Older cubers discussions

[JohnnyReggae]

One reason I push M2 from the very start is I think some people just getting started will get a few successes with OP edges and stick with it for a while because it's what they know and M2 isn't that much harder, just different.

For Q btw, I suggest (U' M')x3 (U' M) (U' M')x4 or (U M')x3 (U M) (U M')x4, whichever direction is most comfortable for your finger tricks. A lot of older references suggest (B' R B U R2 U') (M2) (U R2 U' B' R' B) which is intuitive but not very execution friendly IMO.

Thanks for that Q alg. It's a lot easier to remember and much quicker to execute

I track edges I've traced on the L or R faces with my fingers and simply try to remember any I've hit on the M slice. The competent 3bld folks just develop a natural sense of when the trace is done. I tried using a checksum instead of fingers early on but I don't think it's a fruitful approach, at least not for me.

I was wondering if there was some golden number given any number of corner swaps or rotations. I'll try your method on L and R, it does seem more efficient than trying to trace all the edges. Thanks

 

[pglewis]

I was wondering if there was some golden number given any number of corner swaps or rotations.

For edges you can use a checksum to 11 by counting those already solved, adding one for cycle breaks I think, and maybe there's something special with flipped edges. It doesn't help me know which are still left when I hit a cycle break though, and that's the primary reason I still trace with my fingers.

 

[chtiger]

Is there a very quick way from looking at the numbers when you are done with memo ? I know it could be anywhere between 9 and 13 typically. Is there an easy way to tell because trying to touch each piece trying to make sure they have all been accounted for is tricky

There's 11 to start with (if 11 are solved, 12th has to be). When I was new to blind, since times weren't a concern, before starting the edge tracing, I'd check all the edges to see how many are flipped or solved. Subtract that number from 11 (and take note of flipped ones). This is the number of targets for the trace. If you start a new cycle, add 1 to the target number. When you're done, add the flipped edges to your memo. And it helps to memo in chunks of 4 (or similar), then it's easier to tell how many you've already done.

 

[JohnnyReggae]

There's 11 to start with (if 11 are solved, 12th has to be). When I was new to blind, since times weren't a concern, before starting the edge tracing, I'd check all the edges to see how many are flipped or solved. Subtract that number from 11 (and take note of flipped ones). This is the number of targets for the trace. If you start a new cycle, add 1 to the target number. When you're done, add the flipped edges to your memo. And it helps to memo in chunks of 4 (or similar), then it's easier to tell how many you've already done.

This is what I was looking for. Something to give me an indication that I am done without having to trace each piece exactly. I do also like Phil's idea of keeping track of L and R edges. I am going to try use both at the same time.

Thanks guys

 

[pglewis]

Finally had a fairly consistent 3x3 session and got my PB Ao50 down in the mid 22s. Next big milestone would be a sub :20 Ao12.

 

[openseas]

This is what I was looking for. Something to give me an indication that I am done without having to trace each piece exactly. I do also like Phil's idea of keeping track of L and R edges. I am going to try use both at the same time.

Thanks guys

Yes, for edges, 11-solved/flipped + cycle break, for corners, 7- solves/twisted + cycle break.

In general, total number of stickers -1 (buffer) - solves/flipped/twisted + cycle breaks.

For flipped cases, trends are breaking into cycles, meaning force a flipped sticker into a cycle and start a new one from the flipped part. 2 edge comms are considered to be faster than one flip alg, most cases.

 

[JohnnyReggae]

Yes, for edges, 11-solved/flipped + cycle break, for corners, 7- solves/twisted + cycle break.

In general, total number of stickers - solves/flipped/twisted + cycle breaks.

For flipped cases, trends are breaking into cycles, meaning force a flipped sticker into a cycle and start a new one from the flipped part. 2 edge comma are considered to be faster than one flip alg, most cases.

J-Perm has just done a video on getting faster with 3bld with a bunch of tips, including dealing with flipped edges and corners ... i'm starting to enjoy this little venture into blind solving...

 

[h2f]

I thought I'll let you know. I like UF as a buffer in 3bld. As well as new memo/exe order CEEC. I got first sub-1 yesterday 59.66. 5 comms for edges and 4 for corners. I have to learn algs.

 

[Tom Joad]

First ever sub15 Ao5. Never thought I’d see the day...

A surge of adrenaline, fist pump of delight and obligatory bad solve to follow.

Funny I hadn’t cubed much for a couple of weeks and it came out of nowhere. Perhaps actually bothering to lube the cube for a change paid off!

I guess my new target is an Ao5 of 14 seconds or close enough...

 

[pglewis]

First ever sub15 Ao5. Never thought I’d see the day...

A surge of adrenaline, fist pump of delight and obligatory bad solve to follow.

Funny I hadn’t cubed much for a couple of weeks and it came out of nowhere. Perhaps actually bothering to lube the cube for a change paid off!

I guess my new target is an Ao5 of 14 seconds or close enough...

View attachment 10547

And here I was happy with a counting low 17 single yesterday lol. Great batch of solves there!

 

[chtiger]

Something kinda crazy happened earlier this week. I had only had three sub-13 3x3 singles ever, with a PB of 12.76. I got a 12.85, which was my 3rd best ever, then 5 solves later got a 12.54. Almost had 2 of my top 4 fastest solves ever in the same average (still would have been just shy of PB avg). First one was OLL skip, second was PLL skip.

 

[JohnnyReggae]

First ever sub15 Ao5. Never thought I’d see the day...

A surge of adrenaline, fist pump of delight and obligatory bad solve to follow.

Funny I hadn’t cubed much for a couple of weeks and it came out of nowhere. Perhaps actually bothering to lube the cube for a change paid off!

I guess my new target is an Ao5 of 14 seconds or close enough...

I've found that after taking a break from my normal practice, my first dozen or so solves back are always pretty good. I've put it down to just being more relaxed and turning slower because I feel a little unsure, and it almost always is a good few solves. Then of course my brain kicks into overdrive and it all falls apart until I can reign myself back again.

 

[One Wheel]

I had a crazy 3BLD success yesterday: I forgot the second half of my edge memo, so I knew that I had executed an even number of targets and didn’t have parity, I went ahead and executed corners then came back and recalled the rest of the edges. Not fast: 6:40, my first two for the WC were in the 4:20-4:30 range DNFs, but it was solved. I really feel like if I could just sit down and do 100 3BLD solves in a week or two I would significantly up my success rate and likely drop my times close to 3 minutes.

 

[efattah]

After another year of untimed solves I tried going back to timed solves. I was definitely rusty under the 'pressure' of high TPS, and it took several weeks of timed solves to get back into the flow, but I was rewarded eventually with a single excellent session with lots of pb's.

From the same late night session:

Ao5 13.93
Ao12 14.45 PB
Ao25 15.04 PB
Ao50 15.76 PB
#solves 139

Generated By csTimer on 2019-07-16​

avg of 12: 14.45

Time List:
1. 12.60 F' U' F2 U F2 U2 L2 B2 D' R2 U B2 U2 L' B2 D' L D2 U' F L
2. 13.71 L2 D' L2 D U2 B2 U' R2 B2 L2 B2 D2 F R B' D B D2 L U F
3. 17.99 R B2 R' U2 L' U' F D R' U2 R D2 R2 D2 R' D2 F2 L2 D2 F
4. 13.70 B2 L2 F' D2 F' R2 F' L2 B2 D2 F D' L R' D2 F2 R F2 D R U
5. 15.12 B R B2 L' F2 R2 U2 R' B2 R' F2 B' R U B2 R2 D2 U R
6. 15.53 U2 F R' D' L2 B2 D2 R2 D' B2 D L2 F2 U' R' B L R B2 R2 D'
7. 14.11 B L2 F L2 F R2 B L2 R2 B' L' U2 F D' F' L2 F U2
8. 14.20 R U2 B2 L2 B2 L' D2 B2 F2 L2 B' D' U' B' D L2 F' R2 U2
9. (11.38) L' D' F' R2 B R2 D2 F D2 L2 B L2 F2 D' U2 R' F2 L' F' D
10. 13.62 F2 L2 U F2 L2 U2 B' L' F2 L2 D2 F' U2 F' R2 B' U2 R2 U2 R
11. (18.01) F2 D2 R F2 D' U2 L2 B2 U' B2 U' R2 F' R2 U B R' F2 D'
12. 13.96 F L' D L2 D2 F2 R2 F2 L2 D B2 R2 F' U' B' L F2 L D F2


Cube was a Gan 356 X.

When I last did timed solves a year ago, I can't recall my exact pb's but I think they were around Ao12 15.40 and Ao50 16.60, so I gained around 1 second off my averages after a year of untimed solves.

The thing that is most noticeable is that it is now possible to make serious errors in the solve and still get 14.XX, and I am now able to get 11.XX with only mildly lucky solves and 12.XX with non-lucky solves. To get sub-11 still requires significant luck and sub-10 solves require very high luck mixed with flawless lookahead. The above set of 139 solves had two 11's and six 12's in the same sitting which also felt very unusual compared to a year ago.

What is disturbing about getting faster is that I noticed that lookahead failures and/or algorithm 'fetch' delays in memory now have utterly dramatic negative consequences. Even pausing for 3/4 of a second has a major negative impact.

What is most frustrating is that even after 3.5 years of speed cubing I still have a number of what I call 'problem-cases.' Algorithm cases where I struggle to fetch the algorithm from memory, struggle to execute it and often screw up, and it seems no amount of drills fixes the problem for those particular 'bad cases.' So far my only explanation is that certain algorithm patterns don't lend themselves to memory and execution and in these cases my best bet is usually to just 'delete' the old algorithm from memory and find a new one which fetches/executes more smoothly, even if the new algorithm is longer and apparently slower than the old one.

 

[Tom Joad]

After another year of untimed solves I tried going back to timed solves. I was definitely rusty under the 'pressure' of high TPS, and it took several weeks of timed solves to get back into the flow, but I was rewarded eventually with a single excellent session with lots of pb's.

From the same late night session:

Ao5 13.93​

Ao12 14.45 PB​

Ao25 15.04 PB​

Ao50 15.76 PB​

#solves 139​

​

Generated By csTimer on 2019-07-16​

avg of 12: 14.45

Time List:
1. 12.60 F' U' F2 U F2 U2 L2 B2 D' R2 U B2 U2 L' B2 D' L D2 U' F L
2. 13.71 L2 D' L2 D U2 B2 U' R2 B2 L2 B2 D2 F R B' D B D2 L U F
3. 17.99 R B2 R' U2 L' U' F D R' U2 R D2 R2 D2 R' D2 F2 L2 D2 F
4. 13.70 B2 L2 F' D2 F' R2 F' L2 B2 D2 F D' L R' D2 F2 R F2 D R U
5. 15.12 B R B2 L' F2 R2 U2 R' B2 R' F2 B' R U B2 R2 D2 U R
6. 15.53 U2 F R' D' L2 B2 D2 R2 D' B2 D L2 F2 U' R' B L R B2 R2 D'
7. 14.11 B L2 F L2 F R2 B L2 R2 B' L' U2 F D' F' L2 F U2
8. 14.20 R U2 B2 L2 B2 L' D2 B2 F2 L2 B' D' U' B' D L2 F' R2 U2
9. (11.38) L' D' F' R2 B R2 D2 F D2 L2 B L2 F2 D' U2 R' F2 L' F' D
10. 13.62 F2 L2 U F2 L2 U2 B' L' F2 L2 D2 F' U2 F' R2 B' U2 R2 U2 R
11. (18.01) F2 D2 R F2 D' U2 L2 B2 U' B2 U' R2 F' R2 U B R' F2 D'
12. 13.96 F L' D L2 D2 F2 R2 F2 L2 D B2 R2 F' U' B' L F2 L D F2


Cube was a Gan 356 X.

When I last did timed solves a year ago, I can't recall my exact pb's but I think they were around Ao12 15.40 and Ao50 16.60, so I gained around 1 second off my averages after a year of untimed solves.

The thing that is most noticeable is that it is now possible to make serious errors in the solve and still get 14.XX, and I am now able to get 11.XX with only mildly lucky solves and 12.XX with non-lucky solves. To get sub-11 still requires significant luck and sub-10 solves require very high luck mixed with flawless lookahead. The above set of 139 solves had two 11's and six 12's in the same sitting which also felt very unusual compared to a year ago.

What is disturbing about getting faster is that I noticed that lookahead failures and/or algorithm 'fetch' delays in memory now have utterly dramatic negative consequences. Even pausing for 3/4 of a second has a major negative impact.

What is most frustrating is that even after 3.5 years of speed cubing I still have a number of what I call 'problem-cases.' Algorithm cases where I struggle to fetch the algorithm from memory, struggle to execute it and often screw up, and it seems no amount of drills fixes the problem for those particular 'bad cases.' So far my only explanation is that certain algorithm patterns don't lend themselves to memory and execution and in these cases my best bet is usually to just 'delete' the old algorithm from memory and find a new one which fetches/executes more smoothly, even if the new algorithm is longer and apparently slower than the old one.

This is reassuring!

I just put my timer away with a plan to do exactly the same thing for a year. Seems like the way forward to me. I noticed immediately that I started enjoying rewatching videos (jperm first, how to be sub20, sub 15, sub12, sub10) There’s one hour of great tips right there and there’s no way I am already acting upon all of it just because I watched it one time years ago.

Also started finding white cross solvers on cubesolv.es and reconstructing their solves.

Within a day I sourced seven f2l algs which I wasn’t using, which I am now drilling, It will take quite some time to incorporate these into my solves. Slow solves seems like the only way to do this.

On your last paragraph: any particular examples? I currently had the idea floating around my head that this was just a recognition problem with pairs of olls for me. I plan to get eight cubes in front of me with the two cases from each of the four angles and then learn the auf-oll as one for each of the eight similar looking cases. Should master my weakest oll recognition this way.

Where are you with first pair prediction in inspection? This is my weakest area and something I hope to focus on in my year of untimed solves.

4 years 3 months cubing for me, just got my Pb Ao5 of 14.7, so not quite up to your standard. I would be delighted to take one second off in one year. *picks up cube*

 

[MarcelP]

Finally had a fairly consistent 3x3 session and got my PB Ao50 down in the mid 22s. Next big milestone would be a sub :20 Ao12.

View attachment 10536

Sub 20 was a big milestone for me.. After that it went all bullocks.. Every now and then I practice for a few days and then I seem to get stuck at 19:xx on my Ao100's.. I guess I have hit my limit.

 

[SpartanSailor]

I do a lot of untimed solves and focus on smooth turning and really try to look ahead a lot. But i don’t think I could gona year without timing myself. I do notice improvements when I do mostly untimed solves. But for me, starting the timer makes me change—I turn frantically and don’t focus on look ahead.

I’m still chasing sub-20, officially. I have a comp in two weeks. I’d love to get a sub-20 ao5 and a new PB single. The following week is US nationals. I don’t expect to do well there based upon my habit of letting my nerves impact my solves. I’ll be on staff too, so, instead of relaxing and getting warmed up, I’ll be working the entire time.

I’m impressed to see people that can regularly get sub-15s. Especially, those that routinely average less! Never mind the even faster times—blows my mind.

 

[efattah]

This is reassuring!

I just put my timer away with a plan to do exactly the same thing for a year. Seems like the way forward to me. I noticed immediately that I started enjoying rewatching videos (jperm first, how to be sub20, sub 15, sub12, sub10) There’s one hour of great tips right there and there’s no way I am already acting upon all of it just because I watched it one time years ago.

Also started finding white cross solvers on cubesolv.es and reconstructing their solves.

Within a day I sourced seven f2l algs which I wasn’t using, which I am now drilling, It will take quite some time to incorporate these into my solves. Slow solves seems like the only way to do this.

On your last paragraph: any particular examples? I currently had the idea floating around my head that this was just a recognition problem with pairs of olls for me. I plan to get eight cubes in front of me with the two cases from each of the four angles and then learn the auf-oll as one for each of the eight similar looking cases. Should master my weakest oll recognition this way.

Where are you with first pair prediction in inspection? This is my weakest area and something I hope to focus on in my year of untimed solves.

4 years 3 months cubing for me, just got my Pb Ao5 of 14.7, so not quite up to your standard. I would be delighted to take one second off in one year. *picks up cube*

After a year of untimed solves, my inspection phase did not improve whatsoever. The reason is simple; during the year of untimed solves I would cube as a relaxing break, and for me, visualizing the pieces during inspection is not relaxing (rather mentally draining), so I would never bother any type of visualization in my 'untimed inspections', hence inspection never improved. Lookahead did improve though. However, some things got worse. Drills are still unquestionably the best method to improve (in my opinion), but drills feel like work so I rarely do them. In large algorithm sets (say 50+ algorithms) the odds are not 1/50 of each algorithm appearing; they are skewed and some rare cases may come up much less. Without drills, 'low probability' cases come up very rarely, and during the year of untimed solves I gradually began to totally 'lose' certain algorithms that appeared rarely. When I started timing myself I had to do some drills and re-learn the now 'lost' algorithms. I use about 300 algorithms and estimate that I 'lost' about 15-20 algorithms over the course of the year that had to be re-learned at the end.

Another strange thing is that there always seems to be a better algorithm. I can search 5-6 top cubers' sites and find what I think are the top 5 algorithms and pick the best one, then a year later I do another search and find other cubers' pages that I didn't find before that have way better algorithms for the same case. I'm not sure if people are just developing new algorithms or if I just didn't find the good ones in the first place.

One thing that was a little frustrating was that over the course of the untimed solves, occasionally I would try a high TPS untimed solve, and occasionally I would get extremely lucky on the solve. I estimate I probably had a dozen or so sub-10 singles during the year, I really wish I knew how fast they were, but I will never know.

 

[pglewis]

But for me, starting the timer makes me change—I turn frantically and don’t focus on look ahead.

One thing that was a little frustrating was that over the course of the untimed solves, occasionally I would try a high TPS untimed solve, and occasionally I would get extremely lucky on the solve. I estimate I probably had a dozen or so sub-10 singles during the year, I really wish I knew how fast they were, but I will never know.

These are the two main reasons I still break out the Stackmat for any practice session of decent length. I usually don't bother to track the times, I'm perpetually in so much flux that averages longer than 100 aren't too useful to me. I'm fairly comfortable ignoring the timer and not letting it change my solves, if I decide to use something new and get a :35 out of it, big deal. At my stage it's the encouragement of "hey, that's like the third :16 I've had today" and of course catching that rare lucky single. Running session, though... I'll let those break me nearly every chance.

 

[pglewis]

Within a day I sourced seven f2l algs which I wasn’t using, which I am now drilling, It will take quite some time to incorporate these into my solves. Slow solves seems like the only way to do this.

I'd love to hear some of the top ones you've dug up.

Still working on "edge in wrong slot / corner in U face" cases here. I can only handle one or two tweaks to my F2L at a time and it does take a lot of solves to get the changes worked into the the routine so that's been my "new thing" for weeks. I've also recently traded triple-sexy for (U R U' R')x3 and equivalent for all for slots, which isn't automatic yet but it feels more ergonomic for me.