Is Perfectly Competitive Innovation Compatible with the Free Software Ethic?

In a truly remarkable contribution, Boldrin and Levine demonstrate that under certain conditions "first-copy costs" of innovation can be funded by revenues from sales of copies without resort to government franchise such as patent and copyright. Advocates of "free software", software licensed under terms which do not exploit monopoly rights of copyright or patent, have hailed this result as vindication of their position. However, on closer look, the Free Software ethic is based on assumptions about the production and distribution of software that challenge the applicability of the Boldrin-Levine theory to software. In particular, (1) software in use by definition must be copied from the distribution medium to a hardware display, leaving the distribution medium free for further copying (the technical analysis of this point was anticipated by Quah, and is omitted); (2) the free software ethic presumes that users of software "enjoy" sharing with others, causing a "lemons" market-like breakdown as low-value users share without extracting revenue; (3) similarly, developers enjoy developing and sharing software, so that the market for the early versions is poisoned by anticipated rapid introduction of cheap improvements; and (4) an example is presented where the non-convexity "binds" so that competitive innovation is insufficient, while a "small" amount of IP protection induces the innovation. We conclude that the Boldrin-Levine theory basically makes it plausible that the optimal amount of IP protection is much smaller than traditionally believed, but does not in general make it useless