Home - Cube Corner - Blindsolving
	Single algorithm 3x3 blindsolving

News: This page is somewhat obsolete, since I've improved my system a lot (it's now easier and more powerful). I just leave the old page online in case you're interested. For actual use, I recommend my new method.

First let me clarify that this is not about solving the whole cube by applying some algorithm once, but rather about how to solve it using the same single algorithm multiple times.

I don't yet use this, the main objective for writing about it is that I want to show it's possible to solve the cube (blindfolded) using only a single algorithm. I've read on stiff_hands's page that five algorithms are required, but one is enough as well. Actually of course you don't need to know any algorithms if you can invent what you need on the fly, e.g. with simple intuitive commutators doing 3-cycles (that's how I solved the Megaminx blindfolded).

The general idea is to solve the cubies one by one. This way you will not have parity problems and we'll also do orientation and permutation together at once. Usually you have to apply the algorithm around 25 times. Not blindfolded I can solve the cube in a little less than three minutes using this method.

My first approach (the 528 bytes program) for Tom's CubeSolver Contest uses basically the same idea, but it uses four main algorithms (orient/permute edges/corners).

To demonstrate it I'll use a scrambling algorithm from the Sunday Blindfolded Rubik's Games Contest:

F' L2 F' U2 D R D U' L' B' F2 L2 U2 L2 B2 F2 R2 U2 D2 L2 U2 L2 F2 U D'

This is my color scheme and how the cube looks after scrambling (thanks to Werner Randelshofer for the applet):

The main algorithm I'll use swaps the edges UL and UR and also swaps the corners URF and UBR. I chose it because I can execute it fast and it has a fairly "free" edge (UL) and fairly free corners. You'll see what I mean with "free" in a moment. Here's the Peter Jansen version of the algorithm (his 19-1):

(R U R' U') (R' F R2 U' R' U' R U) (R' F')

We'll first solve the edges, but it really doesn't matter. It always works the same way: We find out where the UR edge has to go, bring that edge to UL with some setup-algorithm A, apply the above algorithm, and apply the inverse of A. It is important to note that orientation does matter here. The UR edge is not the same as the RU edge. It's the same place, but the other orientation. For the corners, I use three letters of the faces in clockwise order. Here, URF is different from RFU. It's the same place, but a different orientation. And FUR is the third. Ok, let's start.

At place UR we have the green/white edge. We use the setup-algorithm (D F L' F') to bring its place to the UR place. Then apply the main algorithm. And finish with the inverse setup-algorithm (F L F' D'). The green/white edge is now at its correct place and we can go on to the next edge. Note that the setup-algorithm must not change the URF, UR and UBR cubies!

Next we have an exception. The UR place contains the blue/yellow edge. This is the edge that belongs there, but it is oriented the wrong way. What we do is instead of bringing the UR edge to its correct place is we bring some not yet solved edge to the UR edge (from where it will reach its correct place in the next iteration). In this case, the UL edge already contains a wrong edge, so we just apply the main algorithm without any setup-algorithm.

I'll not describe the remaining edges in detail but rather give you the whole solution in the following table. If you start with the scrambled cube and then always apply the given algorithm A, then the main algorithm and then the inverse of A, you'll solve all edges:

Note: I used too long setup algorithms at first. Below the next table you'll find the setup algorithms I use now.

The setup algorithms I now use are listed in the next table. Lower case letters mean you have to turn the face layer together with the slice layer next to it.

The corners are solved the same way. I look what's in the UBR corner and bring its goal place to URF. Then apply some setup-algorithm A, then the main algorithm, then the inverse of A. Note that A must not destroy the UL, UR and UBR cubies!

Also note that you'll of course also destroy the UL and UR edges again by swapping them in each iteration. However, this is no problem, just solve the corners and when they're solved, the UL and UR edges will be solved automatically (thanks to the parity issue ;-).

I was pretty strict here. Of course you can also switch the roles of UBR (my buffer position) and URF which might lead to shorter solutions. Also, you don't need to solve all edges first and corners afterwards. You could even mix it or solve an edge and a corner at the same time. But this is more complicated.

Since I've put this in my blindfold section I should probably say something about how to use this for blindfold puzzling. I'd recommend doing the edges first and corners afterwards (or the other way around). Don't mix it. When memorizing, memorize the left column values of the above tables. For corners, you only need to memorize the first two out of the three colors. Use the first characters of the colors to build a word you remember. For example, "yellow/green" is "YoGurt" for me. I replaced "orange" with "Cool" (actually "Cube orange") because you should only have consonants as first letters. Otherwise "yogurt" might also stand for "yellow/orange". This is similar to the master system for memorizing (but I'm not going through numbers, I use colors directly). You then end up with sequences of around 25 objects you need to remember in order. Of course you can also memorize other ways and still use this method for solving.