Hi. I'm going to have to go with "wait, lets add more options...". Here's why:
x11, scrypt, primeWhy just 3 when you can have 5? Actually, there's no real limit to how many algorithms you can run simultaneously. The only trouble is that, if an algorithm's hash power is too small, it can easily become a victim of difficulty hopping. This is where a large miner or pool starts mining when the difficulty is low. They get many blocks in a short time, then the difficulty soars, and they stop mining. The hash rate drops dramatically and miners take forever to find the next block. The difficulty drops, then the big fish jumps back on. Repeat. In multi-algo this is less of a problem as the other algos keep the blockchain running smoothly, but the big miner is getting cheap coins in the end, which leads to dumping, and weaker network security.
Of course, the more algorithms a coin uses, the smaller the portion of block rewards per algorithm = lower hash rate = easier target for difficulty hopping.
Our experience with Myriad shows that 5 algorithms seems to work fine, even on fairly low hash rates. However, in the past we did have issues with scrypt and sha256d (not our GPU or GPU/CPU algorithms). We solved this by enabling merge mining on scrypt and sha256d. Hash rate skyrocketed, and the network has been running great ever since.
x11, scrypt, prime, sha2562.5 ASIC algos, and 1.5 GPU algos. I dunno... that's a lot of ASIC
x11, scrypt, lyra2rev2Why leave out sha256d? It has the most mature ASIC market today. If anything, you would be better off having sha256d than scrypt or x11. Also, the issue I have with lyra2re2 is it's mainly just used by Vertcoin. Vertcoin devs publicly state they will change the algo in the future, so lyra2re2 has rather limited support.
x11, x13, x15, x17This one has merit. x11 ASICs are starting to come online, and an x13 ASIC would appear to be only one step away, x15 two steps away, and x17 three. This could have the effect of allowing a very even, gradual shift from the GPU to the ASIC phase. Essentially, that's what every coin wants.
I'm not voting for this option yet, but the community should consider it.
scrypt, prime, x13
lyra2rev2, x13, primeSeem like fairly random choices. Why x13? Why not neoscrypt or keccak for GPU? Neoscrypt is a more popular GPU algorithm used in Feathercoin and others.
keccak, x11, prime, scryptAgain, why scrypt and x11 but not sha256d?
scrypt, x15, lyra2rev2, blake2s, primeWhy x15? Seems random... again, why not neoscrypt or keccak?
My picks...I am a bit biased here, but personally I would like to see a flock of multi-algo coins all merge mining with each other and supporting each other. Therefore, it would be nice if Verge used one of the Myriad GPU algorithms. They are fairly supported now that they're used by Digibyte and soon also by Auroracoin. So with that, I would go with:
sha256d - best ASIC algo, no brainer
scrypt
x11
prime
neoscrypt
groestl - Myriad GPU algo
All merge mined.
6 algos. 2.5 ASIC, 3.5 GPU. Groestl should be fairly easy to implement in an ASIC. Neoscrypt and prime will probably take a significantly longer time to see ASICs. Plus, Verge could claim to be the first 6 algo coin. I think that has great marketing potential