Funding Art vs Funding Software
I've proposed two new systems to fund artists in a world where we have legalized sharing (noncommercial redistribution of exact copies) of published works. One is for the state to collect taxes for the purpose, and divide the money among artists in proportion to the cube root of the popularity of each one (as measured by surveying samples of the population). The other is for each player to have a “donate” button to anonymously send a small sum (perhaps 50 cents, in the US) to the artists who made the last work played. These funds would go to artists, not to their publishers.
People often wonder why I don't propose these methods for free software. There's a reason for that: it is hard to adapt them to works that are free.
In my view, works designed to be used to do practical jobs must be free. The people who use them deserve to have control over the jobs they do, which requires control over the works they use to do them, which requires the four freedoms (see http://www.gnu.org/philosophy/free-sw.html). Works to do practical jobs include educational resources, reference works, recipes, text fonts and, of course, software; these works must be free.
That argument does not apply to works of opinion (such as this one) or art, because they are not designed for the users to do practical jobs with. Thus, I don't believe those works must be free. We must legalize sharing them, and using pieces in remix to make totally different new works, but that doesn't include in publishing modified versions of them. It follows that, for these works, we can tell who the authors are. Each published work can specify who its authors are, and changing that information can be illegal.
That crucial point enables my proposed funding systems to work. It means that if you play a song and push the “donate” button, the system can be sure who should get your donation. Likewise, if you participate in the survey that calculates popularities, the system will know who to credit with a little more popularity because you listened to that song or made a copy of it.
When one song is made by multiple artists (for instance, several musicians and a songwriter), that doesn't happen by accident. They know they are working together, and they can decide in advance how to divide up the popularity that song later develops—or use the standard default rules for this division. This case creates no problem for those two funding proposals because the work, once made, is not changed by others.
However, in a field of free works, one large work can have hundreds, even thousands of authors. There can be various versions with different, overlapping sets of authors. Moreover, the contributions of those authors will differ in kind as well as in magnitude. This makes it impossible to divide the work's popularity among the contributors in a way that can be justified as correct. It's not just hard work; it's not merely complex. The problem raises philosophical questions that have no good answers.
Consider, for example, the free program GNU Emacs. Our records of contributions to the code of GNU Emacs are incomplete in the period before we started using version control—before that we have only the change logs. But let's imagine we still had every version and could determine precisely what code contribution is due to each of the hundreds of contributors. We'd still be stuck.
If we wanted to give credit in proportion to lines of code (or should it be characters?), then it would be straightforward, once we decide how to handle a line that was written by A and then changed by B. But that assumes each line as important as every other line. I am sure that is wrong—some pieces of the code do more important jobs and others less; some code is harder to write and other code is easier. But I see no way to quantify these distinctions, and the developers could argue about them forever. I might deserve some additional credit for having initially written the program, and certain others might deserve additional credit for having initially written certain later important additions, but I see no objective way to decide how much. I can't propose a justifiable rule for dividing up the popularity credit of a program like GNU Emacs.
As for asking all the contributors to negotiate an agreement, we can't even try. There have been hundreds of contributors, and we could not find them all today. They contributed across a span of 26 years, and never at any time did all those people decide to work together.
We might not even know the names of all the authors. If some code was donated by companies, we did not need to ask which persons wrote that code.
Then what about the forked or modified variants of GNU Emacs? Each one is an additional case, equally complex but different. How much of the credit for such a variant should go to those who worked on that variant, and how much to the original authors of the code they got from other GNU Emacs versions, other programs, and so on?
The conclusion is that there is no way we could come up with a division of the credit for GNU Emacs and justify it as anything but arbitrary. But Emacs is not a special case; it is a typical example. The same problems would arise for many important free programs, and other free works such as Wikipedia pages.
These problems are the reasons I don't propose using those two funding systems in fields such as software, encyclopedias or education, where all works ought to be free.
What makes sense for these areas is to ask people to donate to projects for the work they propose to do. That system is simple.
The Free Software Foundation asks for donations in two ways. We ask for general donations to support the foundation's work, and we invite targeted donations for certain specific projects. Other free software organizations do this too.