Speedcubing - What are the limits?

[Deleted member 36445]

Do I really need to say anything about this or does it speak for itself?

i checked the facts and magnets weren't even invented until 2015

 

[Waffles]

I was looking through the predictions. WHY ARE THEY ACTUALLY SO ACCURATE?????

 

[DuckubingCuber347]

i checked the facts and magnets weren't even invented until 2015

"You lie" - Sun Tzu, Art of War

 

[Deleted member 36445]

View attachment 16014
"You lie" - Sun Tzu, Art of War

"he never said that and even if he did, it isn't true." - a liar

 

[Megaminx lover]

My predictions:
(single)
2x2 - 0.4
3x3 - 3.9
OH - 7.5
4x4 - 19.0
5x5 - 42.0
6x6 - 1:25.0
7x7 - 2:10.0
3x3 BLD - 18.0
4x4 BLD - 2:00.0
5x5 BLD -4:30.0
Pyra: 1.30
Mega: 35.0
WF: 20.0
FM: 16
Square1: 6.5
Clock: 3.0
3x3 MBLD - 45/45


(average)
2x2 - 1.3
3x3 - 4.9
OH - 8.5
4x4 - 25
5x5 - 48
6x6 - 1:25
7x7 - 2:05
Pyra: 2.0
Mega: 43.0
WF: 26.0
Square1: 10.0
Clock: 4.5

The estimates from 8 years ago- whether they have been met or not
2x2 - 0.4- No (?)
3x3 - 3.9- Yes (2018)
OH - 7.5- Yes (2015)
4x4 - 19.0- Yes (2018)
5x5 - 42.0- Yes (2016)
6x6 - 1:25.0- Yes (2017)
7x7 - 2:10.0- Yes (2017)
3x3 BLD - 18.0- Yes (2017)
4x4 BLD - 2:00.0- Yes (2016)
5x5 BLD -4:30.0- Yes (2017)
Pyra: 1.30- Yes (2017)
Mega: 35.0- Yes (2016)
WF: 20.0- Yes (2018)
FM: 16- Yes (2019)
Square1: 6.5- Yes (2017)
Clock: 3.0- No (Only 2021)
3x3 MBLD - 45/45- Yes (2018)
Wow a lot of these have been broken by 2019.

 

[Cubing Forever]

Didn't some people do some tests showing that using average movecounts for certain methods (CFOP, Roux, ZZ, etc) and I specifically remember them saying that sub4 on CFOP (60-65 average HTM) would actually start to melt the plastic?

i.e., ~12+ tps = no more plastic cube

uhh..idk if this is true anymore I guess(sub 4 fullsteps are possible and exist)

 

[Yepala]

This is gold

 

[abunickabhi]

In BLD solves I believe sub-20 will be posible if there will be developed 5cycle method to solve 4 pieces at once.

Sorry for the massive bump, but I would like to add a clarification to this, that only single buffer 3-style is enough to achieve sub-20 in 3blind. The hardware and fingertricks are excellent now so the challenge of sub-20 is very much achievable.

My personal comment is that UF5 would be required for global sub-10 in 3BLD with splits of 3+7.

 

[abunickabhi]

Let's all just remember that an optimal 5-cycle is likely the same number of moves as two 3-cycles perhaps with a cancellation.

This is completely false statement.

The average movecount of UF5 is 10.4 STM (speed optimal and not move-optimal ; also the algs can be decomposed to comms).
The summation of two 3-style algs is 17 moves STM, so even with 2 move cancellation, we are no where near the limits of UF5.

Also I agree with 2025 hardware, 10+ TPS is possible with UF3 and FF 3-style edges, but I do think with a little bit more development of the partial UF5 algset, 10+ TPS would be achievable on all of the algs. The recall of these algs and the fluidity of execution is one major practical factor stopping this method from becoming viable ever, unless someone does deep practice with it. Also, please note that UF5 is used along with UF3 in solves, so it is not a complete replacement to 3-style.

 

[Deleted member 36445]

No lower than sub 6.

hmmm

 

[Caleb Valenzuela]

I feel that sub-4 is the limit because i don’t think consistent sub-3s are possible

 

[Chicken-Cuber]

I feel that sub-4 is the limit because i don’t think consistent sub-3s are possible

No way cubing hardware has not advanced to the max yet. The ZB method is not even close to its max potential. Also in the future we will probably use smart cubes which will let us be faster. So I believe that sub 2 is the absolute limit. Humans can never beat computers though.

 

[Chicken-Cuber]

Didn't some people do some tests showing that using average movecounts for certain methods (CFOP, Roux, ZZ, etc) and I specifically remember them saying that sub4 on CFOP (60-65 average HTM) would actually start to melt the plastic?

i.e., ~12+ tps = no more plastic cube

Sorry for triple post but I wanted to comment on this, ummmmm let’s just say robot turn cube doesn’t melt it

 

[Caleb Valenzuela]

No way cubing hardware has not advanced to the max yet. The ZB method is not even close to its max potential. Also in the future we will probably use smart cubes which will let us be faster. So I believe that sub 2 is the absolute limit. Humans can never beat computers though.

i actually disagree. Cubing hardware for 3x3 is pretty much maxed out already. Even if they added more features to “improve” the performance, the best it’s going to do is shave off a few milliseconds . The best you can do for ZB recog is 0.3 seconds. Sub-1 F2L is nearly impossible (unless you get some yiheng 1.05 scramble officially) and ZB execution would take around 1-1.5 seconds. Unless we are talking about singles, a sub-2 average will not be possible.
So let’s do some math with ZB:
A less ambitous estimate for F2L would probably be 1.5 secs - 2 secs. Then let’s say ZB solvers manage to recognize every ZB in 0.3 seconds. Then execution would take about 1-1.5 seconds. that would equal to about 2.8 seconds. But of course, if you’re turning at Yiheng’s speed, lockups are pretty much inevitable so you need to account the very few milliseconds for that. So all in all an accurate “maxed out” ZB user would be able to solve a cube in low-3s.

We even have to consider the fact that the reason we are even close to sub-3 is because every scramble had:
Rotationless F2L
Atleast one skip (if you count xuanyi’s 3.05 as a PLL skip)
Little to no lock-ups
Low movecount
Sub-2 was literally only possible because they were done on smart cubes (with absurdly easy scrambles)

 

[OtaMota]

Sub-2 was literally only possible because they were done on smart cubes

Yiheng recently got a non-smartcube scram sub-2 single

 

[Caleb Valenzuela]

Yiheng recently got a non-smartcube scram sub-2 single

where did you find it

 

[OtaMota]

where did you find it

It happened months ago lol

 

[Caleb Valenzuela]

It happened months ago lol

LL layer skip lol still proves that you need an insane amount of luck to get sub-2 officially with a really lucky scramble

 

[ruffleduck]

i actually disagree. Cubing hardware for 3x3 is pretty much maxed out already. Even if they added more features to “improve” the performance, the best it’s going to do is shave off a few milliseconds . The best you can do for ZB recog is 0.3 seconds. Sub-1 F2L is nearly impossible (unless you get some yiheng 1.05 scramble officially) and ZB execution would take around 1-1.5 seconds. Unless we are talking about singles, a sub-2 average will not be possible.
So let’s do some math with ZB:
A less ambitous estimate for F2L would probably be 1.5 secs - 2 secs. Then let’s say ZB solvers manage to recognize every ZB in 0.3 seconds. Then execution would take about 1-1.5 seconds. that would equal to about 2.8 seconds. But of course, if you’re turning at Yiheng’s speed, lockups are pretty much inevitable so you need to account the very few milliseconds for that. So all in all an accurate “maxed out” ZB user would be able to solve a cube in low-3s.

We even have to consider the fact that the reason we are even close to sub-3 is because every scramble had:
Rotationless F2L
Atleast one skip (if you count xuanyi’s 3.05 as a PLL skip)
Little to no lock-ups
Low movecount
Sub-2 was literally only possible because they were done on smart cubes (with absurdly easy scrambles)

Okay you might want to consider that future cubers are one-looking scrambles though. We already have Chinese kids two looking (and even one looking) some solves. It just needs to be consistent. Execution of ZBLL definitely is not 1.5s (that is slower than 10TPS), I think 1s flat average is very reasonable. Sub 3 avgs is theoretically feasible but it would be several years in the future.

 

[Caleb Valenzuela]

Okay you might want to consider that future cubers are one-looking scrambles though. We already have Chinese kids two looking (and even one looking) some solves. It just needs to be consistent. Execution of ZBLL definitely is not 1.5s (that is slower than 10TPS), I think 1s flat average is very reasonable. Sub 3 avgs is theoretically feasible but it would be several years in the future.

i guess if cubers could predict ZBLL it can be possible but sub-2 imo just sounds really far-fetched