Trademarks are a legal thing - they are in effect "owned by someone". While the _code_ is open-source, that doesn't mean other people can come along and just use your trademark. Nor does it mean that a project originator should just have to "give up" the trademark to anyone who comes calling.
If you want the trademark to be owned by a neutral body, then by all means create a body to hold the trademark. That's a choice for the project originator (who usually holds the trademark). In this case there's no dispute that Oracle "owns" the trademark.
Leaving aside the legal implications of a trademark for the moment, your hypothesis is flawed at a social level. Who, after all, is "the community". Let's take an imaginary product - which was created by Fred and is now being developed by a "community" of 10 developers. There are say 200 users of the product.
Now 3 of Fred's developers decide to go and work on a fork. Does Fred give them rights to use his product name? What then does he call his product? How do customers differentiate between the two products with the same name? Since anyone can fork at any time, should we have 25 projects with the same name in the same product space?
Ok, 3 seems a little low - what about 6? What if 3 of those developers joined the project in the last month? Does their leaving count as more or less? What about if 9 developers leave?
How best then to determine which is best for the community of users? Who determines which project is "more true" to the original project? what criteria do we use to measure"trueness"? Does the language dictate trueness? Does the location of the original project lead?
In other words, your point is completely moot. Oracle gets to keep the name because they own it. period. (Although it seems they might now be prepared to give it up.) People have the right to fork the code at any time they like, but it's not in our interest to allow them to use the same name. The public is best served, and protected, by changing the name of forked projects. Anything else is just worse.
It is not that easy - that's why I added the comment - to start the discussion about open source and trademarks.
From wikipedia (sorry):
"Trademark law is designed to fulfill the public policy objective of consumer protection, by preventing the public from being misled as to the origin or quality of a product or service. By identifying the commercial source of products and services, trademarks facilitate identification of products and services which meet the expectations of consumers as to quality and other characteristics."
Trademarks exist to protect consumers, not the "owning" companies. In case of open source projects, this does not work so well and there are problems you have mentioned in your reply.
The "protection" you speak of is that consumers can associate a producer with a product. In this case the producer of the original product was Sun which is now part of Oracle. Thus someone who gets an update for "Open Office" is getting it from the one true source for Open Office.
The freedoms provided by the Open-Source license allow someone to fork that code to create a "new improved product" - but that product can't be called "Open Office". In this way consumers can't think they're getting x and end up with y.
If the consumers are concerned about Open Office, then it's their choice to go to Libre Office. They are in effect opting-in to a change. Changes of this nature have to be opt-in, there's no way around that.
Allowing multiple people to use a trademark, or worse encouraging court battles over who is the "one true heir" to a trademark would not be in the consumers interest.
Clearly a fork can not have any claim on a product name because if it did there exists the possibility for infinite forks, and infinite products with the same name. This in no way serves the consumer, and specifically it does not facilitate the identification of the product, not does it serve to identify quality or other characteristics.
In some countries, when trademarks are transferred, there must be a continuity in the underlying product or the trademark is lost. (You cannot just buy a famous trademark and ship something completely different with a sticker placed on it.)
Anyway, as you stated in the last paragraph, there may be an infinite number of forks, and that is the core of the problem with trademarks and open source. At the moment of forking, the source codes is identical. Then they start to differ. Which one is truer to the original one? Software is not a shoe. I see the continuity of a software project in the philosophy behind it.
Who is the true producer (the one that guarantees the "quality" for the consumer) in this case? The organization or the community? Which part of the community?
In my opinion, this is a big problem that needs to be solved in the future. Maybe using a trademark are not a good idea for an open source project...
"In some countries, when trademarks are transferred, there must be a continuity in the underlying product or the trademark is lost."
If you say so, but it's certainly not the case in the US, or most other countries. Ford for example regularly resurrects names like Mustang, or GT40, and there is seldom any continuity in the underlying product. BMW bought Mini from British Leyland, and developed a completely new car, which pays homage to the Mini shape, but there is no continuity between the two products.
In software there are endless accounts of "complete rewrites" which ship under the same name as an earlier product.
I think trademarks work perfectly well in the Open Source space. The project originator (usually) owns the trademark - others are free to get their own name. In this case it's pretty cut & dried. Oracle owns the product Open Office. Anyone can fork it, but the one true "Open Office" is the one from Oracle.
Now you may argue the merits of the fork surpassing the leader, and if customers buy into the argument they can "opt in" to the new fork. That's their choice. I don't think this hurts consumers, and I don't think it hurts Open Source.
"In some countries, when trademarks are transferred, there must be a continuity in the underlying product or the trademark is lost. (You cannot just buy a famous trademark and ship something completely different with a sticker placed on it.)"
[Citation needed]
In some ways you're technically correct I guess, but you're phrasing it in a way that leads to depictions of reality that are widely inaccurate. The thing is that you can't apply trademarks from one business domain to another willy nilly; but this has nothing to do with the transfer of them. (e.g. if I sell apples under the brand name 'chiquita', that may be a problem, trademark-wise; but selling tv's under the name 'chiquita' may not be). This has nothing to do with the transfer of the trademark.
Then again, I don't claim to know anything more than the generalities of trademarks in just a few countries, so I'm interested in examples of your points from specific legal systems.
Wikipedia is incomplete here, or flat out wrong, depending on how you interpret the article. Trademark doesn't exist solely to protect consumers, it has multiple functions, of which protecting consumers is one, and protecting companies against unfair competition is another. So your last sentence is incorrect - it does exist, amongst other reasons, to protect the owning companies.
If you want the trademark to be owned by a neutral body, then by all means create a body to hold the trademark. That's a choice for the project originator (who usually holds the trademark). In this case there's no dispute that Oracle "owns" the trademark.
Leaving aside the legal implications of a trademark for the moment, your hypothesis is flawed at a social level. Who, after all, is "the community". Let's take an imaginary product - which was created by Fred and is now being developed by a "community" of 10 developers. There are say 200 users of the product.
Now 3 of Fred's developers decide to go and work on a fork. Does Fred give them rights to use his product name? What then does he call his product? How do customers differentiate between the two products with the same name? Since anyone can fork at any time, should we have 25 projects with the same name in the same product space?
Ok, 3 seems a little low - what about 6? What if 3 of those developers joined the project in the last month? Does their leaving count as more or less? What about if 9 developers leave?
How best then to determine which is best for the community of users? Who determines which project is "more true" to the original project? what criteria do we use to measure"trueness"? Does the language dictate trueness? Does the location of the original project lead?
In other words, your point is completely moot. Oracle gets to keep the name because they own it. period. (Although it seems they might now be prepared to give it up.) People have the right to fork the code at any time they like, but it's not in our interest to allow them to use the same name. The public is best served, and protected, by changing the name of forked projects. Anything else is just worse.