I got a reply from Carlos Serrano, who works at the Lawrence Berkeley National Lab in the US, explaining his experience regarding Free and Open Source technology development. I copy bellow his comments, which Carlos kindly agreed to share.
At US National Labs, and LBNL in particular, the mandate of Technology Departments has traditionally been fundamentally different from institutions like CERN. The philosophy at US Labs was that if an invention is created using US taxpayers’ dollars and that invention can be transferred to industry for them to make a profit, then US Labs should get a return on the investment through royalties or other means. The mandates have been revised lately to be much more vague on those aspects. The current situation is that (at least at LBNL) if we want to publish an invention under some sort of Open Source license, we need to first justify that that invention doesn’t have the potential to generate some revenue to the Lab. This hasn’t been a problem for us thus far and we have published both software and HDL in this fashion (treating HDL as software and using a BSD license). Some comments:
- We haven’t yet tried to publish hardware files (schematics, gerbers, BOMs, etc.) but we intend to try at some point,
- We haven’t gotten any opposition to release code (software and HDL) to the public. That was after discussions with our Tech Transfer showing them that the developments in each case were not considered to have much of a real commercial value, which we truly believed,
- There are restrictions on what licenses US Department of Energy Labs can use to publish their software/HDL under. There is a list of DOE-accepted licenses and if you intend to publish under any of those, then each lab’s tech transfer department has the authority to decide to do so. If the license you would like to use is not part of that list then DOE needs to review the license and decide whether it’s generally acceptable for use in our Labs. One problem we’ve encountered with this issue is that we are not allowed to publish under GPL (of any version), because of a clause on releasing patents which is considered dangerous by DOE lawyers. The argument is that it is not explicit enough in that the license only affects to that piece of software in particular and does not grant rights over all patents owned by our organization (the University of California, Berkeley in our case). Whether one shares this view or not does not really matter, those are the rules,
- We’ve published code before as I mentioned, and we’ll continue to do so, but it’s not a huge priority for us because: 1) it requires a substantial amount of effort navigating through the bureaucracy, 2) all of our designs are owned by definition by the US DOE, which means we can automatically share our developments with any other DOE Lab, which is basically 99% of our collaborators, and 3) we are required to keep track on the impact of the material we release and count downloads, etc. and thus far the impact on the code we have released has basically been very close to 0. This doesn’t mean we’ll stop publishing what we do, but you can get a sense that it’s not very high on our priority list.