Thursday, October 31, 2002

The real debate

I am beginning to feel a little guilty about the time that Lawrence Lessig is spending with me in conversation about his concept of "opaque creativity." But I continue to believe that this is an important issue, and that Lessig is a person in a position to have a significant influence over the way in which the marketplace thinks about this idea. So I was gratified by the continued seriousness with which Lessig has addressed my concerns in his latest reply.

Lessig returns to the fundamental issue, why should software (or really any kind of opaque creativity) be treated differently from other types of creativity? Why shouldn't all types of creativity receive the same copyright protection under the law? The argument has two parts:

95 Years
Lessig writes that at 95 years, copyrights for software do not provide a balance for the public good -- in 95 years software will have no value, in fact the software product will probably not be able to run on any available hardware. If I can summarize the argument, 95 years might as well be forever. This is the same argument which I agree with that Lessig has been making about copyright generally. Copyright should be for a "limited term." This is true whether the copyright is for a book or a film or a software program. I think there is an interesting argument in saying that different types of creative output should have different lengths of time of protection, because of the inherent "shelf life" of the material. But this is a very different argument than that of saying that software should be escrowed and source code should be released. One is an economic argument about the value that the public enjoys in exchange for copyright protection. The other is an argument about what the public should be allowed to do with the property once it is in the public domain.

Opacity vs. Transparency
This then is Lessig's primary argument. Works of opaque creativity are fundamentally different from works of transparent creativity. With a book, I have all of the "code" before me, transparently. I can reproduce all or part of a manuscript and I can embed any part in a new work. With software, or other works of "opaque" creativity, it is very difficult for me to do anything other than reproduce the product in its entirety. And if the manufacturer has included security elements which require a licensing key to activate the software, reproduction of the product may not provide me with the use of the product. This is a reasonable objection, but the solution need not be Lessig's proposed source code escrow scheme. Consider the three most common uses of copyrighted material that has passed into the public domain -- duplication, derivation, and partial use.

In the case of duplication, software can clearly be duplicated in its entirety whether or not the duplicator has access to the source code. An issue remains in that mere duplication of a software program may not be sufficient to allow for use of the software. Thus some system should be in place to provide the public with access to a version of the software that does not require a security key to unlock functionality, or some ability to generate such keys once the software has passed into the public domain. This is not unlike the problem that Lessig mentions regarding the film industry -- how does one make sure that some copy of a film is available to be duplicated at the point when that film passes into the public domain. I would suggest that this problem should be solved for all categories of copyrighted material, not just such works of "opaque creativity."

In the case of derivation, the concern is more complex. I submit that the process of creating a derivative work from what Lessig calls "opaque creativity" is no different from that in creating a derivative work from a book or other example of "transparent creativity." Furthermore, these derivative works are being created all the time, and software companies have been unable to protect themselves from this type of derivation (see the supreme court case Lotus v. Borland).

Take for example the works of Mario Puzo, since I recently read that Random House is seeking an author to write a sequel to The Godfather. What would this new author have to go on to create this derivative work? The author could read the previous novels, learn about the characters and their motivations, get a feel for the dialog, understand the plot... but the author wouldn't have Puzo. Thus the author would have to "reverse-engineer" the thinking process needed to construct a believable set of characters, plot, dialog, etc.

The authors of Lindows did the same thing -- they looked at how other operating systems displayed icons, user menus, startup and shutdown scripts... and they created a new user environment for the Linux operating system. Would having access to the Windows or Macintosh source code have made their job easier? Yes, just as having access to Puzo as a collaborator on a new novel would make it easier for the new author to create a sequel.

There is a "means of production" for every creative work. In the case of The Godfather it was the mind of Mario Puzo. In the case of many modern "romance" novels, it is an automated book generating computer program. In the case of Toy Story it is a set of algorithms for rendering 3D graphics. In the case of a software program it is the source code. Software should be treated the same as any other creative work -- the resulting work should be made public when the copyright term expires. But the government should not create a system which compels artists, writers, movie producers, and software programmers to make public their means of production.

Partial Use
It is this last category of use that is certainly the most difficult to answer. For a book this might mean reproducing a single chapter. For a film it might mean showing a scene. For a software program it might mean using one function. In the case of the book or the film, I might have trouble getting a hold of the source material, and there may be some technical skill required in copying a film scene, for example. But once I have the source material it is possible to make this partial use of a creative work. With software, this is not a straightforward proposition. I take this then to be the real center of Lessig's case that software escrow is necessary to protect the public good.

The interesting problem with software is that this is exactly where the rules of copyright and the rules of patents/trade secrets intersect. No one will be interested in duplicating trivial functions, such as the way a window is displayed. The interesting functions to reuse from one software application to another will be those that do real work. For example, if I wanted to produce a photo editing application, I would be very interested in taking filters from Adobe's Photoshop and including them in my application. But this represents the core know-how of the Photoshop team, the proprietary algorithms that are the magic that makes Adobe a leader in digital photo manipulation. Furthermore, these algorithms may be the basis of not just one program, but many different programs and indeed, for some companies, an entire business.

To my mind, this is where software patents come into play. If Adobe wants to protect this core knowledge, they have two choices. One is to file a patent, describing the key algorithm. While this would give them protection from competitors for some period of time, it would also put the algorithm into the public domain once that time has expired. At that point, I can write my own software that implements Adobe's algorithm. Alternatively, Adobe can keep their algorithm secret -- BUT then they have no avenue for complaint if I am able to figure out how they achieve a given effect and duplicate this filter in my competing software.

So to summarize this argument -- there are two kinds of features that one could extract from a software program. The first is so trivial that there is no economic value in creating a system which provides for its use. The second is so crucial to a software company's business -- in a sense it IS their business -- that it should remain in the realm of patents and trade secrets and not be subjected to disclosure simply to receive copyright protection for a given specific product.

A few nits...
To respond to Lessig's two final points in his reply:

(1) I don't agree that "we've converged" -- only that I understand a little more clearly what is proposed. I agree that the state should provide protection commensurate with the economic value that is derived by the public after the limited time of such protection. I believe that I addressed this issue above under the heading "95 years" and I believe that we are both in agreement that copyright terms for all works should be shorter. I continue to believe that software should not be treated differently from other creative works.

(2) I am happy to agree to disagree on the (not mystical) difference between atoms and bits as I don't believe it is fundamental to the issue of real importance. The cost of digital duplication rapidly approaches zero, and I would argue is zero from an economic perspective. How much did it "cost" to copy this document to your computer? However the cost of physical duplication of any object will never approach zero. This creates a key economic difference.

Non Nits
I entirely agree that the US Congress has enacted copyright laws which are not in the best interests of our Republic but are instead designed to serve the financial objectives of a small number of powerful companies.

I entirely agree that the protections that our government provides to creators of intellectual property should be commensurate with the benefits that the public ultimately enjoys once those works enter the public domain.

I firmly reject the notion that software developers should be singled out amongst all creators of intellectual property and forced to divulge the means of production and the algorithms which enable them to practice their craft.

I also reject the application of this principal to all creative works -- I do not want to live in a society which compels me to register the contents of my brain with the federal government, in a "secure" escrow agreement which enables future members of the public to create, for example, articles like this one -- the rough equivalent for a writer to the demand on software programmers that their means of production and algorithms be placed in escrow.

I hope that together we will come to a clearer agreement on "opaque creativity" and why it is or is not different from other types of creativity. And more importantly, what should be advocated for the protection of each category of creativity.

No comments: