On 21/07/13 13:02, Ben Finney wrote:
Adam Bolte <abolte@systemsaviour.com> writes:
On Thu, Jul 18, 2013 at 12:23:08PM +1000, Ben Finney wrote:
It depends on what compromises the trademark owner is able/prepared to make. My gut feeling is that if the trademarks do not permit modification and redistribution in such a way that there is no longer any clear association with the original trademark, they belong in non-free - if anywhere.
Right. The conflict, of course, is that this completely undermines the purpose of trademark. Specifically, the purpose of preventing uses of the mark that would mislead consumers about the provenance of a product.
That purpose of trademark is, in my view, of benefit to society. Yet full software freedom of the recipient is *also* of benefit to society.
I think that trademark has a significant benefit to society, which is to limit the tendencies of vendors to misrepresent their modified works as though being whatever the customer is looking for — even if that vendor has made incompatible or undesirable changes which are contradictory to what the customer would expect from the brand.
Some copyright licenses attempt to clumsily use copyright law to do this, e.g. the 3-clause BSD license has as a condition that no-one may use the name of the copyright holder to “endorse or promote” the redistributed work.
Other copyright licenses have explicit permission to combine the copyright license's terms with trademark terms that restrict the use of marks, e.g. the GPLv3.
The usefulness of trademarks is pointed out here, and I did not see this clearly earlier. The problem with trademarks is that it assumes that I would trust the application brand more than the people distributing the software. I would always put more trust in my distribution than any application, and if I didn't I would get the build directly from the application's official website, or grab the distribution source package and inspect the list of patches. Seems that's probably just me. :) So to summarise the benefits of trademark for a second, they might be as follows. To the company: * increased brand recognition * some clear association of a product with a company For the end user: * recognisable name (easier to discover) * brand that the user trusts All the concern seems to be on that last point - the company with the trademark wants to ensure that a quality software build is associated with the brand, but not necessarily a bad build with unsupported patches, etc. This goes against free software, hence the problem. What if there were a way in GNU/Linux distributions to easily identify unofficial builds of trademarked software to the user? Maybe have an included system that simply prompts the user to accept execution of unofficial builds on first program execution, and puts a symbol next to the application launcher to remind the user as such. Regardless of how it would be implemented, if there were some standard for trademarks in free software that required unofficial builds to set a flag that would somehow make it obvious to the user, would that solve all concerns, and enable the trademark holders to relax their restrictions for distributions which enable it? The Mozilla Corporation could then just dictate that unofficial builds that wish to use our trademarks must make it clear that the build is unofficial to the end user by doing such-and-such (which the aforementioned program would take care of automatically). I think that would probably satisfy the DFSG too.
But those either ignore or punt the issue to trademark. The question still remains: what restrictions on the freedom of any recipient are acceptable in exchange for preventing the societal harms trademark law is designed to address?
I don't know what the maximum acceptable restrictions would be, but I agree that it would be great to have some guidelines to clearly define it.