As you may have read already, we do not plan on distributing massive amount of fpgas and we will in the end offer our design in open source. And per laseeks recommendation and our spitballing we decided it would be best to have a hard limit to how many we carry at any given time. I agree the danger of a 51% (even if we were only solo mining) is possible and should be avoided at all cost.
I've seen you mention this one before (recommending your FPGA design, if/when it happens, should not be used on pools), but I've had trouble following the logic. If your hash rate per unit of up-front hardware cost is the same or worse than GPU's (and given my background on this topic, I have good reason to bet on worse), there's no particular 51% danger here. Someone could equally well build a GPU farm for the same or lower cost if that was their goal. Recovering ROI from lower power consumption over the long term for choosing an FPGA approach over a GPU approach would make no sense for someone attempting a 51% attack.
I think the only situation where there's a 51% concern from the general public's use of any hypothetical FPGA implementation (and not, say, a PayPal-funded 51% attack) is if you're claiming you have an FPGA approach that achieves a very high performance advantage over GPU's per unit of hardware cost. I personally think you have an uphill battle even hitting the GPU performance/cost ratio. Time will tell though.
I guarantee Sapphire, Gigabyte, EFX, etc.. will manufacture and stock larger quantities of boards containing Radeon GPU's than you'll likely have a market for in the Litecoin world!