This is a kind of subjective question:
Is it worth it to learn all 57 OLL algs? (versus Winter Variation) I feel that WV is better, but is the overall move count lower? (because after all it's breaking OLL into two steps: EO then WV) and if the move count is relatively the same, which has better recognition and execution?
...
well here's my opinion
this is likely superior to 2 look OLL, but not to 1 look OLL. the reason is because EO + CE pair + WV + PLL is 4 looks, while LS + OLL + PLL is 3 looks.
if you assume you've already learned all the OLLs, you could do this for solves where the EO is done, like by influencing EO while inserting other pairs. however, this is much more advanced so as a learning thing, that wouldn't work. for ZZ or Petrus, WV is always a viable option.
CE pair + WV is 2 looks, since you can't do WV on any LS case, only when the pair is made. And if the pair is like the R U R' case, you need to learn more algs from summer variation. otherwise, you'd have to do something like R U' R' U2 to reduce to WV, increasing the average moves to be more than LS + OLL + PLL. recognizing corner orientation in WV could also be tougher than recognizing OLLs.
and once you've learned all the OLLs, it might seem a waste to have all the edges oriented during LS when you could be doing WV. worry not! there is COLL to take advantage of that.
If you compare WV to VH, they're quite similar. Easy to execute cases, easy to learn, etc. However, they add an extra look to the F2L LS, therefore reducing their potential. That's why the ZB method has ZBLS, which is 1 look VHLS. However, that is too hard to learn, and what would make your method theoretically much faster is if you used a "ZBLS" version of WV (1 look instead of 2). However, the alg count would be ridiculous.
