Also, I'd like to find some opinions on COLL + EPLL vs OCLL + PLL. I recently finished learning COLL, and looking at a recent thread, I realized (stupid of me for not realizing sooner) that EPLL's only real benefit is theoretically faster recognition, considering that most sub-15 cubers have all their PLLs sub-2, or even sub-1. I guess I'm more doubtful of ZZ-VH now, and I'm curious as to how other people feel.
I don't do ZZ but I can give you a fairly unbiased view of ZZ-VH as I use pure VH myself (with some shortcuts, etc.).
The Vandenbergh-Harris method was originally designed as a stepping stone to the Zbrowski-Bruchem method. I think of it as more genius than the ZB system, but that's for another time. However, even co-creator Dan Harris has questioned the method's viability as a standalone speedsolving system. But the jump to full ZB is really too big to make without a stepping stone, so the VH system survived on CubeZone, CubeStation, cosinesystems, and Jason Baum's Cubing Site. VHF2L is rather slow for what it accomplishes though with edge control prior to the last slot it becomes much faster and more usable (or you can skip the stupidly long/slow cases and use partial edge control + OLL), but sometimes that isn't possible (and that would lead to a whole new analyses that I find fascinating but don't want to expand on right now).
The modern-day selling point of the VH system is no longer its potential as a ZB stepping stone nor its uses for 2x2x2 (CLL completely outclassing COLL for 2x2x2) but rather the step not involving a V or an H (make it a pun if you want and call me wrong): COLL. Many people know COLL (it's only 40 algs) or parts of it (most commonly choosing to not learn the Sune/Anti-Sune cases) but don't use (LL)EO methods. They just use it for the times when EO is available. When faced with a last layer with edges oriented, these people choose COLL over OCLL. This should signify that one is faster than the other. But unfortunately we cannot look at it like this. The temptation of the 1/12 chance for a PLL skip and the high TPS of EPLL clouds or judgment of speed (I am very much a victim of this). Also, the ability to avoid N, V, F, and E (or whatever non-Edge PLL bothers you) just adds to the temptation. At the same time we are giving up consistently fast PLLs like T, J, and A, yes, but guaranteeing an EPLL is usually faster (and more fun). Then there are those (like myself) who started with COLL because ZBLL was the ultimate goal. However, realizing that it would be too difficult (unlike myself) or too time-consuming, they left it at COLL, content with the speed and/or move count it brought with it.
Many people do not learn COLL because they feel it is not as fast as OLL+PLL. While orienting edges is a pain, let us assume that that has somehow already been done. Do you choose COLL or OCLL? Many people feel OCLL is faster than COLL because OCLLs are the fastest COLLs from that orientation. This is not necessarily true, and is even false in my case. When starting out, COLL recognition comes very slowly. You see the OCLL, set it up for recognition, then determine the corner permutation. My COLL learning recognition was very slow because I was basically on my own. My recognition evolution went something like:
Stage 1: OCLL -> CP
Stage 2: OCLL -> 2FSP (2 face sticker pairs: using U and F sticker pairs to determine the CP after determining the OCLL)
Stage 3: 3FSP (using sticker pairs/anti-pairs on U, F, and sometimes R, I know what COLL I'm dealing with)
I've only been in "Stage 3" for about 4 months, but in that time I've noticed that my COLL recognition is faster than my OCLL recognition. Searching for yellow is a pain but just seeing where things are opposite/adjacent/alike is much easier. Multi-color recognition is faster than specific shape recognition. So COLL recognition can be faster than OCLL recognition. EPLL recognition is also MUCH faster than PLL recognition, especially since after you get going with COLL, you know which EPLL is coming up and how it with be AUF'd. I can BLD the last layer (and often do to avoid pops/lockups) as soon as I see the COLL (though there are 8 sets I cannot do this for).
Performing OCLLs is faster- no one can debate that. But the recognition can be faster with COLL, making the speed difference small. Then you get a PLL or an EPLL. Again, small difference. But there is much, much more room for error with full PLL. At high speeds, the difference in time is minimal, sometimes favoring one method, sometimes the other. Some cases are bad. Some orientations are amazing (H and L are my favorites). Some are poor (Pi). When you get a skip every average of 12, though, I think COLL is well worth it.
For the record, my weighted COLL+EPLL average 3 weeks ago was 4.53. My weighted OCLL+PLL average was 5.02. My weighted OLL+PLL (this took me almost a week to finish) was 6.31. Started from the last F2L slot with R U2 R' insertion. Hope I helped!
EDIT: Ooh! Block of text. Sorry about that. Hope I got my facts right. I hope I convinced you to at least try COLL, because that's the only way you'll know if it's worth it.
.
).
).
