A method of CFOP speedcubing training that yields systematic progress

[TP]

6 weeks ago vs today:

NEW:

What I have learned since last time is the G-perms, otherwise just cross training and slow solving F2L besides regular timed solves.

1s improvement on the cross, 2s improvement on finding and insert first F2L pair and 2,5s improvement on PLL.

Nice to see that the cross training have helped, still an area to improve though since I still can´t execute it blind every time. A bit surprising that it seems to be improvement only in finding the first F2L pair, but no difference in execution. I haven´t learned anything new except the sledgehammer/hedgehammer, but still I thought I was finding pairs easier now. No surprise that PLL goes faster, G-perms come up a lot and for the rest of the PLL algs I have improved on average 0,5s in execution. My PLL execution average is now 2,95s.

Roughly 6s improvement on 6 weeks, I am content with that.

I gotta say that you where spot on in your analyse Speedpicker and that your post helped me. I haven´t followed all the tips (metronome isn´t anything that suits me) but it was helpful and very interesting to read and I can still use it to improve more.

How does the OLL/fingertrick project go?

 

[speedpicker]

Thanks for asking about the OLL project, its up to 119 pages! Its become a real magnum opus, I hope people find it useful. Im going to upload it as a pdf and mobi for ereaders and other electronic devices so people can have it on their phones, tablets or kindles or whatever. I just need a few more solid hours to finish it off but work commitments have stalled it somewhat. Keep tuned, it will be out soon(ish)!

 

[Methuselah96]

Is there anyone still working on a program to implement this style of training?
If not, I would be willing to take up the challenge to base it off of Prisma Puzzle Timer as a fluent Java programmer.
PM me if this is still available.

 

[TheNextFeliks]

Can you give me the spreadsheet and explain how to use it please?

 

[speedpicker]

The spreadsheet can be found in the thread here: http://www.speedsolving.com/forum/showthread.php?39406-A-method-of-CFOP-speedcubing-training-that-yields-systematic-progress&p=805512&viewfull=1#post805512

I'm fairly sure how to use it is explained pretty thoroughly in the thread. Is anything there unclear?

 

[TheNextFeliks]

The spreadsheet can be found in the thread here: http://www.speedsolving.com/forum/showthread.php?39406-A-method-of-CFOP-speedcubing-training-that-yields-systematic-progress&p=805512&viewfull=1#post805512

I'm fairly sure how to use it is explained pretty thoroughly in the thread. Is anything there unclear?

Sorry I didn't know it was there. It is explained well.

 

[TheNextFeliks]

Two questions:
For it says f2l and first pair. Should this be cross and first pair?
Second. For "full f2l" should this include cross?

 

[speedpicker]

Oops, small error there, apologies. Yes it should be cross and first pair, and yes full f2l is including cross, so cross and four pairs.

 

[ketchuphater999]

Is there anyone still working on a program to implement this style of training?
If not, I would be willing to take up the challenge to base it off of Prisma Puzzle Timer as a fluent Java programmer.
PM me if this is still available.

I am working on an application, for Mac. I assume people would like it if you did make a Java version, as it runs on both Mac and Windows.

Thread for the mac one: http://www.speedsolving.com/forum/showthread.php?41766-TurboTimer-prototype-versions!

 

[PianoCube]

I decided to try this out.
I got the individual times by filming an average of 12 and then watched the video carefully to find all the times I needed.


The "baseline solve" is actually 18.42, but I wrote in the wrong time and didn't notice before now.

Seems like I should focus on cross and PLL recognition + auf. I'm able to do the cross a bit faster, but that would make 1st pair recognition worse.
Maybe it's time to try to learn 2 sided PLL recognition.

Spoiler: All the PLLs

It said "Do 7 clean executions of each", so I didn't count those with much lock ups or mistakes. G and N are those I find most difficult to do "clean".
I should practice some more G Perms.

 

[kp]

Is it just me, or is there nothing inside the spoiler?

 

[Methuselah96]

Which spoiler? If you're talking about the spoiler in the previous post, there is something inside the spoiler.

 

[kp]

Yeah, sorry. I was on an older computer, so for some reason the images weren't being displayed properly.

 

[BaconCuber]

Very good post! You have amazing writing skills! I think an app would be very great, even if it happened to be around $10. I am not a programmer, sadly, but hope you do find one soon.

 

[speedcuber50]

I am a programmer.

I cannot yet produce decent programs in C, but there is a very handy language called BBC Basic for Windows which I'm quite an advanced user of. It makes standard windows applications, inditinguishable from those made in other languages.

Contact me if you're interested.

 

[Jorghi]

I'm making the app right now.

 

[ilikecubing]

Oops,sorry I think I removed it by mistake while editing an error

Here it is,

I)Average of 12 = 14.18

PLL,all are averages of 12
1) A1 - 2.20
2) A2 - 1.87
3) E - 2.94
4) U1 - 1.5
5) U2 - 1.95
6) H - 2.01
7) Z - 2.30
8) J1 - 1.99
9) J2 - 1.91
10) T - 1.75
11) R1 - 2.46
12) R2 - 2.14
13) F - 2.76
14) V - 2.33
15) Y - 2.32
16) N1 - 2.82
17) N2 - 2.98
18) G1 - 2.22
19) G2 - 2.64
20) G3 - 2.27
21) G4 - 2.72
= 48.08

II) 48.08/21 = 2.289(Average PLL execution time)

III) 1.83 = Cross only (average of 12)

IV) 3.59 = cross + F2L1 (average of 12)

v) 5.55 = Full LL including recog/AUF (average of 12)

VI) 9.18 = Full F2L including cross (average of 12)

VII) 2.62 = Full OLL including recog and AUF (average of 12)

VIII) 2.87 = Full PLL including recog and AUF (average of 12)

Also,a video if it helps http://www.youtube.com/watch?v=pRawGiIa_ZE&list=UUAL7oScVnmrW7UcCjJdV_aQ

Posting an update of my progress after about 8 months, I have improved by a fait bit on cross and average PLL execution time. Recognition still needs work though

12th July,2013

I)Average of 100 = 14.11 [Best average of 12: 13.31)

1) A1 - 1.79
2) A2 - 1.58
3) E - 2.65
4) U1 - 1.44
5) U2 - 1.42
6) H - 1.33
7) Z - 1.95
8) J1 - 1.78
9) J2 - 1.33
10) T - 1.45
11) R1 - 2.26
12) R2 - 1.82
13) F - 2.09
14) V - 2.19
15) Y - 1.96
16) N1 - 2.04(R U R' D)
17) N2 - 2.39
18) G1 - 1.99
19) G2 - 1.83
20) G3 - 1.93
21) G4 - 1.89
=39.11



II) 39.11/21 = 1.86(Average PLL execution time)

III) 1.51 = Cross only (average of 12)

IV) 3.51 = cross + F2L1 (average of 12)

v) 5.16 = Full LL including recog/AUF (average of 12)

VI) 9.09 = Full F2L including cross (average of 12)

VII) 2.42 = Full OLL including recog and AUF (average of 12)

VIII) 2.95 = Full PLL including recog and AUF (average of 12)

@speedpicker any thoughts/analysis/suggestions?

 

[Dapianokid]

Can this be adapted to Petrus?

 

[speedpicker]

Tricky to adapt to Petrus as there just arent enough really fast Petrus solvers to make a meaningful sample. Same problem with Roux, you just end up taking 5BLDs splits as "correct" (because no-one else is in the same league with this method) where in fact even he will tell you he has areas for improvement. Using CFOP I could take a wide sample of different cubers with different strengths and styles who are all world class.

 

[Hypocrism]

This highlighted that my PLLs were much worse than I thought, and my F2L was proportionally better than I thought! Time to start drilling PLLs instead of timing solves

At the same time, I have about 6 more OLLs to add before it's finally done. Only took 3 years to learn full CFOP :/

For F2L (I still want to do quite a bit of lookahead training-I feel it's still weak) is it better to vary the metronome times or stick with times that are pushing your boundaries? I've been starting slow every time (60BPM rapidly going up by 20BPM in 5 min chunks to about 160BPM or my limit) so that I practise the basic skills, but this might not be useful or an efficient use of time.