Columns-First

[stinkocheeze]

I came a across an article on the wiki about "Columns First" Method.
There hasn't been much info about it though..... It there a good youtube tutorial, and Is it worth learning?

 

[AndyRoo789]

http://www.speedsolving.com/forum/showthread.php?t=18342

 

[stinkocheeze]

http://www.speedsolving.com/forum/showthread.php?t=18342

sorry. My Search didn't come up with anything

But still, it's 2 pages of argueing

 

[Athefre]

Columns First, the plain "Four Pairs, CLL, etc.", isn't worth learning when Roux exists. The two things that kill Columns First are the view (the edges aren't oriented, that's what makes it so bad) and the amount of S moves (or rotations).

 

[qqwref]

Yes, it doesn't have the potential Roux does, but it's good enough to get a sub-20 average. One version I tried was:
- Four F2L pairs
- OCLL (don't do C*LL)
- Pair up a first layer edge with its center and insert. Do this twice to fix all centers.
- Put in a third edge however you like.
- Insert the last edge while orienting the remaining edges.
- PLL.

Another variation which is closer to Kenneth's intention is to do the pairs, C*LL, then do the centers and two D edges of one slice, and finally do roux-style L6E on the other slice.

 

[Athefre]

Another variation which is closer to Kenneth's intention is to do the pairs, C*LL, then do the centers and two D edges of one slice, and finally do roux-style L6E on the other slice.

Yeah, that's the way I would do columns first as well if I was solving pairs style. What would really help columns first is if people were orienting edges early in the solve. There could be so many easy variations afterwards. For example:

-Insert final pair while orienting U corners.
-Permute U corners while separating M and S slice
-Finish M and S (about 8 moves to do this [4 per slice], not counting parities)

Another way I've considered is on my site. After all edges are oriented and the columns are finished, solve three edges and finish the rest in one sequence. It seems to work really well with a lower move count than simply separating them into their correct layers. Though look-ahead is still difficult.