Is 10 years really that bad?

I've been following the interesting blogversation between Doc Searls, Dave Winer,  Larry Lessig, and Charles Cooper surrounding Larry's proposal to reduce the copyright of software to 10 years, and to hold source code in escrow so that it can be released after the 10 year copyright is over.

One point kept niggling at me, which was the proposition that Charles and Dave made that a 10 year copyright will seriously hurt small and independent software developers.  Charles writes:

I can't think of a better prescription for formalizing the existing constellation of power that favors the Microsofts and Oracles over the small and independent developers.

At this juncture in the history of the software industry, more so than ever before, 10 years doesn't amount to a hill of JavaBeans--not when you're attempting to build up brand, distribution and customer loyalty in an increasingly fragmented and competitive market.
Dave Winer puts it this way:
After giving it a bunch of thought, I think Lessig is going after the BigCo's, probably Microsoft. But he would also sacrifice the independent companies. If we have to publish our source code the users won't pay for it. Ten years isn't enough time to create a new market. So you wouldn't get any commercial innovation in this system. The BigCo's don't innovate.
Huh?  I don't get it.  I'm a software developer - I make my living through the creation of software, and all my experience and training tells me that these arguments just don't add up.  As a thought experiment, ask yourself how many pieces of software are you using that are 10 years old?  And when did you purchase that software?  I can't think of a single example of software that generated revenue 10 years after it was written, unless you're talking about software for the Space Shuttle or some other old piece of hardware; and even then, that software will have bug fixes and new features, more likely than not - and that code will be covered by the 10 year copyright law as well. 

Now ask yourself, how many old pieces of software would I love to have on my current desktop?  Hypercard, perhaps?  Or how about all those old games for the Atari or Intellivision consoles?  How many programmers would be able to learn and innovate on today's programming world if the source for those old non-revenue generating programs were part of the public domain?  How many old data files would you be able to unlock if you were able to run that old DOS accounting program on your current desktop?  What will happen 30 years from now when the MS Word file format is as anachronistic as a pile of punch cards?

Larry Lessig has written a response to Cooper's criticism, and I think he's dead on.   Lessig quotes an 1829 precedent that shows that the founding fathers felt that copyright terms had to be "limited" so that creative works would pass into the public domain "at as early a period as possible":
Why? Because the framers were keen to have others build upon creative work, after copyright assured the "author" a sufficient return for his or her creativity. They believed, in other words, in a public domain, and they required that copyright terms be "limited" so that the public domain would flourish.
Lessig then builds a strong argument to show how his proposal would actually help small, creative, individual developers, and help reduce barriers to entry into large, established markets while still giving the copyright holder a 10 year head start on new competitors.

Dave Winer makes another argument, which I'll call the "supply vs. demand" argument:
If the customers placed a sufficiently high value on having access to source code, or if they felt our copyrights lasted too long, of course we would have to do what they want us to, or retire from the market. So the proponents of this plan are trying to legislate what they haven't been able to gain in the market. It's a weak position for that reason.
I disagree. There's never really been true power or representation by customers in the market for source code and long-term support contracts - it is just too small, fragmented, and uninformed, especially in the desktop computer space.  Of course, big companies who make custom deals can get license terms changed, including access to source code, but we're not talking about a true level playing field in today's world.  Heck, we've had companies declared monopolies that still browbeat their customers with obnoxious clickwrap licenses and liability warranties that would be considered obscene if they were attached to a physical object like a car or a bridge. These actions are not signs of a healthy relationship between software developers and their customers.  Software, like privacy, has always been a supply-side game - the only recourse a customer has is to not play, and in many cases, that is impossible or untenable.

In short, reducing software copyright to a reasonable term would have a very positive effect for the software market and for its customers.  There would be an incentive to innovate, in order to keep opening up new markets and provide better and better code.  And future generations of computer scientists (and historians) would thank us - with this code in the public domain, we could join the other sciences and stand on the shoulders of giants, not stand on each other's feet.