and it’d be important later?

The Creative Commons GUI claims that the right “to Remix — to adapt the work.” is governed by its license.

And remember when we discussed how derived works are limited to re-castings of the original work? Integrating parts of copyrighted works into a new work isn’t (or at least should not be) a copyright violations at all.

The Creative Commons GUI claims the most expansive possible definition of derived work– the kind of ‘everything is ours forever’ approach that the Free Culture Movement is always complaining has gutted the public domain and had a stifling, chilling effect on artistic free expression.

And remember that the GFDL actually claimed the narrowest possible definition of a derived or modified version? It was written to govern software documentation– it didn’t much care about preventing someone from staging “LISP Build: An Interpretive Dance.”

So now you’ve got two ostensibly identical licenses that fall on completely opposite ends of a broad spectrum of what can be done to original material under fair use. GFDL allows virtually everything that’s not related to documenting software, Creative Commons allows virtually nothing– any work derived in any way must still be licensed under CC-BY-SA3.

It may now be occurring to you that the ‘fruit of the poisoned tree’ argument I made regarding John McCain’s speech doesn’t actually work– he stole from GFDL Wikipedia, and GFDL reserves a very narrow definition of applications as also requiring a GFDL license.

But it would work under a Creative Commons license, which uses the broadest possible definition.

Wikipedia switched to the Creative Commons CC-BY-SA3 license on June 15, 2009.

Slow Pressure Over Time

Picture a world where Wikipedia, the most-referenced and least-cited resource on Planet Earth is infected with a disease. And using the content from Wikipedia without proper legal protection (citation and a claim of fair use.) infects any new work it was based on, it’s now a carrier.

CC-BY-SA3 diffuses slowly over time. Blog entries which draw on the site become CC-BY-SA3. From there it might jump into news articles– or art pieces (I like to put original art on my blog entries, and that art would become CC-BY-SA3. Imprudent use of Google Image Search would infect all manner of things.) Three or four generations out, its origins entirely laundered, it’s no longer possible to tell what’s CC-BY-SA3. Failure to note that a blog entry is CC-BY-SA3 doesn’t stop the re-usage of others from being CC-BY-SA3, because failing to abide by the terms of the license doesn’t prevent you from being a carrier.

The production department of a TV movie uses a graphic they got off the web for a sign. The image was CC-BY-SA3. By the expansive interpretation of copyright law that’s been so popular in the last few decades… the movie is now CC-BY-SA3. Anyone can make a copy, freely and legally.

Unrealistic? Maybe not as much as you’d think. Superstar comic book artist Greg Land is notorious for tracing images he finds on Google. Normally this is a minor copyright concern– but if he traces an image from someone’s FLCKR account that was licensed CC-SA3, the entire issue could become CC-SA3. Legal test cases involving Creative Commons indicate that courts will uphold this– and if Land’s employers are unhappy about the fact anyone can now make a copy of their comic, they can sue him for damages.

But what it mostly does is create massive ambiguity. Was Land’s use of the FLCKR photo fair use? Maybe, maybe not. But unless the issue included a fair use claim, there is now a reasonable basis to claim that the issue is now part of the Creative Commons.
And reasonable basises are what safe harbor laws are made of. It took half a decade and a supreme court to win a single legal judgement against The Pirate Bay even when a vast majority of the content it served were blatant copyright violations.
When comic books, movies and episodes of TV show begin to fall into a limbo of “could be available under the Creative Commons,” the burden is now on the alleged copyright holder to prove that the work isn’t creative commons before they can demand it be pulled. (Well, they can demand it all they want… but a site like that now has a genuine legal basis for ignoring them.)

The burden of proof shifts from a situation where individuals have to include disclaimers of fair use (and even then face possible legal reprisals) to one where the large media companies will have to include disclaimers of fair use, trying to ward off their own works becoming infected by the increasingly-omnipresent CC-BY-SA3, which is like the public domain– except that you can’t use it without also being infected by it, becoming this new sort of pseudo-public-domain where anyone can copy.

By 2030, either almost everything published everywhere in the world will be CC-BY-SA3, or we’ll be heading for a series of worldwide Supreme Court showdowns as the same large media companies that spent decades claiming that they owned everything discover that that same expansive interpretation of derivative work they championed may now cripple their ability to own anything.

Cui Bono?

This whole argument is made of sticky-tape and dreams… just like the rest of our legal system. It’s a brief; a possible scenario. But test cases involving the Creative Commons (mostly around FLCKR accounts) seem to indicate that this basic premise is legally valid, and the text of the licenses… says what it says.

It’s possible that the world will go on functioning with no problems whatsoever. The infectious nature of the CC-SA3 license will be ignored, just like everyone ignored how Wikipedia wasn’t properly compliant with the GFDL’s attribution clauses.

The real question you might want to ask yourself then is why there was such a big push for Wikipedia to change licenses? Because CC-BY-SA3 is a slightly better license, still horribly unsuited for wikis? Because getting every major source online to use Creative Commons allows for portability of content? It doesn’t. CC-BY-SA3 content cannot be ported to a CC-SA3, CC-SA-NC or CC-BY project… each license is it’s own thing, legally, they just share a common umbrella name. Is it because the ‘must include the full text’ proviso that everyone ignored was too cumbersome?

This is a video of the November 30, 2007 iSummit in San Fransisco where Jimbo Whales announced Wikipedia had brokered an agreement for the switchover.

Why does the room go up when Jimbo makes his announcement? Jubilant shouts of “we nailed it?” Why does Lawrence Lessig hug Jimbo Whales and say this announcement is the greatest event in his life after the birth of his children?

This move was pushed for, hard, by the radical arm of the Free Culture movement. And while Wikipedia’s important and all… why is change from one piece of legalese to an ostensibly identical piece of legalese causing blissed-out joygasms from the people who most understand what these licenses mean, and are most vocal about the need for radical change in how copyright is handled?

The real question is what do they see that has them so excited? Is it the year 2030?