Can Anti-DRM Clauses in Content Licenses be Free?

This essay has been written as a required deliverable for the course T-76.5753 Law in Network Society.

The GNU Free Documentation License 1.2 and the Creative Commons 2.5 suite of licenses, the best-known “free” or “open” content licenses, both contain anti-DRM clauses. The Debian community, which is very strict in its interpretation of what is free, tends to consider the anti-DRM clauses in these licenses non-free in terms of the Debian Free Software Guidelines. See the summaries by Manoj Srivastava on the GFDL and by Evan Prodromou on CC. (Of the Creative Commons licenses only Attribution and Attribution-ShareAlike have any chance of being formulated as free.)

Are anti-DRM clauses a good idea? Are the current clauses merely badly drafted and an anti-DRM clause in general could be free? Or is any anti-DRM clause inherently non-free?

The Purpose

Let us consider why anti-DRM clauses exist. The entire purpose of DRM (Digital Restrictions Management according to its opponents) is to incarcerate a work and to deprive the recipient of the work of the freedom to do things with the work without contrived technical barriers. Therefore, DRM is inherently contrary to the GPL-style notion of freedom that requires freedom to be passed on downstream.

Moreover, anti-DRM clauses are also anti-anti-circumvention clauses. Content industry incumbents have lobbied for Draconian anti-circumvention laws that tend to make some part of the process of the circumventing “effective” technical measures a criminal offense and the copyright holder cannot waive his “protection”. This means that entangling a work in DRM puts the recipient at the risk of criminal prosecution if he exercises his freedoms otherwise granted by the license. Putting your neighbor in such a position could be considered unethical. Certainly, it defeats the point of granting a license to modify and redistribute the work.

The Arguments Against

Three arguments have been presented against the anti-DRM clauses of the GFDL and the CC licenses. First, it has been argued that the clauses are worded in such a way that they prohibit encryption or any kind of access control (such as basic HTTP authentication or removing the read permission in a Unix-like file system). The second argument is that the formulations cover private copies. The third argument is that prohibiting the use of DRM even if a copy without DRM is provided alongside the copy with the DRM goes beyond the restrictions of GPLv2 (and GPLv2 is assumed to be about as restrictive as you can get while still being on the free side).

DRM vs. Access Control and Privacy Measures

The first argument seems almost like a straw man. However, since it has actually been presented, it needs addressing. The essence of DRM and the essence of access control are crucial different. The essence of DRM is to let a person see or hear a work while technically preventing him from making a copy. (The technical absurdity of this is protected by law.) Access control or encryption for private communications, on the other hand, either prevents access to the bits or grants access without trying to technically control what happens with the bits once the access has been granted.

The Debian-style notion of freedom does not require one to provide copies of a work to everyone. However, when someone is furnished with a copy, it needs to happen on free terms. Therefore, as far as the first argument goes, an anti-DRM clause could be free if it were worded so that anyone to whom a copy of the work is provided must be given access for all purposes otherwise granted by the license without technical measures intended to curtail some activities. The problem is with the wordings in the licenses that deal with “technological measures that control access or use of the Work” or “obstruct or control the reading” of the work in general instead of scoping the clauses to controls devised against parties who have been furnished with copies or to whom the work has been provided for ephemeral viewing (e.g. by broadcasting or streaming). I believe that as far as the access control argument goes, the licences could be made free by drafting them better.

Private Use

Also, private DRM-infected copies for technical reasons should obviously be allowed in order for a license to be free, because it is generally non-free to restrict what people do with copies they keep to themselves.

Using DRM and Providing Also a DRMless Copy

Being able to rerecord the work or to extract it from a copy without committing a crime (or using a tool the creation or distribution of which is a crime) is essential for exercising the rights provided by the Creative Commons suite of licenses, because the licenses do not guarantee the recipient access to the preferred format of making modifications to the work but instead it is up to the recipient to exercise freedom with the copy she is given. (Since the Creative Commons licenses are intended to be applicable to music and movies and the preferred format for making modifications to these may be over a thousand times larger than the typical distribution versions in byte size, requiring the source to be provided would be impractical.) The GFDL, on the other hand, requires the recipient to be given access to a “Transparent” copy of the work anyway, so the anti-DRM clause of the GFDL is mainly an anti-anti-circumvention clause that mitigates the risk of the recipients of copies to commit a crime and works as a poison pill against DRM-only systems.

Devices that require DRM are not a big problem presently. DVD players will play discs that don’t use CSS, iPods play also DRMless MP3 and AAC files, and Adobe Reader does not require PDFs to be encrypted. Therefore, it would make sense for those who object to DRM and anti-circumvention legislation to produce a large pool of compelling content that comes with an anti-DRM poison-pill clause. That way the pool of such content would work as an incentive against creating devices and player software that refuse to play DRMless content.

Even if a device requires content to be DRM-infected, applying DRM privately e.g. just before transferring content from a desktop computer to a player device would be allowed if the anti-DRM clauses were fixed not to apply to private copies (and one didn’t distribute preloaded players). This might not be possible if the DRM system requires a central authority to sign the files.

What this really boils down to is this: Is a license free if it forbids you from distributing DRM-infected copies even if you provide a DRMless copy of comparable or better quality also? The DRMless copy would most likely be in a different format, but then so is a source CD provided with a printed book. It is well-established that ensuring downstream freedom is free, so the case where only a DRM-infected copy is provided could be prohibited without making the license non-free.

Given enough money anyone can still have a book printed. However, the controllers of the DRM system might refuse to sign or otherwise bless modified works thereby de facto limiting downstream freedom to replace the work with a modified version. Having to go to central authority to exercise freedoms and not being able to apply DRM on one’s own would de facto fail the Dissident Test downstream (and protecting freedom downstream is free), so the discussion needs to be limited to DRM that one could apply on one’s own but distribution of DRM-infected versions is merely for convenience like precompiled binaries for software. (However, just because something is allowed privately does not mean that the results should be distributable. GPL has set the precedent by allowing private derivatives that have proprietary parts.)


So would a ban on relatively useless but still legally dangerous DRM discriminate against certain people or a field of endeavor? Personally, I am inclined to think that it wouldn’t in a way that would be non-free. After all, it is free for the GPL to discriminate against people who make the choice of refusing to accept the GPL and against the field of endeavor of mixing GPL’ed code with proprietary code. At least at present, using a player that requires DRM is in practice a choice that does not correlate with e.g. one’s ethnicity, profession or religion. Also, it is a choice that does not inherently correlate with any field of endeavor, since there is no regulation forcing any field of endeavor to apply DRM to content (even though in the U.S. certain players need to support DRM). This is in contrast with software replaceability regulations in e.g. radio devices.

So yes, I am inclined to think an anti-DRM clause in a content license can be free if it only deals with distributing copies and it does not attack boolean access control but measures that try to restrict the nature of access once some access has been granted. Such clauses may even be a good idea in the sense that they may discourage the creation of DRM-only players.