There is a better method that I will be pushing for at the same time that I release my own FPGA bitstreams, and that is for all coins to adopt dynamically changing algorithms like Timetravel10, X16R, X16S, X11Evo. These can be mined by FPGA's and GPU's, but no ASIC can ever beat an FPGA on those algorithms, making them the only real 'ASIC-Proof' algorithms invented thus far. The technical details of why that is the case are complicated.
If coins adopted those algorithms, they would never have to go through the trouble of 'forking' since no ASIC can beat a DIY-FPGA on those algorithms anyway.
These algorithms probably have insufficient resistance to the ASIC. It is not enough just to shuffle the SHA3 candidates to get rid of ASIC. It is quite possible to expect the announcement of the miner for these algorithms.
I believe that protection from ASIC should include many huge multiplexers. Those muxes are already in FPGA (although they are statically configured at Place and Route stage), but they will take up a lot of space in ASIC. The downside of this solution is the need to build a bitstream for every state of multiplexer and load a bitstream at the beginging of every block.
In order to distribute computing power equally for everyone, it makes sense to use somehow the bandwidth of the PCI Express link with at least 8 transceivers of GTX in the POW function.
Further, it makes sense to choose a useful function for the POW. For example, the Needleman Wunsh or Smith Waterman algorithms. Thus this altcoin may have a real price.