The use of BSD also allows us to market the OS by saying something like, "It's based on BSD, a little like Mac OS X." Anybody with half a brain would see through this, but technical users aren't the target market. Techies already have a first-rate open source operating system. It's called Linux.
There are no great surprises when it comes to the desktop. I would make use of the Gnome project, which I believe to be one of the best and simplest open source desktop interfaces. However, I would sponsor the creation or adaptation of a toolbar-based program launcher/dock. This is the fashion right now, of course; the next version of Windows will have such an interface, and OS X has had its Dock for years. The toolbar will be used to launch programs, and also minimize programs. Pretty simple, really, although I'd want my effort to be ultra-intuitive and easy to use.
In terms of supplied software, we are again looking at a Linux-like collection--Firefox (or maybe Google Chrome, although I'm unsure if it is being ported to BSD, or even if its feasible). I'm not 100% sure I'd include OpenOffice.org, but would investigate tying into an online office suite via something like Prism and Google Gears. However, I'd want both beefed up with encryption, to guarantee complete data privacy. I may sponsor a browser plugin that encrypts/decrypts on the fly, so that any data stored online is secure. For what it's worth, it seems to me that this is the only way that online applications can progress.
Perhaps the most surprising aspect of my new OS would be the inclusion of Wine, along with a backroom project consisting of developers who create scripts and tweak the code so that the new OS runs the majority of popular Windows software and games. This is not entirely unlike Codeweavers, in fact, the commercial offshoot of Wine.
The compatibility scripts and tweaks would be part of the updates downloaded to each computer on a regular basis. (Needless to say, support would be as free of charge as the software itself.)
The goal wouldn't be 100% Windows compatibility, because that's just too high a bar, and would set us up for a fall. The goal would be to support the most popular applications and games. By publishing honest lists of what does and doesn't work, as with the hardware lists, we could earn the respect of the user base, and not make promises that we couldn't keep. Of course, I envision a community arising around the OS, who could also contribute to this effort.
In terms of program compatibility, the goal would be to offer a halfway house between commercial operating systems like Windows, and open source operating systems like Linux. I think this would pay off, because in my experience people are intrigued by the open source approach, and receptive to its concepts, but simply scared off by Linux.
We'd also have software repositories packed full of precompiled open source software, of course, just like any decent Linux project worth its salt. The best of open source would be available.
A major goal of the OS project would be for it to run quickly, even on modest hardware. I would set an arbitrary ceiling on the hardware we expect the user to have (probably something like 1GB of RAM and at least a 1.5GHz CPU).
Optimization and efficiency is something of a fashion right now, and both Windows 7 and Mac OS X 10.6 are promising substantial performance improvements.