Communities developing modern open source and free (as in freedom) software now have more than 30 years of experience behind them. During this time software licenses like the GNU General Public License, BSD license and a small handful of others have seen such wide use that they became de-facto standards when it comes to releasing code to the public. Although there are still some gray spots regarding legal issues, as a document from Free Software Foundation Europe nicely explains, the community has developed some stable views on what is acceptable and what isn't under those licenses.
Logo by Mateo Zlatar
The situation is somewhat more complicated in the field of hardware. Hardware mostly falls outside of copyright law that makes software licenses possible. There is a fuzzy line between uploading a RTL design to a software-programmable circuit and etching a printed circuit board based on a schematic, but generally imposing restrictions on manufacturing and distribution of physical objects falls under the domain of patent and trademark law. While in most countries the author of a creative work is automatically granted copyright free of charge, patents and trademarks that would allow you to set any specific rules for use of your design documents are neither automatic nor cheap.
If you look around, practically all popular open hardware projects ignore this issue and simply apply either a software license like GPL or a Creative Commons license to their design documentation: for instance, RepRap decided on GPL, while Arduino uses Creative Commons Attribution Share-Alike for their board designs. This doesn't always have the desired effect though. For instance when someone sells products based on a modified design under GPL, he doesn't need to provide the modified documentation nor credit the original author, since he is not distributing their copyrighted works.
There are attempts to address this issue. First of all, there is now a draft of the Open Source Hardware Definition that aims to be what Debian Free Software Guidelines are for software, outlining the basic requirements that need to be satisfied before a design can be called open source. There are also several license texts that attempt to focus specifically on hardware projects. Ones that appear the most complete and widely used are the TAPR Open Hardware License from the Tuscan amateur packet radio group and the recently released CERN Open Hardware License from the European organization for nuclear research.
Both of these, however, contain some terms that might be surprising for anyone used to software licenses. For instance, both require you to make some reasonable effort to contact the original author or authors when distributing derived works or manufacturing products based on their designs. While a nice provision for one-man efforts, this puts an upper bound on how large a project using this license might become. Imagine trying to contact all the contributors to the Linux kernel today. TAPR has even more terms that limit its scalability. For instance, it specifically requires attribution on PCB artwork, which can waste costly PCB area. It also requires you to distribute "before" and "after" versions of all modified files, which can again become quite unwieldy when you have to distribute the whole history with every copy.
On the other hand, what I miss in these licenses is stronger terms for practical modifiability. While GPL clearly defines that source code must be in the preferred form of the work for making changes in it, TAPR merely says that you must also include open format versions (such as Gerber, ASCII, Postscript, or PDF) if your tools can create them. The openness of the format does not however make the design easily changeable. In fact, all of the listed formats can be compared to a compiled binary in software world. CERN license makes no such constraints at all, but then, neither do Creative Commons licenses, so all are more like the permissive BSD license than the GPL of the software world in that respect. Modifiability of the hardware design files is arguably less important than having source for a binary computer program, since even a PNG image of the schematic will enable you to understand the design, but if you want to have a lively community around your open hardware project, I think keeping designs easily modifiable is a must.
Given all of the above, I find that currently existing open hardware licenses still fall short of specifying good terms for redistribution of the documentation. While they do state the top most requirement for open hardware, that documentation for physical products must be shared, I think it's questionable whether that is actually enforceable in the way specified. From what I've seen, CERN license shows the most promise of addressing these issues and I will closely follow the happenings around it. In fact they already have a list of issues to be addressed with the next version that more or less covers all of my concerns, so I'm keeping my fingers crossed. In the mean time, I'll probably stick to GPL or Creative Commons for my projects involving hardware designs.