Could someone clarify this part:
"Instead of deploying complicated, CPU-friendly algorithms, we use use simple GPU and ASIC friendly algorithms. And as soon as an ASIC is on the verge of being made, we simply change the algorithm by automated hard forking"
If you want to be asic resistant, why choose an asic friendly algo ? And to what will you change it if asics are going to be used.
ASIC friendly algos are also GPU friendly, so are very fast on the GPU and have negligible speeds on CPU, which has it's advantage.
Blake is a GPU friendly algorithm which provides maximum protection, cause GPU power is the most expensive to lend.
The disadvantage is that it's simple so ASICs can be made for it without much R&D.
Since we can always hard fork before a reputed company claims to be able to produce such an ASIC, we'll change the algorithm (in the next hard forking schedule) to something which does not have an ASIC like Ink, SHA-3, X13, Quark etc... depending on what's available at that time when an ASIC is on the verge of being made.
