@Ken: I've been looking at your Ahtlatl FPGA board and I think it's very interesting, especially when combined with all the support documentation and examples you and your students are working on -- I'm planning on looking into this further in the not-so-distant future.
Several of our customers, our engineers, and engineering students use FPGA boards to make quick prototypes and test fixtures. In some cases it's for a one-off prototype, demo, or as part of a production test system. In others, it's an application where the chosen processor simply "ran out of gas" and just didn't have the throughput required to handle the processing in a timely fashion. And sometimes we've used an FPGA board to provide a quick solution to a component obsolescence problem. That replaces the obsolete part with a current production part and uses the FPGA as an intermediary between the processor and the new IC. The FPGA emulates the software interface on the processor side and translates signals from the processor into those compatible with the replacement part. While adding an FPGA might seem expensive, modifying the software, retesting, and requalifying the system may be far more costly and time intensive than the cost of requalifying the replacement hardware. This is especially true in the case of an FDA approved medical device or other application which has similarly long and painful requalification cycles.
For our purposes, we needed a better way than using the various demo and eval boards, which always had stuff we didn't need, and were ususally too big or had inconvenient connections for prototyping. That's a major reason why we implemented a bare bones design for the Ahtlatl FPGA board we developped for OEM ans student projects. It has just an FPGA, power supply, USB / JTAG configuration interface, and two 50 pin 0.1" header connector postions. (See http://DIYchips.com and http://htevp.com ) The typical eval board I/O was left off the FPGA board and placed on a separate PC board, so that most of the FPGA pins are available to the user. The board was also designed with constant impedance traces placed for high speed differential signals and selectable logic signal levels on the two 50-pin connectors to handle level translation.
Overall, an FPGA board is probably the closest thing we've got to a "universal digital translator / adapter" between two digital devices that otherwise won't play well together!
I remember when the first FPGAs appeared on the scene. 8 x 8 arrays of programmable logic cells where each cell held a 4-input lookup table (LUT), register, multiplexer, and not much more. Applications for these little scamps where pretty much lmited to gathering glue logic and implementing simple stste machines.
Look at them now -- dual hard core MUCus, programmable digital, programmable analog, huge amounts of on-chip memory, vast numbers of high-speed multipliers, large numbers of high-speed tranceivers, high-speed external memory interfaces...
So yes, I agree that development boards based on these modern devices can make ideal prototyping platforms for a vast range of applications.