(Re-)considering The Limits

[KJiptner]

At the beginning, the limits of the Fridrich Method (CFOP) were considered to be arount 17secs as you can see on Jessica Fridrichs webpage:
http://www.ws.binghamton.edu/fridrich/system.html
And Zborowski mentioned that the only way of getting 15sec avg is to learn his method:
http://www.zborowski.republika.pl/

Yesterday we've had a new WR: 11.76 Avg of 5 for a 3x3x3 ... with the standard CFOP system.
Are the limits for this method reached?
And what can be considered the limits with a fully mastered ZB now? (if anyone should ever reach this state)

[ArunJangity]

YOU ARE INCORRECT, JIRI SAYS THE LIMIT WAS 10-12 SECOND

http://www.ws.binghamton.edu/fridrich/hints.html#limits

[tenderchkn]

Those limits are severely outdated. The 17 was before optimized algorithms. Fridrich did say that the limits are 10-12 seconds (look at her limits of speedcubing" page).

And she's about right. Macky gave the same estimates on his site (cubefreak.net).

There's really no way to average more than 5 turns per second while during a CFOP solve. If you try to go faster, there's no way you can see the next pair in time. For a ZB solve, you have to go slower during the F2L (see Chris Hardwick's site) because you only insert three pairs instead of 4. The last one has to be done slower. Additionally, the LL algorithms, while more efficient, are nowhere close to being as optimized as the fastest OLLs and PLLs. So for a ZB solve, the limit for turns per second is probably 4.

So if you take that a CFOP solve is a little over 50 turns and a ZB solve is about 40, then 10 seconds/solve is really the limit. But ultimately, the most important factor is the cuber, not the system. Macky was heads and shoulders above everyone else for two years. And now there's another leap in times. The algorithms are probably as optimized as they're going to get, it's just how well one learns to execute them.

[Johannes91]

Originally posted by tenderchkn@Jan 8 2007, 05:16 PM
So if you take that a CFOP solve is a little over 50 turns and a ZB solve is about 40, then 10 seconds/solve is really the limit.

How would you average 40 moves using ZB??

And why do you guys ignore all the other methods? It is indeed possible to average around 40 moves, if you don't build a cross but start solving the pieces directly. I know that most speedcubers disagree with me, but I think 8-9 seconds on average is possible.

[KJiptner]

YOU ARE INCORRECT, JIRI SAYS THE LIMIT WAS 10-12 SECOND

okay, okay. i see. no need to scream around.

sorry, for my carelessnes. but she actually says limits for speed cubing... not cfop... but okay...

we could also consider other methods...
I just took fridrich into consideration because it is the most used (and so the most "tested"/optimized) and ZB because it's the most advanced, i think.

I think 8-9 seconds on average is possible.

limit of 8-9 sec...well seems a little bit humanly impossible at first... but we'll see. time has proven a lot of things.

[tenderchkn]

Originally posted by Johannes91+Jan 8 2007, 06:07 PM--></div><table border='0' align='center' width='95%' cellpadding='3' cellspacing='1'><tr><td>QUOTE (Johannes91 @ Jan 8 2007, 06:07 PM)</td></tr><tr><td id='QUOTE'> <!--QuoteBegin-tenderchkn@Jan 8 2007, 05:16 PM
So if you take that a CFOP solve is a little over 50 turns and a ZB solve is about 40, then 10 seconds/solve is really the limit.

How would you average 40 moves using ZB??

And why do you guys ignore all the other methods? It is indeed possible to average around 40 moves, if you don&#39;t build a cross but start solving the pieces directly. I know that most speedcubers disagree with me, but I think 8-9 seconds on average is possible. [/b][/quote]
That&#39;s what ZB claims on his site. I know that actual CFOP solves are around 60 moves and ZB solves are probably close to 50 for most people, but since we&#39;re talking about theoretical limits, we might as well assume that we have cubers who can solve close to the theoretical optimum. And the point that I was making is that it doesn&#39;t matter what method one uses to solve the cube - the limits remain the same. It is not the method, but the cuber that is the limiting factor.

I don&#39;t mean to leave out Petrus, but my point remains the same. Being completely color neutral makes Petrus a more efficient solving method, but just like ZB, that does no translate into faster speeds. No matter how you do the F2L, you will be limited in some way (cross in Fridrich and the 2x3x3 in Petrus). It&#39;s just the fact that a lot of Petrus solvers choose to be color neutral that makes it more efficient.

[watermelon]

I agree with Johannes on this one. Although 8-9 sec avg might seem unlikely right now, sub-12 avg would have seemed impossible a few years ago, yet multiple people have gotten it.

However, 8-9 sec avg with CFOP does seem very unlikely (but not impossible). I think that an avg of that speed would require another, more efficient method.

[Erik]

Have you considered roux?...
The roux method itself is somwhat about 40 moves. And the last steps are not very much optimized yet.. I see potential...

[Athefre]

I just took fridrich into consideration because it is the most used (and so the most "tested"/optimized) and ZB because it&#39;s the most advanced, i think.

It isn&#39;t the most advanced. I think Heise is much more advanced than ZB. More "algorithms" (and a lot of time spent finding them) doesn&#39;t mean more advanced.

Originally posted by "Erik"
Have you considered roux?...
The roux method itself is somwhat about 40 moves. And the last steps are not very much optimized yet.. I see potential...

Yeah, people leave out Roux even more than Petrus. It&#39;s about 48 moves for most people that use "normal" Roux (same blocks every time). I use Non-matching Roux and use 43 moves on average in speedsolving. If you are color neutral, I think 38 would be a good number for most people. For Heise, it&#39;s the same numbers IMO.

Like Johannes has told people many times, 38 moves at around 4 moves per second means a less than 10 second solve. Andrew Kang, the Unofficial record holder, does 4.6 moves per second on average so if you divide that into 38....

[Me]

I think rather then reconsidering the limits i think that it could be possible that we have reached the limits.
Slowly but surely there are longer periods of times between new records being set. A few years ago new records were set with each competition that came along. Now new records aren&#39;t as common. I think that it&#39;s accurate to say that the Speedcubing community as a whole has reached its limits with most of the puzzles out there. No doubt the single solve records will be broken, but there will eventually be a record that will blow the others away like a lucky sub-10 around 7 or 6.xx. However as far as averages (of 5 in this case) go it seems that the limits have been reached.