GuRoux
Veteran
- Joined
- May 6, 2013
- Messages
- 1,713
- Location
- San Diego, California
- WCA ID
- 2014TANG03
- YouTube
- Visit Channel
how is that going with roux?
Switching Methods Progress Thread
- Thread starter IRNjuggle28
- Start date
IRNjuggle28
Veteran
Ehh... it's OK. I'm a lot better at filtering pieces for F2B now. One thing I'm doing that is likely a bad habit is that I solve 2 cross pieces, and then do the 4 F2L pairs, instead of doing first block/second block. It makes it easier to remember what I'm doing, but maybe I should just do it right instead of making Roux more CFOPpy. I also really need to memorize my L6EO algs. I know less than half of them. I also don't even really know 2 look CLL. I need to learn better algs. That's actually one of the reasons I looked towards Roux. I heartily dislike learning algs.
With CFOP, I average about 23, and finish F2L at about 13. With Roux, I finish F2B + CLL at 20something seconds, and then spend forever reinventing the L6E wheel. I'm probably at 45-50 seconds right now. And it would instantly drop to low 30s if I memorized the L6EO algs and learned to recognize L4C better.
IRNjuggle28
Veteran
It's possible that you'll have an easier transition to Hoya if you haven't practiced your previous method for a while since the cube broke. Lol. Maybe it'll be helpful!
Knut
Regular
Currently I am averaging about 26 seconds with CFOP. I heard about Roux a little while ago, and when a tried it (ish) it really seemed to click with me. I'm thinking of switching over, but I can't seem to find many Roux resources. Does anyone have any links to CMLL algorithms and something with a good explanation of the last six edges?
Renslay
Legend
Here you can find many good things about CMLL and LSE (as well as other things about Roux):
http://grrroux.free.fr/method/Intro.html
http://rouxtorial.webs.com/
http://wafflelikescubes.webs.com/rouxmethod.htm
Some extra CMLLs:
http://www.speedsolving.com/forum/showthread.php?20355-CMLL
http://www.speedsolving.com/wiki/index.php/CxLL
Also look for "all cmlls sub 1" in youtube. You cand find some really fast algorithms for CMLLs.
https://www.youtube.com/watch?v=0M7Q2fABvuE
https://www.youtube.com/watch?v=rPwUEiDwuqs
rowan
Regular
This was talked about awhile ago, but for what it's worth, I'm fully CN with Roux. There are definitely fully CN Roux solvers. I started doing solves about 3 weeks ago and I'm around 1:00. I didn't switch methods though, the first method I learned was Roux and it's currently the only method I can solve with.
I agree basically with everything Kirjava said in this thread and the CN discussion thread. I think the recognition approach for CN is *different*, not necessarily worse than the approach for opposite CN or color fixed. To say there are no benefits for being a CN Roux solver is pretty uninformed, though. The average move count for the first block is lower when you're CN. Any discussions about the efficiency of recognition are really too abstract to reach a definitive conclusion. Comparisons don't really exist because as Kirjava has stated many times before, no sub-10 solvers have successfully switched and either way their personal experience would not be enough to form any concrete conclusions about recognition. I think the more important focus is practicing with whatever choice you do make, though, and trying out lots of methods only seems to be a hinderance to speed.
As a side thought: Perhaps there aren't many CN Roux solvers because Gilles Roux suggests that you don't be CN? This is pure speculation though, I could be incorrect about that influence.
Cool Frog is also fully CN with Roux and ~13 seconds.
I agree basically with everything Kirjava said in this thread and the CN discussion thread. I think the recognition approach for CN is *different*, not necessarily worse than the approach for opposite CN or color fixed. To say there are no benefits for being a CN Roux solver is pretty uninformed, though. The average move count for the first block is lower when you're CN. Any discussions about the efficiency of recognition are really too abstract to reach a definitive conclusion. Comparisons don't really exist because as Kirjava has stated many times before, no sub-10 solvers have successfully switched and either way their personal experience would not be enough to form any concrete conclusions about recognition. I think the more important focus is practicing with whatever choice you do make, though, and trying out lots of methods only seems to be a hinderance to speed.
As a side thought: Perhaps there aren't many CN Roux solvers because Gilles Roux suggests that you don't be CN? This is pure speculation though, I could be incorrect about that influence.
Cool Frog is also fully CN with Roux and ~13 seconds.
ChickenWrap
Veteran
Not to be rude, but it isn't that hard to be CN with Roux when your average is 60 seconds. Once your average is significantly faster (sub 20), CN isn't feasible/efficient for everyone.
Obviously, a select few are fast at it, though.
rowan
Regular
Ah, yeah, of course I realize this, I'm really slow.
But as others have stated before, I don't think the lack of many sub-20 CN Roux solvers really says anything about the abilities of CN with Roux.
I'm not sure how you can assume that once I reach sub-20 I will find CN inefficient. I'm curious how many people who started CN with Roux end up switching out of it because they find it inefficient, I doubt many people would, honestly. It seems that the best way to do anything is to pick it and stick with it. Any amount of practice will increase efficiency.
Last edited:
jeff081692
Veteran
I think the best way to see the benefit if any of color neutrality regardless if you are currently CN or not, is to do something like an average of 100 for the first step with the same scrambles 6 times (one for each color and rotating before doing a scramble again so you keep your familiar colors). CN only gives you an advantage in the first few moves at the expense of having worse recognition for every other step until you catch up to your normal speed. So once you have done all 600 solves of the cross or cross+1 for fridrich you can easily compare how you would do if you only solved one color or fully CN and anywhere in between. The averages for each individual average of 100 would likely follow a trend and the best solves from opposite colors would show you being dual CN.
I only picked 100 because that should have enough data for you to really start to see trends when comparing things with lucky cases not being as big of a factor in changing averages.
And people who are sub10 could change but when you solve multiple events and have other things you could improve the difference between CN and non CN is rarely worth the effort. The setback of CN in the start is a lot more forgiving on your times than switching methods though so it could probably be done in 1-2 years of steady practice even for the fastest cubers.
I only picked 100 because that should have enough data for you to really start to see trends when comparing things with lucky cases not being as big of a factor in changing averages.
And people who are sub10 could change but when you solve multiple events and have other things you could improve the difference between CN and non CN is rarely worth the effort. The setback of CN in the start is a lot more forgiving on your times than switching methods though so it could probably be done in 1-2 years of steady practice even for the fastest cubers.
Last edited:
IRNjuggle28
Veteran
I think that is a terrible way to gauge whether or not color neutrality is helpful.
Problems:
Doing the same scramble 6 times is silly. You'll practically have the entire solve memorized. The point of that experiment is to find out whether or not neutrality hinders recognition (which is already an overly simplistic way to look at it) and you won't need to recognize anything by the 6th solve because you'll have already done it 5 times. Do you really think that solving the same scramble over and over won't affect the time it takes to solve?
Someone who is color neutral would not have any data for not being color neutral. If they did an average of 100 with the same color all 100 solves, their recognition would not be improved or worsened. It would be the same. Because they don't have a color bias. Their times would worsen because of having a less efficient cross most of the time, but their recognition would not be improved the way it would with someone who was not color neutral. A color neutral solver doing a fixed orientation AO100 does not magically develop a better color filter for the color they decided to solve.
600 solves isn't that many. The standard deviation of averages of 100 is way too high for this to work unless you repeated the experiment dozens of times.
There are more things you said that don't make any sense, but I'll stop. I'm feeling a little bad for Kirjavaing at you, but some of the stuff you said was really far out there.
IRNjuggle28
Veteran
So I have a confession to make. I have decided to not use Roux as my main 3x3 method, despite starting this thread because I'm was going to switch to that from CFOP.
I've learned most of Roux (I just need to memorize some more LSE orientation algs. I understand the method) I'm going to learn it the rest of the way so I can teach it to other solvers, and I still do like the method better than CFOP, but I've decided it's just not for me. I solve mostly big cubes. Mostly 5x5 and 7x7. I'd rather not spend a bunch of time practicing a method that I can't use for my favorite events. However, I'm glad I learned Roux, and I'm sure I'll occasionally do some Roux solves just for fun.
Renslay, thank you. That was perfect.
I've learned most of Roux (I just need to memorize some more LSE orientation algs. I understand the method) I'm going to learn it the rest of the way so I can teach it to other solvers, and I still do like the method better than CFOP, but I've decided it's just not for me. I solve mostly big cubes. Mostly 5x5 and 7x7. I'd rather not spend a bunch of time practicing a method that I can't use for my favorite events. However, I'm glad I learned Roux, and I'm sure I'll occasionally do some Roux solves just for fun.
Renslay, thank you. That was perfect.
jeff081692
Veteran
@IRNjuggle28 It's fine. But I don't think you understood that you wouldn't have solves memorized because you are approaching a new cross configuration each time you did the same scramble. Another way of thinking of it is just doing the cross+1 of one scramble on each color then moving on to the second scramble (the reason I said that stuff about rotating before scrambling again is that you get the same effect of being CN while still solving with your familiar starting color(s). at the end you have results from if you were color neutral compared to fixed or dual CN on that sample size of solves.
You are right about someone being CN potentially having worse times not because of recognition but the efficiency of the cross and that is exactly what I am designing this thing to measure. Someone who is CN would have data of not being CN because they have less chances for an easy cross/xcross and that would be reflected in their results. At the same time someone who is not CN would have data on what it's like being CN because when you adjust the color scheme by rotating before each scramble they get extra opportunities for easy crosses that would improve their average of 100 cross+1.
The last thing I have to say is that even with one scramble there will always be a cross that is faster/easier to execute than the others just like some PLL are faster than others. The difference is probably really small but for someone like me who is inefficient in the cross that difference is often 1-1.5 seconds. If I had efficient crosses the difference in CN vs non CN would be smaller assuming recognition was even across colors for the rest of the solve.
You are right about someone being CN potentially having worse times not because of recognition but the efficiency of the cross and that is exactly what I am designing this thing to measure. Someone who is CN would have data of not being CN because they have less chances for an easy cross/xcross and that would be reflected in their results. At the same time someone who is not CN would have data on what it's like being CN because when you adjust the color scheme by rotating before each scramble they get extra opportunities for easy crosses that would improve their average of 100 cross+1.
The last thing I have to say is that even with one scramble there will always be a cross that is faster/easier to execute than the others just like some PLL are faster than others. The difference is probably really small but for someone like me who is inefficient in the cross that difference is often 1-1.5 seconds. If I had efficient crosses the difference in CN vs non CN would be smaller assuming recognition was even across colors for the rest of the solve.
Last edited:
IRNjuggle28
Veteran
Ahh. Yes, I did miss the fact that each cross would be different. I thought you were saying do the same scramble 6 times and scramble with a different color on D each time. That makes much more sense. However, I still don't really see what you're trying to prove that hasn't already been figured out. We know that you'll have a longer cross. Getting experimental data on just how much longer doesn't sound helpful. People have already figured out mathematically exactly how long the average optimal cross will be with AND without color neutrality.
jeff081692
Veteran
I see what you mean. It was just a random thought that came to me and since someone mentioned CN I just posted it. There are benefits for me at least of comparing the averages rather than optimal move count maybe others find it pointless and I honestly don't expect anyone to waste practice time doing what I proposed it was just a hypothetical that is one way to see what you would average at that point in time.
I would like to know how long the average optimal cross for quad-CN is compared to full CN though.
IRNjuggle28
Veteran
That would be interesting, yeah. I have no idea, but I bet you can get someone to crunch the numbers of CN vs quad-CN for you in the probability thread: http://www.speedsolving.com/forum/showthread.php?20384-Probability-Thread
Quad CN is rare, though. You still have to build different color filters for the E layer colors, which you don't have to do for bi-CN. And because of that, I doubt it's really that much easier than being fully neutral. If someone is quad CN, they might do well trying out full CN. Agree/disagree?
jeff081692
Veteran
Well the thing is I am dual CN with blue and green and while training for CN I liked practicing on white more and so there is an imbalance where my white/yellow is 1 second slower than my blue/green but my orange/red is 3 seconds slower than my blue/green. Being dual CN it is easier to become quadCN because you are used to doing opposite colors so even if I practice white cross all day my yellow improves just as much. to become full CN would require twice the amount of effort because not only do i have to get 1 second faster on white/yellow but also 3 seconds faster on orange/red. It's much easier for me to pick up 1 new set of colors to adjust to for the short term and once I become comfortable with quad CN I could decide if I want to go full CN but I won't just switch to CN right away and it could be that 4 choices out of 6 could be enough to take advantage of most of the easier starts to solves.
Also the time it would take to surpass my normal average with Quad CN is faster than if I went full CN. Another thing that is nice is you most likely get extremely close to the benefits of full CN with quad cn. Whereas with dual CN you miss opportunities with 4/6 colors limiting yourself to 2 each time. quad CN lets you take advantage of the best cross most of the time and the chance of the easiest cross being on the 2 colors you didn't adjust to would be a minority instead of the majority like it is with dual CN.
My filter for white/yellow cross is really good compared to orange/red. I still insert into the wrong slot with orange/red but with white/yellow I can see an edge and know instantly where it goes.
So to summarize it is easier to be quad cn than being fully neutral because to be fully neutral you have to adjust to two different opposite color schemes instead of one, decreasing the work required by half.
Last edited:
IRNjuggle28
Veteran
I'm guessing that white pieces are just easier to filter period. There's a reason that most solvers that aren't neutral solve white cross. When I started with the beginners method, I was color neutral, but when I started F2L, the white pieces stood out, so I decided to solve white only. (This was before I'd read anything on how useful being color neutral is) I've since reconsidered that and started solving yellow also. At one point, I realized that I'd solved yellow almost exclusively for a while, and had to restore my white filter because I'd actually gotten faster on yellow than on white. I caught that before it became a problem, and now, I don't have any bias between white and yellow in terms of solving times or recognition. And I'm completely happy with only having 2 cross colors. But anyway, the point I was getting to is that white, and maybe yellow, are just easier to recognize. It would explain the disproportionate number of white cross solvers as well as why you're doing better with white/yellow than with orange/red.
It'll be interesting to hear how quad CN goes! It's fairly untested, as far as I know. I should probably explain better my doubts about it, although what you said does make a lot of sense.
My (completely unsupported) theory for why solving 4 colors might not be a good choice is based on the fact that the benefit you can get from having more crosses to choose from would likely increase at a fairly steady rate. The benefit you would get from solving 2 cross colors instead of 1 would be greater than the benefit from solving 6 cross colors instead of 5, sure, but not that much greater. The cross move count would increase at a fairly steady rate, although the growth would accelerate some. But the color filter number is a different story. (I'm only going to mention the E layer because cross and LL are pretty easy to be color neutral with. It's F2L that's difficult) For Bi-CN, you have one E layer color filter. For quad-CN, you have 2. For full CN, you have 3. But the growth rate in difficulty of that is not even close to as constant. I would theorize that the difference in difficulty between having one (Bi-CN) and two (quad) color filters is greater than the difference between 2 and 3(full CN). The difficult thing seems like it might be having more than one, instead of a steady increase in difficulty as color filters are added. In summary, the relationship of difficulty of recognition and ease of cross solutions will be worse at quad-CN than it will at Bi-CN or full CN.
It is a theory. Nothing more. And you have more experience with neutrality than me, so I'm sure you would know better. But... opinion?
jeff081692
Veteran
I can tell you for sure that the reason I am faster at white/yellow is because I practiced white more. There is research out there that suggests that recognition for perceptual motor skills improves as a direct function of practice. Also the way I practice CN stuff is by doing cross solved scrambles and finishing through F2L. This way I can get the most F2L practice out of my time. I don't have time to do lots of solves so I stick to targeting my weaknesses in addition to a stable amount of full solves each day.
The difficulty of recognition has never really been intimidating for me because I know exactly where I should be looking during F2L and improving lookahead recognition is the only thing holding me back and it's not like I have to learn any new F2L algs or fingertricks to improve that last second like I did with my original colors. Also going steady and looking ahead makes things much easier and I can usually do F2L without pauses aside from some inconsistencies about 1 every 3 solves. Once I get more comfortable with slow lookahead I am sure I would have no problem speeding my turning.
Also for benefits: the benefit of going from 2 to 4 is better than going from 4 to 6 and so that is why doing half the work of being CN by becoming quad CN makes more sense to me, for someone who never started with full CN. I am sure the best method for some is to dive right into CN but that is not my style really.
One thing I have going for me is that I specialize in 3x3 and my goal is extremely ambitious so any little advantage I could achieve would be worth it for me and I imagine only a few others. Because I only do 3x3 I treat the white/yellow as a separate event in my mind until it earns it's place in my normal averages. Even when green/blue is faster than white/yellow there is a point that white/yellow has to cross for me to add it to my solves and I know exactly how fast it should be at all times for my goals.
Also for your theory I can put this another way. For someone who started quad CN and someone who started dual CN the difference in work to go from full CN and quad cn respectively is the same because they are both learning only 1 filter (I don't know if this reflects your point). Now maybe someone who is CN has more work cut out for them because they also have to maintain good recognition with 3 filters and so consistent practice seems more important. But again I think of it in terms of different events and if you practice white/yellow more you get better at white/yellow. I'm actually ok with white/yellow lagging behind blue/green for a while as long as white/yellow is at the point where the ease of the cross evens out the difference between my blue/green solves. At that point when I integrate the colors I keep a consistent average with room for improvement because my white/yellow might converge to blue/green speed.
Either way it will be interesting to see what happens since like you say there aren't many quad CN people but I feel it is the right choice for me.
The difficulty of recognition has never really been intimidating for me because I know exactly where I should be looking during F2L and improving lookahead recognition is the only thing holding me back and it's not like I have to learn any new F2L algs or fingertricks to improve that last second like I did with my original colors. Also going steady and looking ahead makes things much easier and I can usually do F2L without pauses aside from some inconsistencies about 1 every 3 solves. Once I get more comfortable with slow lookahead I am sure I would have no problem speeding my turning.
Also for benefits: the benefit of going from 2 to 4 is better than going from 4 to 6 and so that is why doing half the work of being CN by becoming quad CN makes more sense to me, for someone who never started with full CN. I am sure the best method for some is to dive right into CN but that is not my style really.
One thing I have going for me is that I specialize in 3x3 and my goal is extremely ambitious so any little advantage I could achieve would be worth it for me and I imagine only a few others. Because I only do 3x3 I treat the white/yellow as a separate event in my mind until it earns it's place in my normal averages. Even when green/blue is faster than white/yellow there is a point that white/yellow has to cross for me to add it to my solves and I know exactly how fast it should be at all times for my goals.
Also for your theory I can put this another way. For someone who started quad CN and someone who started dual CN the difference in work to go from full CN and quad cn respectively is the same because they are both learning only 1 filter (I don't know if this reflects your point). Now maybe someone who is CN has more work cut out for them because they also have to maintain good recognition with 3 filters and so consistent practice seems more important. But again I think of it in terms of different events and if you practice white/yellow more you get better at white/yellow. I'm actually ok with white/yellow lagging behind blue/green for a while as long as white/yellow is at the point where the ease of the cross evens out the difference between my blue/green solves. At that point when I integrate the colors I keep a consistent average with room for improvement because my white/yellow might converge to blue/green speed.
Either way it will be interesting to see what happens since like you say there aren't many quad CN people but I feel it is the right choice for me.
Last edited:
PJKCuber
Veteran
I am switching to Roux, so can anybody explain to me the types of Roux CN since I am CN with CFOP? I have no idea what x2/y and x/y2 means.
CriticalCubing
Veteran
Similar threads
