When and how to move from 2-look to 1-look OLL

[mark49152]

There's an assumption there that all 57 are worth learning, but clearly some offer much more benefit than others. I'm too lazy to learn algs that might save me 2 moves once every 54 solves

It might be interesting to extend this project to look at other routes through the last layer - for example, maybe learning OLLCP algs for OLL case A gives better reward than learning OLL for case B, in terms of impact on averages overall. OLL's not the only choice of what to learn next.

I'm also curious to know how often top CFOP solvers stick to a strict CFOP sequence or throw in other steps like edge control, OLLCP, COLL, etc. I'm guessing that the most efficient approach would be to know a variety of routes depending on cases, and the same approach could be taken to measuring the relative cost of different routes and choosing the learning order accordingly.

 

[Niah]

new to the forums here, hi! just want to say alg 41 is one of my favorite algs, double sledge hammers and then double sexy's ((y') (L F' L' F)*2 L U' L' U L U' L), it's quite fun! i learned full oll with reflections, but my average is sadly 40+ seconds . learning oll took me 3 weeks, i'm so surprised how fast people are without full oll. i feel so bad, haha

 

[Lagom]

Im working on a list of scrambles for OLL´s, categorised after what kind of shape they are. So if you learn for example all P-shaped OLL's you´ll have a list of scrambles for those, so you can practice. You guys want that?

 

[Dakotajennings]

Im working on a list of scrambles for OLL´s, categorised after what kind of shape they are. So if you learn for example all P-shaped OLL's you´ll have a list of scrambles for those, so you can practice. You guys want that?

Yes!

 

[PianoCube]

Im working on a list of scrambles for OLL´s, categorised after what kind of shape they are. So if you learn for example all P-shaped OLL's you´ll have a list of scrambles for those, so you can practice. You guys want that?

Like this one?: http://thesixsides.com/rubikscube/algorithms/oll.php

 

[jayefbe]

Like this one?: http://thesixsides.com/rubikscube/algorithms/oll.php

I only see algs on this site.

I would also like scrambles to practice OLLs, although Marcel's program already provides a function similar to that. I believe his scrambles are mostly the inverse of the OLL algs, maybe with some rotations thrown in. So maybe collaborating to make a program with scrambles that won't reveal which OLL it is, while also being able to pick and choose which OLLs you want to work on would be interesting.

 

[Jorghi]

I only see algs on this site.

I would also like scrambles to practice OLLs, although Marcel's program already provides a function similar to that. I believe his scrambles are mostly the inverse of the OLL algs, maybe with some rotations thrown in. So maybe collaborating to make a program with scrambles that won't reveal which OLL it is, while also being able to pick and choose which OLLs you want to work on would be interesting.

Pretty much get a list of 100+ inverse OLL algs from cube explorer.

 

[Lagom]

Here you go https://dl.dropboxusercontent.com/u/95991760/OLL-Training.xlsx
Note that there's two sheets. The second sheet contains the scrambles.

Im gonna update it soon tho. Make sure the scrambles switches places every time you open the document and stuff like that. Maybe even add timer and stuff if I figure out how to...

If anyone knows programming and want to use this idea to make a simple program, let me know

Note that I took all the algs, pictures and comments from www.cubewhiz.com (thank you), so no credit for me about that!

 

[mark49152]

This is nice if you want to set up a specific OLL. Alternatively if you don't want to know what OLL case you will get, use a last layer scrambler like qqTimer.

Also for anyone learning OLL, I recommend reading back to the first post in this thread. That proposes an OLL learning order based on how much time each will save you over 2-look. For those in too much of a hurry, here is a breakdown of how many moves on average each OLL group will save you (basically, don't learn W's first! )

10.00	Stealth
8.50	Squares
8.13	Dots (1/54)
7.50	T shapes
7.31	Small lightning bolts
6.67	P shapes
6.00	Kites
5.88	C shapes
5.58	L shapes
5.56	Knight moves
5.38	Fishes (excluding kites)
5.25	Big lightning bolts
4.25	H
4.13	I shapes
3.81	Dots (rare)
3.44	Awkward
2.63	W shapes

 

[Lagom]

This is nice if you want to set up a specific OLL. Alternatively if you don't want to know what OLL case you will get, use a last layer scrambler like qqTimer.

Also for anyone learning OLL, I recommend reading back to the first post in this thread. That proposes an OLL learning order based on how much time each will save you over 2-look. For those in too much of a hurry, here is a breakdown of how many moves on average each OLL group will save you (basically, don't learn W's first! )

10.00	Stealth
8.50	Squares
8.13	Dots (1/54)
7.50	T shapes
7.31	Small lightning bolts
6.67	P shapes
6.00	Kites
5.88	C shapes
5.58	L shapes
5.56	Knight moves
5.38	Fishes (excluding kites)
5.25	Big lightning bolts
4.25	H
4.13	I shapes
3.81	Dots (rare)
3.44	Awkward
2.63	W shapes

Thank you for that list
Which ones are stealth and kites? Never heard those categories before

 

[mark49152]

Thank you for that list
Which ones are stealth and kites? Never heard those categories before

Stealth = OLL-28, H = OLL-57, the all corners oriented cases.

I separated out kites (OLL-9 & 10) and fishes (OLL-35 & 37) because although the wiki groups these together, they are distinctly different U patterns. Note that the kites can be solved by Sunes with M setup moves so are in the same "family" as the squares and small lightning bolts.

The dots are also separated into those with 1/54 probability and those with <1/54 probability; the move savings for the latter are weighted by their probability which is why they are so low in the list. Although rare, when one actually arises you still save a lot of moves (8-10) so they're good to learn anyway.

 

[Lagom]

Stealth = OLL-28, H = OLL-57, the all corners oriented cases.

I separated out kites (OLL-9 & 10) and fishes (OLL-35 & 37) because although the wiki groups these together, they are distinctly different U patterns. Note that the kites can be solved by Sunes with M setup moves so are in the same "family" as the squares and small lightning bolts.

The dots are also separated into those with 1/54 probability and those with <1/54 probability; the move savings for the latter are weighted by their probability which is why they are so low in the list. Although rare, when one actually arises you still save a lot of moves (8-10) so they're good to learn anyway.

Thank you

 

[pipkiksass]

Mark, you am legend! I subscribed to this thread AGES ago - I got about 1/2 way through reading the OP and my daughter started crying, knew it would be worth coming back to. Are your algs all selected on length, the 'common' algs, one particular source, or 'other'? Would be interesting to have a similar comparison based on time, as well as move count - i.e. finger-tricked 2-look algs vs. the optimum finger-tricked 1-look alg for each case.

Funnily enough, I only know c.20 OLLs, and I'd started learning the W's. Definitely won't bother now!

Should be stickied. Fantastic work, thanks a lot.

Out of curiosity, how far along your through your own learning order have you got so far?

 

[Crowned xerxes]

I don't think it matters.
I average 16 seconds and I know 16-21 plls. And only like 8 olls.

 

[Lagom]

I don't think it matters.
I average 16 seconds and I know 16-21 plls. And only like 8 olls.

Well if you knew the oll´s you´d average like 14

 

[mark49152]

@pipkiksass: Thanks, glad you like it

Are your algs all selected on length, the 'common' algs, one particular source, or 'other'?

They are mostly Badmephisto's with a few replaced by more optimal algs from the wiki where they are significantly shorter. I wanted algs that were representative of execution time, relative to each other; I made an assumption that if there's a longer alg usually preferred, it's because it's finger-tricky and faster than the move-optimal alg and therefore the move count of the optimal alg is more representative. I'm not recommending any of these algs, just using them for estimation purposes.

Would be interesting to have a similar comparison based on time, as well as move count - i.e. finger-tricked 2-look algs vs. the optimum finger-tricked 1-look alg for each case.

Yes agreed - see the above comment, and also note that I used a non-standard metric, counting quarter turns (including slices) as 1, half turns as 1.5, and rotations as 2. That (for me) gave a better approximation of execution time as a function of constituent moves, based on a few tests. Of course execution time is subjective, and this is still only a rough approximation that in particular doesn't account for finger-trickiness; but this whole thing is rough approximation so I didn't want to get too anal about the metric.

Out of curiosity, how far along your through your own learning order have you got so far?

I've learned the "Top 15" and I already knew the other dots. At my blistering 4tps execution rate, that's a saving of ~2 seconds per solve when they come up. I have no plans to learn any of the others, at least until I'm a lot faster and have sorted out some of my bigger weaknesses, like cross & F2L.

 

[Ok ganta]

Awesome...

 

[Alphalpha]

Well, obviously you should learn them when you think they will benefit your solving more than anything else you could spend your practice time on.

I understand this catch-all suggestion and agree wholeheartedly. But, since I searched and found, I will ask my original question anyway. Any suggestions on what speed/avg I should be looking for with 4LLL before I adopt 2L-OLL? Perhaps other's personal experiences trials and successes will help me decide when to start learning.

 

[supercavitation]

I understand this catch-all suggestion and agree wholeheartedly. But, since I searched and found, I will ask my original question anyway. Any suggestions on what speed/avg I should be looking for with 4LLL before I adopt 2L-OLL? Perhaps other's personal experiences trials and successes will help me decide when to start learning.

If you're using 4LLL, you're using 2-look OLL...

 

[josh42732]

I understand this catch-all suggestion and agree wholeheartedly. But, since I searched and found, I will ask my original question anyway. Any suggestions on what speed/avg I should be looking for with 4LLL before I adopt 2L-OLL? Perhaps other's personal experiences trials and successes will help me decide when to start learning.

There is only 7 algorithms for edge-oriented OLL cases and I highly recommend looking at this video about edge control for OLL. It really helped me do 2LLL and brought my times down: http://youtu.be/hgOORsYS07k