Of course, if a gaming group wants to use Harnmaster or HM Gold for their game, the systems should still be for sale and not open source.
Actually, I'd recommend the opposite. If anything that should be open source, it should be HarnMaster. The value is in the setting, and the system can help draw in new customers. I came to harn by way of CGI's free HM3 pdf offer that was made available in January of 2004. Since then I've purchased alot more, but if it wasn't for the free pdf, I probably wouldn't be here today and I'd probably be in the middle of failing to design my own system (or using d20), and not getting very far with my own setting (not that anything has changed in regards to that... as I added a third setting to my list of projects).
WotC didn't make their campaign settings open for a reason, they made the game rules open, because their value is in the network they create. Atleast when it generally comes to D&D, settings and system are usually presented as being fairly intertwined (it's one of the reasons I like Harn because HM and HW are fairly independent of each other). While there are some who might use the online SRD, many eventually purchased their own set of books. Open source HM1/2/3/G would in the long run pay for itself and sell the system, and also sell the setting.
The main concern alot of people have is that an Open Source HarnWorld would threaten the quality of the setting. I haven't heard anyone claim canon violations when it comes to someone's houserules
If CGI/NRC were to experiment with Open Source Harn, I recommend that the rulesets be the first part of that experiment. From there, other materials can be made available to the Open source community.