[om-list] Arguments against Copyleft

Mark Butler butlerm at middle.net
Fri Dec 21 13:05:33 EST 2001


I think the GPL is ideal for two kinds of products:

 1. Pervasive infrastructure products where there is a natural monopoly
 2. Products developed *primarily* by end-users, volunteers, and non-profits

However, it is virtually impossible to base a software company on software that is completely GPL.  The reason is that while customers may be willing to pay for support and minor enhancements, the income from support is not sufficient to support significant research and development.  This is especially true in narrow, low volume markets.

Because of the R&D costs involved, the leading edge software in narrow markets will likely never be completely open source. The best we can hope for is that companies can cooperate in shared libraries where the LGPL and BSD style licenses are the only real options.

The disadvantages to the LGPL are primarily technical in nature. The extra restrictions are designed to prevent proprietary forks, but they create quite a bit of extra work - for example, the libraries *must* by dynamically linked.

The argument depends on the empirical observation that the returns gained in terms of extra participation by allowing the extra freedom of a BSD style license are greater than the losses due to proprietary forks.

Apache and the incredible constellation of surrounding projects is proof positive that open source products can thrive under BSD style licenses.  Companies produce and market proprietary enhanced versions of Apache all the time, but the momentum of the open source line of development is sufficient to encourage all those companies to use and contribute to the main open source line of development.

 - Mark




More information about the om-list mailing list