How To Win (DMCA) Exemptions And Influence Policy

by Seth Finkelstein (EFF 2001 Pioneer Award winner for censorware decryption) [updated to 2005]

But we're in a world where disobedience is treated with felony convictions. The idea that you are going to get lots of civil disobedience against the Digital Millennium Copyright Act is just crazy. You're going to get lots of prosecutions and people going away to jail.

-- Lawrence Lessig, Reason interview

What is this?

The Digital Millennium Copyright Act (DMCA) is extensively known now, being perhaps the most hated Internet law since the Communications Decency Act censorship legislation. The DMCA's chilling effects have been widely publicized, and seem only to increase. One of the few parts of the DMCA which even attempts to moderate its far-reaching restrictions, is a provision which directs the Copyright Office of the Library of Congress to make an inquiry regarding adverse effects of one of the DMCA's prohibition on circumvention. If adverse effects are established, this office then has the power to "exempt certain classes of works from the prohibition against circumvention of technological measures that control access to copyrighted works."

These Library of Congress DMCA proceeding are to be done every three years. The last round took place in 2003. The upcoming round is the next one, for results in 2006. The schedule is that "Written comments are due by December 1, 2005", and in response to those comments, "Reply comments are due by February 2, 2006." These comments are intended to be evaluated in hearings in the spring of 2006.

The key documents involved are as follows:

This is a chance for an ordinary person to influence the process of policy-making, and perhaps convince the Library of Congress to grant an exemption. The author of this article (Seth Finkelstein) was the commenter primarily responsible in 2003 for persuading the Library of Congress to grant one of the four exemptions determined, allowing circumvention in order to access censorware blacklists.

Why do it?

There's a common view of technical people, that participation in government falls somewhere between functionally useless and morally perverse. It's easy to deride the clueless congresscreatures, and take comfort in an idea of building an uncensorable realm located in some other dimension. And it's not encouraging to be one letter in a pile of mail to a Senator, or a few dollars versus the economic clout of copyright industries. But not all aspects of influencing government are so one-sided and unbalanced.

To begin with, making a submission in this DMCA rulemaking process is comparatively easy. Whatever gains are made, are won at a trivially small cost. This is a matter of drafting a letter, with some thought and detail. It's not a multi-million dollar lobbying campaign. It's not a lawsuit which drains someone's life. It's not anyone going to jail. Rather, it's essentially drafting a letter, the same commitment as happens every single day on so many mailing-lists and weblogs.

But unlike letters to congressional offices, these public comments are truly read by the people who make the policy. That is in fact one of the most astonishing aspects visible in the text of the earlier rulemaking results. The policy-maker may not have agreed with the arguments, may in fact have dismissed them; but there's enough referencing and mention of the reasons for the results so as to make it clear that the viewpoints of the public were heard. And this consideration doesn't even require a large bribe, I mean, campaign contribution. Not all parts of the government are equally inaccessible. It turns out these policy-making determinations are surprisingly amenable to informed citizens making a difference.

What is in play?

Only one part of the DMCA is affected here (quoting the 2005 notice of inquiry):

The statutory focus of this rulemaking is limited to one subsection of section 1201: the prohibition on the conduct of circumvention of technological measures that control access to copyrighted works. ... The Librarian has no authority to limit either of the anti-trafficking provisions contained in subsections 1201(a)(2) or 1201(b).

Roughly, the only DMCA aspect which can be softened is the prohibition on accessing material. Nothing about making tools can be changed. That is, this rulemaking CANNOT make distributing DeCSS code, or eBook access programs, into an exempt act. It's clearly stated that the only effects which may be considered, relate to the access circumvention provision. This rulemaking is about doing, not coding or distributing. Those other provisions are a battle for another time.

Nonetheless, even though coding and distributing are not in play here, it's worthwhile to have as strong a record in favor of exemptions as is possible. This is for the simple reason that it may help when those next battles are fought.

What is to be done?

In order to get an exemption, there must be a "showing that the prohibition has a substantial adverse effect on noninfringing uses of a particular class of works.". Proponents have the burden of proof. The specific details are (emphasis added):

In order to meet the burden of proof, proponents of an exemption must provide evidence either that actual harm exists or that it is “likely” to occur in the ensuing 3-year period. Actual instances of verifiable problems occurring in the marketplace are generally necessary in order to prove actual harm. The most compelling cases of actual harm will be based on first-hand knowledge of such problems. Circumstantial evidence may also support a claim of present or likely harm, but such evidence must also reasonably demonstrate that a measure protecting access was the cause of the harm and that the adversely affected use was, in fact, noninfringing. “Likely” adverse effects may also support an exemption. This standard of “likelihood” requires proof that adverse effects are more likely than not to occur. Claims based on “likely” adverse effects cannot be supported by speculation alone. The House Manager's Report stated that an exemption based on “likely” future adverse impacts during the applicable period should only be made “in extraordinary circumstances in which the evidence of likelihood is highly specific, strong and persuasive.”
Nevertheless, as the Register's final recommendation of 2000 explained, the expectation of “distinct, verifiable and measurable impacts” in the legislative history as to actual harm suggests that conjecture alone would be insufficient to support a finding of “likely” adverse effect. ....

Although a showing of “likely” adverse impact will necessarily involve prediction, the burden of proving that the expected adverse effect is more likely than other possible outcomes rests firmly on the proponent of the exemption.

The identification of existing or likely problems is not, however, the end of the analysis. In order for an exemption of a particular class of works to be warranted, a proponent must show that such problems warrant an exemption in light of all of the relevant facts. The identification of isolated or anecdotal problems will be generally insufficient to warrant an exemption of a class of works. Similarly, the mere fact that the digital format would be more convenient to use for noninfringing purposes is generally insufficient factual support for an exemption. Further, purely theoretical critiques of Section 1201 will never satisfy the requisite showing. ... Proponents of exemptions must show sufficient harm to warrant an exemption

Translation: No ranting from any Libertarian Book-Of-Rand. No repetitions of "I own it and I can do whatever I want with it". There is no interest in hearing an argument about whether a license is a valid contract, or the abstract meaning of freedom and property and fair-use rights. That won't work, and will likely just waste the writer's energy and the reader's time.

Further, note the emphasis on using a "class of works". Again, this rulemaking is not about theoretical considerations. One must identify a concrete category of material. In the last round, some proposals were rejected as not being a "class of works". This idea is discussed extensively in terms of requirements, such as (emphasis added):

The Register found that the statutory language required that the Librarian identify “classes of works” based upon attributes of the works themselves, and not by reference to some external criteria such as the intended use or users of the works. The phrase “class of works” connotes that the shared, common attributes of the “class” relate to the nature of authorship in the “works.” Thus, a “class of works” was intended to be a “narrow and focused subset of the broad categories of works of authorship ... The starting point for a proposed exemption of a particular class of works must be the section 102 categories of authorship: literary works; musical works; dramatic works; pantomimes and choreographic works; pictorial, graphic and sculptural works; motion pictures and other audiovisual works; sound recordings; and architectural works.

While the word "class" isn't used in a computer-science sense here, just as a way to think about this for technical people, the portion discussing "shared, common attributes" could almost be taken straight from object-oriented programming definitions. In brief, the above passage might be translated as that the exemption shouldn't focus on what businesspeople call the "use case", but rather what computer-scientists call the "design pattern". That is, while the intended use is important as the basis for the exemption, the exemption itself has to be constructed in terms of a class (of works).

For example, consider censorware blacklists. The "use case" is research, investigation, and so, regarding what censorware in fact has on the blacklists. But the "class of works" is "compilations consisting of lists of websites blocked by filtering software applications".

So don't talk about fair-use as a principle in itself. Rather, focus on practical problems affecting a specific "class of works", as in perhaps "public domain works released on CSS-protected DVD disks".

This is not a situation where quantity (whether votes or money) is the key aspect. Rather, it's a case where a detailed, well-constructed, presentation can have an effect. And this is why an ordinary person can make a difference here. Better, if done properly, the requirements can even play into a technical person's strengths in formulating an argument which needs to meet certain specifications. It's just critical to keep in mind that this concerns empirical effects, not ideological axioms.

In fact, when read properly, the documentation for the rulemaking is in effect a design chart:

In addition to the necessary showing discussed above, in order to make a prima facie case for a proposed exemption, at least three critical points should be established.

First, a proponent must attempt to identify the specific technological measure that is the causal source of the alleged problem, and show why that technological measure “effectively controls access to a [copyrighted] work.”

Second, a proponent must specifically explain what noninfringing activity the prohibition is adversely affecting.

Third, a proponent must establish that the prevented activity is, in fact, a noninfringing use under current law.

The important aspect here is to organize the plea for an exemption into what lawyers call a "fact-pattern":

  1. Problem created by technological measure
  2. Activity needing to be done
  3. Why the activity is legal

For example, in the specific instance of censorware ("filtering software"):

  1. The problem is that the censorware blacklists are encrypted
  2. This prevents in-depth examination of the blacklist contents
  3. Such examination, and publishing small excerpts, is a copyright fair-use of the blacklist

After this organization, there's unfortunately another "specification" (emphasis added):

The nature of the Librarian's inquiry is further delineated by the statutory areas to be examined:

(i) the availability for use of copyrighted works;

(ii) the availability for use of works for nonprofit archival, preservation, and educational purposes;

(iii) the impact that the prohibition on the circumvention of technological measures applied to copyrighted works has on criticism, comment, news reporting, teaching, scholarship, or research;

(iv) the effect of circumvention of technological measures on the market for or value of copyrighted works; and

(v) such other factors as the Librarian considers appropriate.


These statutory considerations require examination and careful balancing. The harm identified by a proponent of an exemption must be balanced against the harm that would result from an exemption. In certain circumstances, an exemption could have a greater adverse effect on the public than would the adverse effects on noninfringing uses identified. The ultimate determination of the Librarian must take all of these factors into consideration. Therefore, a commenter's analysis should also address these considerations.

All of this boils down to a requirement to argue that the exemption won't do more harm than good. That is, it's not enough just to find a fair-use application somewhere, anywhere. It's necessary to argue that the exemption won't turn into a figleaf to justify wholesale copyright infringement.

Again, for censorware, the case might proceed in outline that

The more these kinds of points are substantiated, the better. What's being done here is much closer to an analytic examination of various "test cases", than political theorizing.

How to do it

The Librarian wants comments organized in a certain way:

Comments submitted in the initial comment period should be confined to proposals for exempted classes. They should specifically identify particular classes of works adversely affected by the prohibition and provide evidentiary support for the need for the proposed exemptions.

For each particular class of works that a commenter proposes for exemption, the commenter should first identify that class, followed by a summary of the argument in favor of exempting that proposed class. The commenter should then specify the facts and evidence providing a basis for this exemption. Finally, the commenter should state any legal arguments in support of the exemption. This format of class/summary/ facts/argument should be sequentially followed for each class of work proposed as necessary.

Note programmers see this all the time. It's the basic boilerplate in terms of standards for organizing parts of project. That is, roughly:

And if you have more than one class, make sure you follow best-practices and organize all class-specific material locally within its class, instead of scattering portions globally anywhere. That's quite straightforward.

Submissions can be done very conveniently, using the Copyright Office Comment submission form. It's worth looking over this form before starting, since the arrangement of the form was itself practically a summary of the requirements. Besides submitter's information, there are text boxes for:

The file attachment is described as follows:

Preferred format: text saved as PDF (.pdf). Other permissible formats: Microsoft Word Version 2003 or earlier (.doc), WordPerfect 9 or earlier (.wpd), Rich Text (.rtf), or ASCII text (.txt).

This submission box has a nice little link to a small requirements summary, which is worth reading. The underlying message - no theoretical ranting; they mean it.


Some background information on the details of the DMCA can be found at the Chilling Effects website, particularly the anticircumvention section. And see also the EFF's DMCA archive. It's important to thoroughly read the Copyright Office 2005 notice of inquiry. This is all the typically unpleasant but often necessary task of reading the documentation.


The reason to become involved in this process is very simple. The lawsuit you prevent may be your own.