• Welcome to the Speedsolving.com, home of the web's largest puzzle community!
    You are currently viewing our forum as a guest which gives you limited access to join discussions and access our other features.

    Registration is fast, simple and absolutely free so please, join our community of 40,000+ people from around the world today!

    If you are already a member, simply login to hide this message and begin participating in the community!

The Rice Method

Cool Frog

Member
Joined
Nov 24, 2009
Messages
1,255
Location
Florida, in that one place...
45.25, 27.30, 54.52, 28.98, 36.99, 44.09, 47.46, 32.83, 40.80, 38.45, 31.19, 33.52

current avg5: 34.93 (σ = 3.07)
best avg5: 34.93 (σ = 3.07)

current avg12: 37.96 (σ = 6.35)

Just did an avg 12. no rolling so pretty meh. Lookahead is pretty terrible.
 

rice

Member
Joined
Mar 8, 2012
Messages
29
I feel your pain. I've started doing solves with the good, old metronome. Keep it up!
 

rice

Member
Joined
Mar 8, 2012
Messages
29
I set the bar low with this one! :p
 
Last edited by a moderator:

rice

Member
Joined
Mar 8, 2012
Messages
29
An interesting theory that should significantly improve times: making the first edge pair and storing it in the D layer, solving the corners without breaking up the pair, inserting the first pair, then solving like normal for the rest. This allows us to maximize utilization of inspection time and therefore speed up look-ahead and execution for the entirety of the solve.
 
Joined
Oct 1, 2010
Messages
361
making the first edge pair and storing it in the D layer, solving the corners without breaking up the pair, inserting the first pair, then solving like normal for the rest.

I remember having read that an Josef Jelinek's site.

Also from theory (and from symetry) after doing the corners step it does not matter wich pair of corners it is, that I start with. But when planng a "pair-prepare" I discribed by you, its is easiest, if both edges belong in the opposite layer ... (not sure that si understandable so I do an )

Exsample:
2a1- Yellow-Blue and Yellow-Green form a edge pair in D
1a - Solve white Corner on D
1b - Solve yellow corner on U
2a2- U°M2 to placve Yellow-Blue Yelow-Green
2b- Continue with White-Blue White-Green
2c - solve white an yellow centers along with on yellow and one white edge
3 - LSE

While at first this looks exactly like what you dicribe, It has the following advanteges to me:

2a2 is really just two moves - ( it might be better to do U° L2R2 D to save cube rotation)
2b I am much more comfortable looking for two U-colored(white in my exsample edges)
2c Is now different from 2b because here now one edge is Yellow and the other white, but I can easily benefit from cases where I solve yellow-red and white-orange..
3- I get LSE with white yellow as R and L wich I prefer.

While the first point (2a2) my really saves moves all the other things should have no influence on movecount, just on the way of lookahead...[maybe wrong because after having done columns you have 4 possibilites vor 2c while I restict myself to just 2 plus 2 (with opposit colors )]

EDIT: The most obvious way to benefit from this I forgot to mention - build 4 corner and two edges of the same layer right at the start.
 
Last edited:

rice

Member
Joined
Mar 8, 2012
Messages
29
Building on your idea, oll+phase+sync, I have come up with the fried rice method! :D
1. Starting with white on D, we pair up white/blue and white/green in the DL/DR position
2. Pair up yellow/blue and yellow/green in the DF/DB position
3. Solve corners based on position of 1st edge pair (if white/blue is in DL, white/blue/red will be in DFL)
4. Insert 2nd pair, align corners
5. Rotate cube so red/orange is on D, then make and insert 3rd pair
6. LSE
 

Kirjava

Colourful
Joined
Mar 26, 2006
Messages
6,121
WCA
2006BARL01
YouTube
Visit Channel
An interesting theory that should significantly improve times: making the first edge pair and storing it in the D layer, solving the corners without breaking up the pair, inserting the first pair, then solving like normal for the rest. This allows us to maximize utilization of inspection time and therefore speed up look-ahead and execution for the entirety of the solve.

I try to pair up the first one during the end of the corners/at the same time, a la EOLine.
 

rice

Member
Joined
Mar 8, 2012
Messages
29
Has anyone tried out fried rice? It should ease look-ahead for the edge pairs.
 
Joined
Oct 1, 2010
Messages
361
Has anyone tried out fried rice? It should ease look-ahead for the edge pairs.
Doing the pseudo cross (no center, and unphased axis ) is really quite easy, and after corners the rest of the solve is really a piece of cake.

But having a cross makes corner anticipation during inspection to hard for me, so I think one should solve at least two corner while crossing.
 

rice

Member
Joined
Mar 8, 2012
Messages
29
It's still too early to see how many edge pairs should be made (0, 1, or 2 pairs) before solving the corners. I'm trying to track multiple pieces around the cube but too much can be mind-blowing, so I haven't undergone any serious attempts yet. Has anyone tried to track several edge and/or corner pieces through a solve?
 

rice

Member
Joined
Mar 8, 2012
Messages
29
Welcome to the 1st Annual Rice Method Revival!

I've been thinking about how to speed up this method, but Cube Explorer isn't really suited for generating and solving ricey cases. Is there any demand for a list of cases? I can cook some up, but some might not be optimal.
 

elrog

Member
Joined
Jan 31, 2013
Messages
518
Location
U.S.A.
YouTube
Visit Channel
Cool. I had an idea for this exact method before. Seems like nearly everything has already been thought of. Concerning speeding this up, doing EG without effecting edges to improve look ahead could be good except for it would vastly increase the number of moves for algorithms. Prahaps you could preserve just enough edges that make it easy to preserve. Even preserving just these edges still limits the possiblitities of where other edges can be. I would suggest doing someing similar to PEG algs. They preserve the middle layer and many can be derived from the standard EG algs by treating the Pairs as a single corner.

This doesn't really partain to this method, but I thought I'd post it here. Could any waterman users tell me if it would be just as good a speedsolving method to solve the corners and then the bottom 4 edges (then proceed as in standard waterman) as it would be do do waterman. I think it could be because all of the corners can be planned out during inspection.
 
Joined
Jul 7, 2008
Messages
627
Location
Saskatchewan, Canada
YouTube
Visit Channel
This doesn't really partain to this method, but I thought I'd post it here. Could any waterman users tell me if it would be just as good a speedsolving method to solve the corners and then the bottom 4 edges (then proceed as in standard waterman) as it would be do do waterman. I think it could be because all of the corners can be planned out during inspection.

In terms of look-ahead, no. At least that's my experience.

@rice: I was wondering if you'd ever be back on the forums. In of movecount, there isn't much you can do. With the exception of EG for corners, this method is pretty well optimized. I guess in terms of speed, the most important things would be developing more thorough strategies for look-ahead.
 

elrog

Member
Joined
Jan 31, 2013
Messages
518
Location
U.S.A.
YouTube
Visit Channel
Well, the inspection is the best of any 3x3x3 methods out there. EG is solving all 8 corners with inspection. I think this could help make up for some of the slack in recognition.

@ somerandomkidmike's next post: Hey, I'm just trying to look at this with an optimistic point of veiw. Aslo, what other method solves 8 peices orientation and permutation with inspection alone?
 
Last edited:
Joined
Jul 7, 2008
Messages
627
Location
Saskatchewan, Canada
YouTube
Visit Channel
Well, the inspection is the best of any 3x3x3 methods out there. EG is solving all 8 corners with inspection. I think this could help make up for some of the slack in recognition.

I don't think the problem with recognition has anything to do with the corners, so there is really no point in presenting a more complicated corners solution as an "improvement". Sure, you might save 2 or 3 moves and improve your recognition for the first step, but you're just going to make it more difficult for the next step.

Either way, I don't know where you're getting the idea that the "inspection is the best of any 3x3x3 methods out there". I'm sure if that was true, the CFL method would be adopted as a main method by many more cubers- specifically people that have tried Roux or other corners-first variations.

The real downfall for using this as a speedsolving method is the first-six edges. The solutions are very elegant for a highly intuitive corners-first method, but it suffers from a moveset that is awkward for the step with the hardest recognition. No amount of optimization for the corners will help with that.
 

qqwref

Member
Joined
Dec 18, 2007
Messages
7,834
Location
a <script> tag near you
WCA
2006GOTT01
YouTube
Visit Channel
Having tried CF methods, I agree, but...

...what if you could do them quickly, somehow? I mean, for the best people, the corners can be done in 2 seconds, and so can L6E. If all we have to do is solve 6 edges and the centers in 6 seconds (to get sub-10), maybe there's a way.
 

elrog

Member
Joined
Jan 31, 2013
Messages
518
Location
U.S.A.
YouTube
Visit Channel
Yes, but the WR is 5. That leaves you 1 second for the first 6 edges. Good Luck!! :p

@ qqref's next post: "6 seconds is a lot of time" Not for all of us my friend...
 
Last edited:

qqwref

Member
Joined
Dec 18, 2007
Messages
7,834
Location
a <script> tag near you
WCA
2006GOTT01
YouTube
Visit Channel
That's a single solve. Nobody averages 5 seconds.

And besides, even if this method isn't going to be WR level, being able to sub-10 it would be pretty impressive. 6 seconds is a lot of time; if we can do that, great, but if we can make 3 seconds possible, that would be even better, and 7 seconds would be possible for the whole cube. It's unlikely, but hey, who knows?

I'm thinking maybe some way to get the 4 E-layer edges first, then a D move and placing DB and DF (really DR and DL), then ADF/L6E.
 
Top