Older cubers discussions

[openseas]

Posted this in the wrong thread yesterday.

MBLD PB 7/7 39:00 (29:06)
Been getting annoyed at failures on my 10 cube attempts so went a bit smaller. Still good practice. Went super safe on memo and execution. Previous bests 5/5 and 7/9.

Wow, great!

I have two attempts left this weekend. Last chance for Nats qual.

 

[newtonbase]

Lkkkkxcsrdxv x v have t bgbvc., x v v gnu g

 

[40yearstosolve]

Early on it's all about learning the ropes. Use a scramble generator to get comfortable making arbitrary face turns. It'll be slow at first but you'll also be practicing reading notation, it'll get faster. Finger-trick practice won't hurt a bit at this stage, you're basically learning the alphabet. Especially get comfortable with M, M', and M2 slices, they're incredibly useful and they were awkward for a while for me, personally. Time will melt off surprisingly quickly with practice, the more time you shave off your solves the more solves you'll get during your practice sessions.

I agree with other advice out there to ditch the daisy step early and learn how to solve the first four white edges straight to the bottom. This is an easy habit to break at this stage. As you progress, a lot comes down to what you're able to see and solving the "cross" on the bottom means you're displacing other edges from the bottom that you need to see for the next step.

Next, I would learn F2L instead of doing corners then edges. If you pay close attention to exactly what happens when you do step 4 you'll see that you remove a corner you'd solved earlier, pair it with the edge you're about to insert, then put them in place together. F2L just connects a corner/edge and solves them as a single entity, cutting your work in half. If things are still a little dodgy doing corners first then I'd say you're fine sticking with it long enough that it's reasonably comfortable... but the moment it is reasonably comfortable, ditch it for F2L. It's a dead-end with zero advantage other than being quicker/easier to learn.

Step 5 is called "OLL" (orient the last layer) and Step 6 "PLL" (permute the last layer). You can get a lot faster without changing what you're doing here. I would add another PLL whenever your current set of algs is comfortable and not at risk of being forgotten. For OLLs, most people first focus on the 7 "OCLL" algs that just deal with the cases where all 4 top edges are already oriented. No big rush in my opinion as the first two layers is likely to be the biggest bottleneck for a while.

I just accepted OLLs and PLLs as "magic" for a while. Somewhere along the way things started clicking more intuitively, as my 3D IQ improved and I did a lot of F2L practice. A lot of OLLs are just taking a pair or pairs out and reinserting them in different ways (trace what Sune does with the pair in the front-right slot). Whether or not I understand what a particular OLL or PLL does I still treat them as magic without thinking. For solving quickly you want to reach a state of "don't think, just do".

Superb advice pglewis thank you. I must admit I had a bad few days this week, where I totally forgot when to use the correct last two algos - it's like I went into a mental black hole. But just as odd I came out of that and realised what I was doing again last night.

One of the turn sequences I now do without really knowing what I am doing - so I used to know them as notation, but now I just turn them.

I am almost scared to change any part of what I know for fear of confusing myself and am tempted to retain what I do - even the daisy - until I can almost do my current method without thinking.

The bad habbit I have got into already is that the last four sequences, for the last layer, I have resorted to memorising just two of them - they are performed anti-clockwise/clockwise, but I have just memorised the clockwise version which I must repeat twice if for example it should have been the anti-clockwise that was performed. I know it's not ideal, but perhaps later I will memorise the others.

I will have a proper read of F2L which I remember reading is part of CFOP --I read the Wiki on this site which was amazingly detailed so need to revisit it and digest.

Is completing the first white square part of CFOP, or is there a separate sequence for that? As I understand it you don't complete that first side using an algo, it must be a technique because at that stage there are too many possibilities.

And I must remind myself that I didn't used to be able to complete the cube. I now can, and fairly consistently and most importantly without reading my notes. And recently have gone from 5 minutes to sub 3 minutes.

I reckon I solve somewhere between 10 and 30 times a day so if I keep this up surely it will begin to solidify in my mind.

Thanks again pglewis you have helped me reawaken and get reinspired, cheers.

 

[newtonbase]

Lkkkkxcsrdxv x v have t bgbvc., x v v gnu g

Apologies for nonsense post. My son got hold of the phone. He's also gained be a couple of new friends on Facebook

 

[pglewis]

I must admit I had a bad few days this week, where I totally forgot when to use the correct last two algos - it's like I went into a mental black hole. But just as odd I came out of that and realised what I was doing again last night.

One of the turn sequences I now do without really knowing what I am doing - so I used to know them as notation, but now I just turn them.

I got into speedsolving about two years ago and most of my PLLs are now fully in muscle memory. I can't transcribe the moves without breaking down what I'm doing slowly, which is difficult. Most of them I can't do one-handed for that reason. I have difficulty executing most of them unless I'm actually looking at the case they solve in a real-time solve, my muscle memory is triggered by the visual recognition (and I don't drill my algs like I should). It's a funny thing but familiar to anyone who has learned to play an instrument competently.

I recently went through a spell where I started doing one of my G-Perms wrong out of the blue. It took me a few days of repeatedly making the same mistake to figure out which one it even was. Thinking is so removed from the equation for that step that I often forget or didn't even register what PLL I just used. But there was a time where I needed a cheat sheet and/or had to remember the individual moves.

I am almost scared to change any part of what I know for fear of confusing myself and am tempted to retain what I do - even the daisy - until I can almost do my current method without thinking.

I still say go ahead and bite the bullet on this one. I'll wager that within a week you'll be just as comfortable without the daisy step as you were with it though, like all new things you incorporate, it'll slow you down at first.

The bad habbit I have got into already is that the last four sequences, for the last layer, I have resorted to memorising just two of them - they are performed anti-clockwise/clockwise, but I have just memorised the clockwise version which I must repeat twice if for example it should have been the anti-clockwise that was performed. I know it's not ideal, but perhaps later I will memorise the others.

I don't think this is such a bad thing for the moment and I'm a proponent of using a cheat sheet for as long as needed when you tackle OCLL or new PLL algs. The vast majority of your time is likely tied-up solving the first two layers still. Extra steps in the last layer isn't a big cost right now so you're fine with that and willing to be lazy. As you progress, self preservation will kick in as those extra steps do start to matter more, you get tired of consulting a cheat sheet, and just plain ol' repetition solidifies them. My primary focus with a mid 20s average is still improving F2L and my last layer has mostly just naturally progressed from doing solves and incrementally adding algs as I grew tired of extra steps.

I will have a proper read of F2L which I remember reading is part of CFOP --I read the Wiki on this site which was amazingly detailed so need to revisit it and digest.

Is completing the first white square part of CFOP, or is there a separate sequence for that? As I understand it you don't complete that first side using an algo, it must be a technique because at that stage there are too many possibilities.

I believe what you're doing now is a form of "layer by layer" (LBL), and it's effectively CFOP already IMO... you just turn every F2L case into the single "corner solved, edge in the top layer" case by solving corners first in an extra step. CFOP solves the 4 bottom edges first (cross), then solves the the rest of the first two layers as 4 columns, or "slots", by pairing a bottom corner with a middle-layer edge and putting them in place as a pair.

On a solved cube with yellow top and green front, do R U R'. That will take out the green/orange pair from the front right slot and place it in the top layer. Put the pair back in via R U' R'. That's pretty much the easiest case, you've just solved an F2L pair. There are a lot of F2L cases but they can be generalized some and solved intuitively; you don't have to memorize the 42 cases listed on the wiki in order to get started. Eventually you do want to treat them like "algs" in the sense that you just do them automatically but like all things, that'll come with practice.

And I must remind myself that I didn't used to be able to complete the cube. I now can, and fairly consistently and most importantly without reading my notes. And recently have gone from 5 minutes to sub 3 minutes.

I reckon I solve somewhere between 10 and 30 times a day so if I keep this up surely it will begin to solidify in my mind.

As you approach 2 1/2 mins you're getting twice as many solves during your allotted practice time than when you were at 5 mins. The extra practice makes you faster, which in turn gives you more solves... it can snowball.

 

[newtonbase]

Nicely done. Seems like going from 7 to 10 would be a pretty big jump.

It's not as big as you might think. The biggest issue is time. I've gone slightly over the hour 3 times in 11 attempts at 10 cubes and they were just bad days where I'd have struggled on anything.

 

[Mike Hughey]

Apologies for nonsense post. My son got hold of the phone. He's also gained be a couple of new friends on Facebook

I'm happy to hear this. After I saw your post and that One Wheel liked it, I was afraid I was just missing out on some new method of communication among cubers, so I was feeling old, even on this thread.

It's not as big as you might think. The biggest issue is time. I've gone slightly over the hour 3 times in 11 attempts at 10 cubes and they were just bad days where I'd have struggled on anything.

I agree that jumping from 7 to 10 wasn't that hard - at least for me. On the other hand, increasing upward from 10 has been quite challenging for me. I've been averaging about 1 cube per year that I've actively tried (although it seems like I'm going through a bit of a jump in ability lately, which is nice).

 

[megagoune]

Fascinating. Do you download each timing and collate them all manually?

I just keep using the same session in cstimer. I am now at 5000 solves (maybe 4 to 5 months). From the website you can export all the times to a csv file at once. I then open the file in excel to create the chart.

 

[pglewis]

Apologies for nonsense post. My son got hold of the phone. He's also gained be a couple of new friends on Facebook

My initial thought was "what the heck letter/memo scheme is that?"

 

[h2f]

Apologies for nonsense post. My son got hold of the phone. He's also gained be a couple of new friends on Facebook

I liked the "Lkkkkxcsrdxv x v have t bgbvc., x v v gnu g" post. Seemed very funny.

 

[newtonbase]

I agree that jumping from 7 to 10 wasn't that hard - at least for me. On the other hand, increasing upward from 10 has been quite challenging for me. I've been averaging about 1 cube per year that I've actively tried (although it seems like I'm going through a bit of a jump in ability lately, which is nice).

I'd hope to improve by more than a cube a year but it has been a year since I first got 5 points!
I really need to work out how to get my letter pairs into Anki. The instructions I found were bewildering.im not doing it a pair at a time.

 

[40yearstosolve]

I got into speedsolving about two years ago and most of my PLLs are now fully in muscle memory. I can't transcribe the moves without breaking down what I'm doing slowly, which is difficult. Most of them I can't do one-handed for that reason. I have difficulty executing most of them unless I'm actually looking at the case they solve in a real-time solve, my muscle memory is triggered by the visual recognition (and I don't drill my algs like I should). It's a funny thing but familiar to anyone who has learned to play an instrument competently.

I recently went through a spell where I started doing one of my G-Perms wrong out of the blue. It took me a few days of repeatedly making the same mistake to figure out which one it even was. Thinking is so removed from the equation for that step that I often forget or didn't even register what PLL I just used. But there was a time where I needed a cheat sheet and/or had to remember the individual moves.

I still say go ahead and bite the bullet on this one. I'll wager that within a week you'll be just as comfortable without the daisy step as you were with it though, like all new things you incorporate, it'll slow you down at first.



I don't think this is such a bad thing for the moment and I'm a proponent of using a cheat sheet for as long as needed when you tackle OCLL or new PLL algs. The vast majority of your time is likely tied-up solving the first two layers still. Extra steps in the last layer isn't a big cost right now so you're fine with that and willing to be lazy. As you progress, self preservation will kick in as those extra steps do start to matter more, you get tired of consulting a cheat sheet, and just plain ol' repetition solidifies them. My primary focus with a mid 20s average is still improving F2L and my last layer has mostly just naturally progressed from doing solves and incrementally adding algs as I grew tired of extra steps.



I believe what you're doing now is a form of "layer by layer" (LBL), and it's effectively CFOP already IMO... you just turn every F2L case into the single "corner solved, edge in the top layer" case by solving corners first in an extra step. CFOP solves the 4 bottom edges first (cross), then solves the the rest of the first two layers as 4 columns, or "slots", by pairing a bottom corner with a middle-layer edge and putting them in place as a pair.

On a solved cube with yellow top and green front, do R U R'. That will take out the green/orange pair from the front right slot and place it in the top layer. Put the pair back in via R U' R'. That's pretty much the easiest case, you've just solved an F2L pair. There are a lot of F2L cases but they can be generalized some and solved intuitively; you don't have to memorize the 42 cases listed on the wiki in order to get started. Eventually you do want to treat them like "algs" in the sense that you just do them automatically but like all things, that'll come with practice.



As you approach 2 1/2 mins you're getting twice as many solves during your allotted practice time than when you were at 5 mins. The extra practice makes you faster, which in turn gives you more solves... it can snowball.

Amazing post, so much motivation for me there thank you for all of that it is hugely appreciated, MASSIVELY so!

So, after I replied to your earlier post I immediately researched the cross without the daisy and following the principles I found from this kind chap on Youtube. And now I can’t even remember how I did the daisy - in fact I didn’t have a technique for the daisy, I just used to do it instinctively (which is to say ham fistedly). So now doing the cross as my first step I at least have a method I now follow. I’ll do some timings in the morning when the beer has worn off

—hoping these are good principles to follow, I at least understand these even if they aren’t the best.

Thanks again for the reply, I’ll be re-reading it several times.

Have a great weekend, cheers

 

[mark49152]

I agree that jumping from 7 to 10 wasn't that hard - at least for me. On the other hand, increasing upward from 10 has been quite challenging for me. I've been averaging about 1 cube per year that I've actively tried (although it seems like I'm going through a bit of a jump in ability lately, which is nice).

I agree, except with a different limit. I think adding more cubes is quite easy while maintaining the same comfortable pace, up until you hit the time limit. After that, adding more cubes means going faster, and that is the hard part. I am finding it difficult to push MBLD memo pace without seriously losing accuracy.

 

[Mike Hughey]

I agree, except with a different limit. I think adding more cubes is quite easy while maintaining the same comfortable pace, up until you hit the time limit. After that, adding more cubes means going faster, and that is the hard part. I am finding it difficult to push MBLD memo pace without seriously losing accuracy.

Yes, I guess that's really it. I hit my easy limit at 10 cubes, way back in 2009. After that, it was a slow creep up from there. Except I just recently had a sudden jump, from where 14 cubes was my absolute limit to where 16 cubes is my absolute limit, very suddenly. And the encouraging thing for me is that my 3x3x3 BLD average speed is currently about 30 seconds slower than my fastest speed was when I was at my peak, so I'm hopeful that getting my regular 3x3x3 BLD speed back up would be enough to get me to 18 or 19 cubes - maybe even give me a shot at 20.

 

[SpartanSailor]

I'd hope to improve by more than a cube a year but it has been a year since I first got 5 points!
I really need to work out how to get my letter pairs into Anki. The instructions I found were bewildering.im not doing it a pair at a time.

How do points work for MBLD?

 

[Duncan Bannon]

1 point for each success and negative point for each unsolved cube. So 9/10=8 or 4/7=1

 

[SpartanSailor]

So, 2/4 = 0 and 1/2 = 1?

So in theory, 3/3 for 3pts would beat 3/5 or even 5/8 because those only yield 2pts?

 

[Mike Hughey]

So, 2/4 = 0 and 1/2 = 1?

2/4 = 0, yes.
1/2 actually is DNF. It would be 0 (1 solved minus 1 unsolved equals 0), but we figure that means "you didn't solve multiple cubes blindfolded", so it's a DNF.

So in theory, 3/3 for 3pts would beat 3/5 or even 5/8 because those only yield 2pts?

Yes. Except 3/5 is 3 -2 = 1 point.
5/8 is indeed 5 - 3 = 2 points.

 

[Duncan Bannon]

If I'm correct, a 0 points would be a DNF. And 3/3 is better than a 3/5 and 5/8.

E-Mike Hughey got you

 

[SpartanSailor]

2/4 = 0, yes.
1/2 actually is DNF. It would be 0 (1 solved minus 1 unsolved equals 0), but we figure that means "you didn't solve multiple cubes blindfolded", so it's a DNF.


Yes. Except 3/5 is 3 -2 = 1 point.
5/8 is indeed 5 - 3 = 2 points.

1/2 = DNF makes sense... and sorry for the bad math on 3/5.