So you mean that Xilinx would know in advance someone will use pin number h9 of one of their chips as a port in an Ethernet switch and would have instrumented the chip to sniff Ethernet frames as they go through that port and send them somewhere? OK...
Would they do that for all possible data transmission standards (SATA, firewire...) and all possible pins in all FPGA families? I agree at some point you have to trust someone, and this is a point I would be comfortable with.

For critical applications, one can use a White Rabbit switch. White Rabbit is a technology developed at CERN and other institutes and companies. The switch PCB is Open Source (licensed under the CERN Open Hardware Licence) and all the switching happens inside an FPGA for which all VHDL sources are available under LGPL. There is already one company commercializing it, but the sources are all available for any other company to build it, test it, commercialize it and provide support. The terms of the licence give no privilege to any single vendor. No royalties, no patents. Plus the HDL can be customized for particular applications (low latency, redundancy...).

I meant "to manufacture UTC" instead of "to manufacture GPS". Sorry.

The authors have documented their whole procedure here: http://www.ohwr.org/projects/cngs-time-transfer/wiki The author of the bogus paper assumes the people who designed GPS and those who use it in metrology labs around the world to manufacture GPS do not know anything about relativity. He also proceeds to an analysis without checking his very basic premises first with the authors of the neutrino velocity paper, or anybody close to the actual experiment. Is it that hard to check one's assumptions first?

OSHW hosts a definition, not a license. It tells you "if you want to call your license Open Hardware by our definition it needs to fulfill these points". CERN OHL is a license that complies with these conditions.

I am not an expert, but I think the big difference with Physics is that there is a lot of money involved (patents for new drugs, etc). In Physics, CERN and others had the vision of Open Access http://library.web.cern.ch/library/OpenAccess/ for scientific publications, a mode in which the editing expenses are paid by the authors so that readers can access freely. This was possible because the big publishers were not really given an option: if they said no, physicists would publish on already-existing free places like arxiv.org which give comparable (if not better) visibility. Since there are rarely any immediate money-making applications for Physics papers, an atmosphere of openness could develop and once Physicists got used to it there is no way back.

CERN and others have already designed Open Hardware in the OHR. See for example the SPEC board http://www.ohwr.org/projects/spec/wiki the 100 Ms/s ADC mezzanine http://www.ohwr.org/projects/fmc-adc-100m14b4cha/wiki and the Rhino http://www.ohwr.org/projects/rhino-hardware-01/wiki. All of these, and others, you can buy from a number of vendors, and more are in the pipeline. Have a look at the OHR projects page: http://www.ohwr.org/projects

Here's a non-hobbyist site for control and data acquisition hardware: http://www.ohwr.org/