October 30, 2002  ·  Lessig

Alot of us have been talking up the importance of spectrum policy, and about the extraordinary opportunity that free spectrum (or unlicensed spectrum) creates for the next great internet revolution. Sarah Lai Stirland has a great piece framing the debate. I think Michael Powell might get this. If he does the right thing, it could be the most important governmental decision affecting the internet in 40 years.

October 29, 2002  ·  Lessig

Ted Shelton ends his reply with a nice question: why should the protection for software “be different from copyright property in the first place.” That is the issue on which the only real disagreement lies (the other disagreements are illusory, caused I trust by my own lack of clarity).

Here’s what motivates my thought that copyright gives too much protection to software just now: Think about patents for drugs. There is an active generic market for drugs. Why? Because the patents expire long before the usefulness of the drug does. That means the generics get to free ride off of the invention of the drug companies. But that free riding is a Good Thing. If the patent term is long enough, it has had its effect: it has incented [not really a word but it should be] a new discovery; but after giving the inventor his due (as defined by the term of the patent), the public gets access to the invention “without restraint.” (Story, J.).

Patents for drugs thus strike a balance: the inventor gets a portion of the total value of the patent, and the public gets a portion. And because the public gets the portion equally (meaning there can be competition in the production of the drugs), prices for drugs fall significantly, and competition is enhanced.

My sense is that at 95 years, copyrights for software produce no similar balance. Sure: they produce value to the public to the extent that they induce the creation of new software. And sure: people can reverse engineer software to figure out how it ticks. But both of those “bargains” for the public are much less than the bargain that a drug patent gives the public. (A drug patent too creates a new drug (that may save lives, or reduce pain), AND AS WELL after a period of time, a useable and valuable invention becomes the public. And a drug patent too creates an invention that other people could reverse engineer, but it AS WELL includes instructions (the patent) for how to practice the invention.)

So too is the “bargain” of a software copyright less than the bargain of a copyright for something other than software. For when (or I guess I should say if) an ordinary copyright expires, the text it produced can be copied and actually used. That’s because texts, on their surface, carry all the code necessary to make them run within one natural language. Software is different. When the copyright for Windows 95 expires in 2090, there will be no machine that can run it. It will have zero value then. And indeed, all software written in 1995 will have zero value then, if all that is “free” then is object code. (Sure, there may be emulators that emulate computers of the last century, and they may well be able to run Windows 95, but we don’t need an emulator of the 18th century to read Ben Franklin’s essays. Everything’s a difference of degree, and some differences of degree (like this one) become differences of kind.)

Software copyright thus gives the public less than patents, and less than nonsoftware copyright. That was the motivation for my proposal: software should be “different from copyright property” because the public is not getting the same sort of bargain with software that it is getting with nonsoftware copyright, or with (some) patents.

Thus I proposed two modifications: First, shorten radically the term of protection. I said 10 years, but I’ve no magic in that number. Could be shorter. Could be longer (slightly). The point is the term should be short enough that there is something valuable at the end of the bargain that others can build upon. Second, make sure there’s a translatable text at the end of the bargain so that there’s something others actually could build upon. Thus the escrowed source code. Two changes designed to create more of a bargain in the context of software, so the public could benefit from more public domain code, and others could more easily build upon formerly copyrighted software.

Now one could well say, “but it’s my property. I deserve it for ever. No reason you should force me to give it to the public after 10, or 95 years. I created it. I get it forever.” But then you would have to say the same thing about drug patents. And you would also have to amend the constitution. For again (and this is why I insist it is just confusing to use the word “property” here):not all “property” is alike (even if all “property” gives what Ted calls “ownership”). While the constitution says that if the state takes your property, it must give you “just compensation” (Amendment 5), the same constitution says that if the state gives you (what we call) intellectual property, it MUST take it after a “limited time” without compensation. So you may well be against limited times, and the bargain with the public implicit in that. But my argument is grounded in the view that the monopolies (“exclusive Rights” is the constitutional text) the government grants in the name “promot[ing] Progress” should be balanced in the way patents are, and copyrights used to be.

Two other nits on Ted’s response.

(1) We’ve converged on a description of the kinds of protection the law might grant software under the proposal I’ve outlined: (1) the modified form of copyright (shorter terms, escrowed source), (2) if not (1), then some protection against simple mechanical copying of object code, (3) if not (1) or (2), then trade-secret. This structure says: the state gives you more protection the more you give the public after a limited time. Whatever the details, that’s got to be the form.

(2) We’re nowhere close to an agreement about the mystical difference between atoms and bits, but I don’t think it matters to the argument above at all. I will just note this: Ted makes lots of the idea of the government holding the source, and then giving it away after the term expires. But there is nothing new or special in this. Until Hollywood got a nice exemption from this, the government required, e.g., deposits of films that were copyrighted for the sole purpose of assuring that after the copyright expired, there would be a source for others to come and copy. Obviously, the cost of copying is greater, but here’s the point about economics that we are just not connecting about: so long as all competitors face the same cost, why does it matter that the cost is $1 or $100. The product will be sold at the marginal cost. This is what the competitive system is all about. Or put differently: would the complaint go away if the government charged everyone wanting a copy $100? Or put differently again: I understand there is a difference between atoms and bits: the question I am asking, and not yet seeing an answer, is why does it matter to the economics. Some inventions are easy to copy after the patent expires (the Cotton Gin); some inventions are hard (drugs). Some copyrighted work is easy to copy after the copyright expires (a novel); some copyrighted work is hard to copy (a 1930 film). But whether easy or hard, the economics is the same.

October 26, 2002  ·  Lessig

This letter by Adam Smith on behalf of the “New Democrat Network” asking Cybersecurity Czar Richard Clarke to avoid GPLd software deserves a response. Here’s a short one by me. And if you agree, then you should respond here.

The essence of Congressman Smith’s argument is that the government should not fund basic research that results in GPLd code because it “prevent[s] companies from adopting, improving commercializing and deriving profits from the software.” GPL does this by demanding that any modifications of GPLd code that are distributed must be distributed with the source and that makes it impossible for companies to establish “commercial IP rights in any subsequent code.”

The trick in this argument is to make it sound as if it is “companies” versus GPL, or the “private sector” versus RMS. But of course there is no such division. There are, no doubt, some companies that because of their business model, cannot adopt GPLd code. Microsoft is one. But there are plenty of other “companies” who have no trouble dealing with GPLd code—IBM is one. The difference is therefore not between “companies” and the GPL. The difference is between companies willing to pay the price of GPLd code and those not willing to pay the price of GPLd code. Giving up proprietary control is the price GPLd code demands, just as all-the-money-in-the-world is the price Microsoft would demand for similar access to its OS source. Some can pay that price; some cannot.

So if Smith is being principled, then properly stated, Smith’s principle comes down to this: That the government should not fund any research that results in code that some companies could not, consistent with their business model, adopt.

If that is his principle, then it follows that the government can’t fund projects that result in proprietary code (since there are some entities (say, the Free Software Foundation) that can’t, consistent with their business model, accept that code), or more radically, it means that the government can’t fund research that results in patents (since there are some business models that can’t pay the price of a patent). The only research the government could support, on this theory, is research that produces work in the public domain.

That is an interesting but radical principle. The government funds all sorts of research that results in patents, and in proprietary code. So the real question for Congressman Smith is this: Does he believe the government can’t support proprietary or patented work if he believes it can’t support GPLd work? Is he advancing a principle, or just FUD about GPL.

If this is not a principle, then why he is speaking for the New Democrat Network. I understand why the congressman from Microsoftland pushes legislation to protect Microsoft. But it is wrong to link that pork with the NDN.

October 25, 2002  ·  Lessig

I’m trying to think about other things, but when my Dad won’t stop sending me articles about the case, it gets hard. Today he sent me this piece by Professor Marci Hamilton which concludes that “Mickey has the better of the argument.” But I wonder whether Mickey would have the better argument if the real arguments that we make are considered.

Jeffrey Rosen in the New Republic is the only commentator I’ve seen (though I’ve been trying to ignore this, my father notwithstanding) who has actually framed what makes this an interestingly hard case. The question is not what the Copyright Clause means in the abstract. The question is what the Copyright Clause means, given the Court’s method for reading enumerated powers. Given the Court’s insistence on the limits on Congress’s power in the context of the Commerce Power, our argument is that the same principles should produce limits on Congress’s power in the context of the Copyright Clause. Again, Rosen frames the question perfectly well: Why would you read the express limits of the Copyright Clause (“limited times”) to produce no meaningful limit on Congress’s power, but read the implied limits in the Copyright and Commerce Clause to impose real limits on Congress’s power?

It is also odd to find in Hamilton’s piece the following statement: “indeed, there is so little history relating to the Copyright Clause that any argument from original intent is doomed from lack of information.” That claim does a great disservice to scholars such as Ray Patterson who has written extensively about what the framers thought. So too does it ignore the extensive history by Tyler Ochoa, Mark Rose, and Edward Walterscheid in their historians’ brief. Indeed, it does a disservice to Hamilton’s own work, for we had relied upon her writing in our opening brief to conclude, as she wrote, that the Copyright Clause is thus not so much “pro-author but rather anti-publisher.” At least in her writing, she seemed quite confident about what the framers meant, and, imho, with good cause.

Hamilton may in the end be right. She is an extraordinarily talented lawyer and legal scholar whose work has taught me a great deal. It may well be that though everyone seems to concede that CTEA is bad policy, the constitution must permit it. But if that’s the answer, then I confess I’ve missed something in my understanding of modern American constitutional law. Taking this Court at its word, and even praising them a bit for their commitment to constitutional fidelity (as I have), “limited times” ought to have a meaningful limit.

October 24, 2002  ·  Lessig

Aaron makes the nice point that there should be an analog to “fair use” in the context of patent law. This is actually a point explored by a number of legal scholars, most extensively by Maureen A. O’Rourke in a 2000 paper that does not seem to be online. But he grounds his claim on the “the conflict between these two have, in copyright at least, traditionally been resolved by the concept of ‘fair use.’”

Not true, or better, that’s government speak. “Fair use” is one mediating device between the First Amendment and copyright. The Court has never held, and it would make not sense to hold, it is the only mediating device. It is a central part of our First Amendment claim in Eldred that there are other mediating devices between the First Amendment and copyright — i.e., duration. And that just as the government concedes that if “fair use” were changed by Congress, or if Congress decided to copyright ideas as well as expressioin, then “undoubtedly” ordinary First Amendment analysis would apply, so too if Congress changes the duration of existing copyrights, ordinary First Amendment analysis should apply. This was the argument of the author of the notion (adopted by the Supreme Court in Harper) that there is a “definitional balance” between the First Amendment and copyright–Nimmer–as he concluded that retrospective changes of copyright violated the First Amendment and the Copyright Clause.

October 23, 2002  ·  Lessig

Aaron points to a wonderful page by John Mark Ockerbloom at the University of Pennsylvania listing books that are in the public domain elsewhere but not, because of the Copyright Term Extension Act, in the United States. Check out the books you are not allowed to download. Makes me proud to be an alum.

October 22, 2002  ·  Lessig

Dave bravely (given the excitement about Mitch’s latest contribution) defend’s Don Park‘s concern that the Open Source Applications Foundation will fuel an “erosion in the sense of value for software.” That is an important and valid concern, but it needs to be kept in context.

The more I try to understand the resistance of good-souled sorts to open source or free software, the more it reminds me of the resistance by good-souled sorts to limits on copyright in, e.g., music. Sure, there are institutions you would anti-trust that oppose free software (msft), just as there are institutions that should be anti-trusted that oppose limits on copyright in music (riaa). But the puzzle is to understand the valid and serious concerns of those whose motives one should trust — Dave Winer, e.g., in software; Jenny Toomey’s Future of Music Coalition, e.g., in music.

Park’s statement helps, but it helps by isolating a legitimate, yet distinguishable, concern.

I don’t see how anyone could on principle oppose having the source code for a program available. Dave insists it’s not necessary, but that’s a very different point. It’s also not necessary to have the formula for a certain chemical compound (because one could always reverse engineer it) (I know squat about chemistry, so that might be complete bs), but just because it is not necessary does not mean it would not help. All things being equal, having the source code helps for many purposes. (“Many,” not all; helps, not “necessary”). So opposition to having source code out there must be grounded on the view that source code also hurts, and that its harm outweighs any benefit.

So how might it hurt?

Well, one way it might hurt is by making it harder for companies to behave strategically against their competitors. That was the charge against msft: that it used its power over its code to play games that illegally harmed its competitors. That power was enabled in part by closed source code. That’s not to say one couldn’t reverse engineer the code — Ed Felten did it. But the ability to reverse engineer is different from the access to the source.

I take it no one opposing free software would offer that as a justification, and certainly, Dave-type opposition is not grounded in that reason.

That leads to a second reason to oppose open or free software — that it would destroy or change the software-writing business. This seems to be Park’s concern: If everyone expects code to be free, then the ability of certain sorts to get paid for writing code is threatened. Not all coders, but some. The people who would get paid for writing software would be the people who sold devices (e.g., Apple); the ability of independent sorts to write and sell software would in turn then be weakened.

Notice the parallel argument existed in the early days of the debate about copyright and the net. (Barlow‘s amazing article is still a powerful read on this.) In that debate, some suggested the answer was for musicians to sell t-shirts, or spend more time on tour. Understandably, that response didn’t make musicians very happy.

But I think the key in both contexts is first to isolate the point, and recognize what drives it. If there were a way to assure coders — especially independent coders — got paid even though the source of their code was open, then it would be hard to oppose open code. And while it might seem odd to imagine how that is possible, we should recognize that our economy already has about a billion ways in which it secures payment to creators without locking up the creativity. Some of those would be bad (moving music back to the patronage system, for example); but not all of these would be bad. And if we could devise a way for coders to get paid, including coders independent of companies like IBM, while allowing the source code to be free, then this legitimate concern of good-souled skeptics could be met.

Professor Terry Fisher is devising such a technique in the context of music. Pester him to publish, because it is truly brilliant. Equivalent geniuses should be crafting a similarly brilliant solution for code.

October 22, 2002  ·  Lessig

Ted Shelton continues to do web logging credit in his careful and fair reply to my reply to … well you get the picture. And more importantly, he does this space credit for his patience in waiting for a reply. I apologize for the delay, but because his last post has helped me understand and frame this a bit better, the conversation continues.

First, let’s get rid of a point on which there is absolutely no disagreement. I had said the ultimate question is what system would create the best incentives to “create and spread knowledge.” Ted adds to this that in his view, there is an important “economic objective” built into that. I absolutely agree. The very best thing that copyright does (in my view) is enable small creators (coders as well as writers) to operate independently of patrons by giving them an economic incentive to create. But for copyright, many of these creators would only create as commissioned. This is one of the great points made by Neil Netanel in a long but wonderful piece in the Yale Law Journal, and it is an important point for those of us perceived to be “anti copyright” to repeat again and again.

I, in fact, would go further than Ted about this. It’s not just, as he says, paraphrasing Churchill, that capitalism is “the worst form of incentive, except for all the rest.” I would say it is positively great — so long as balanced by other sources of creativity, and balanced in the traditional ways. So here, at least, unless Ted wants to go pinko on me, we have complete agreement.

Ted’s second point does require me to state more clearly the kind of regime I imagine. I had said we shouldn’t give “copyright protection” to code unless a copy of the source code was deposited with the copyright office, and that after the term expired, that source code would then be made available to the world. If the coder didn’t like this trade (protection in exchange for the creation after the term expired), then I said the coder could rely upon trade secret law to protect his code. That, I said, was close to the offer patent law makes: In exchange for a limited term monopoly (20 years) the inventor must describe his invention clearly enough so that others could copy it. The difference is that my scheme hides the description for 10 years, while the patent system makes it available 18 months after the patent is applied for. (Thanks to David Makman for correcting my careless mistake).

Ted replies that “without the legal status of property,” however, this is no real option, because the code protected by trade secret law could still be “endlessly copied … with no need for access to the” source code. And hence, even if the source code were protected, there would be no protection for the real value of the program.

Agreed. But we could hammer out a real agreement if we could give up this word “property” for a moment. All “property” is simply a bundle of rights associated with a particular rights holder and a particular object, whether virtual or real. These rights differ radically across “properties,” and it is therefore sometimes misleading to speak of them as if they are the same. Thus, e.g., a patent, like a toy pony, is a form of “property” in the sense that with both, the owner gets to decide under which terms to alienate the property, and if someone invades that property right without the owner’s permission, there is hell to pay. But a patent is obviously a very different kind of “property” from a toy pony because after 20 years, the property is automatically the public’s, but after 20 years, a toy pony is still private.

So, without using the often-misleading-word “property,” here’s how we might restate where I believe there should be agreement. I would still not give “copyright” protection except upon the deposit of the source code. But note, “copyright” protection protects against much more than simple copying, and it is that larger bundle of rights that I would deny a coder unless the source code were deposited.

I would allow the coder the protection of trade secret law, and it may well make sense to supplement that protection with a protection against mechanical reproduction of the digital object — such that the effective protection was against simply duplicating the code and distributing it free of any restriction. That would give the coder a real choice (between copyright and trade secret law) but also give the public something more for the more extensive set of rights that copyright law provides.

This twist to what I said before may mean we should balance the rights a bit differently still. This “mechanical duplication” right need not run long, and should not run as long as the copyright. But these are details for experts with slide rulers, not me.

Ted’s third point I’m not sure we can yet agree upon. He insists there’s a difference in kind between the state giving away the source code for a product, and the state giving away the knowledge necessary to make the product. The second is what happens when a patent expires; Ted insists that’s different from what would happen if the state also gave away the source code.

I still disagree. Though I’ve made a career insisting that the digital is importantly different, I don’t think it’s different here.

Ted nicely illustrates his point with an analogy to the cotton gin. He says, “all of the algorithms needed to create a spreadsheet would be the equivalent of the information needed to build a cotton gin. But the implementation, the source code, IS the cotton gin itself. Requiring the software developer to give away the source code is the equivalent of the government giving away a complete working cotton gin to everyone who wants one once Whitney’s patent expires.”

I guess I don’t see what is supposed to follow from this analogy.

One follow-on might be “and this would be bad because it would destroy the ability of the coder to continue to sell the product at all.” If that’s the point, then the same point could be made about many patents. And anyway, what’s the harm from that? The inventor got his monopoly; the public now gets a free resource which before it had to pay for.

But in any case, I don’t believe this would follow for most code anyway. Sure, giving away the source code enables anyone to simply release the product. But that in turn creates an incentive for everyone to invent a better implementation of the product. And as each “derivative” work would be entitled to its own protection, the incentive to improve would be rewarded as well.

So what, Ted, am I missing here? And are we agreed on points 1 and 2?

October 21, 2002  ·  Lessig

I’ve just finished replying to the more than 500 emails I received about the Eldred case. Despite my con-law musings, it is extraordinarily hard to read these emails without a profound sense of just what’s at stake. Missive after missive describes the burden of the existing system, and how 20 more years of control will stifle creative work. These half-a-thousand emails should be worth one word at least — REVERSED.

In a bunch of these emails, however, there were two recurring questions that I wanted to clarify.


Following a comment to my account, a bunch of people asked why it was so clear that the 1790 statute did not support the government’s case. I had said that of all the questions in this case, this one I was “100% certain” of. If the 1790 Act “extended” copyrights, why isn’t it precedent for CTEA?

The reason is this: CTEA extended copyright terms in exchange for nothing from the companies and authors who benefit from the extension. It is an extension without a “quid-pro-quo” — the authors giving something in exchange for something they get. As we argued, and the government simply ignored, in 1790, many would have believed that they had a common law copyright (perpetual) as well as any state statutory copyright. The effect of giving them the benefit of the federal right was to terminate this state right. IN EXCHANGE FOR the federal right, they gave up a state right. The 1790 Act thus fits the “quid pro quo” requirement in a way the CTEA does not.

But, as one person emailed, the SG said that the Supreme Court held in 1834 that there was no common law copyright. True enough. But that doesn’t resolve at all what the people who passed the 1790 Act thought. We are not arguing that they, in 1790, necessarily believed there was a common law copyright. Just that they thought it uncertain, and believed they were terminating any rights that existed then in exchange for the federal right.

But, as another careful reader asked, what about the Patent Act (from 1793) that the government pointed to? Didn’t that show that the Copyright Act of 1790 did not require a “quid pro quo”? For that statute expressly required the surrender of a patent to get the benefit of the federal act. True enough again. But there is zero authority for the idea that a state right survived after the act of 1790. The same case in 1834 holds as much. The statutes may well be different but the effect was identical: both terminated state rights and gave a federal right in exchange.

1st Amendment claims?

A bunch asked whether the First Amendment claim is dead, or whether we had given it up, or had it taken away. The answer is clearly “No.” Again, there are certain questions we wanted to focus on at the oral argument; we would have been happy never to even mention the First Amendment. That’s not because that claim is not alive and crucial to our case, but instead because it is, or should be, clear enough when they think through the mechanics of the case. If we prevail on the copyright clause claim, then there is no need for them to reach the First Amendment claim. If we don’t prevail on the copyright claim, then they must reach the First Amendment. But when they get to that step in their thinking, it will be extremely hard for them to avoid applying ordinary First Amendment analysis to this statute. (Justice O’Connor said she thought it was odd to think of a law passing under the copyright clause, but failing under the first amendment. But the government concedes this possibility — if the law removed, say, fair use — so the only real question is how the First Amendment should police changes in term.) If they apply “definitional balancing” which is how the Harper case is framed, then the question will be how “definitional balancing” applies to changes in copyright terms, and as the author of definitional balancing said, “definitional balancing” should invalidate retrospective changes in terms. And if they apply ordinary First Amendment intermediate scrutiny, then there is no presumptive pro-speech benefit that could outweigh the burden on speech. Either way, the First Amendment stands as an effective backstop to the copyright clause argument. But as that argument is narrow, and more easily cabined to just changes in terms, we expect, if we prevail, that’s where the Court will decide it.

Again, thanks to all who wrote in support of the case and with kind words for what has been done. Even the one defender of Mickey was polite in his strong opposition.