AI new cubing method

[Nikcubing13]

So I asked chatgpt to invent a NEW Rubik's cube solving method and this is what it came up with I think this is the reason u shouldn't ask AI for cubing stuff lol
"

The Prism Method​

The idea is to reduce the cube into a "prism" before solving.

Step 1: Make a Cross​

Same as CFOP.


Goal: 4 edges + center.

---

Step 2: Build Four Columns​

Instead of solving pairs, solve the four corner-edge pairs vertically, creating four 1×1×2 columns around the cross.


Each column is independent.


After this stage, the first two layers are solved, but in a different order than CFOP.

---

Step 3: Orient Last Layer Edges​

Use only R, U, F moves to flip the four top edges.


Now every edge is oriented.

---

Step 4: Place Last Layer Corners​

Move the four corners to the correct locations without worrying about orientation.

---

Step 5: Orient Corners​

Twist all four corners.

---

Step 6: Permute Last Layer Edges​

Cycle the four edges into their correct places.

---

Algorithms Needed​

• 5–10 edge orientation algs
• 4 corner placement algs
• 7 corner orientation algs
• 4 edge permutation algs

Around 20–25 algorithms, much fewer than full CFOP.

---

Why it could work​

• Independent "column" building is intuitive.
• Easy lookahead because each column is separate.
• Last layer is broken into simple stages.
• Lower algorithm count than CFOP.

---

Possible Downsides​

• More moves than CFOP.
• Columns might interfere with each other.
• Could require rotations during the build phase.

---

Even crazier idea: Ring Method​

1. Solve the middle layer first as a complete ring around the cube.
2. Solve the entire bottom face.
3. Solve the entire top face.

It completely ignores the traditional layer-by-layer order and would feel very different to solve."
Isn't the first one just the begginer's method lol and is the ring method even something that has a reason to exist or is even possible

 

[KubikPanda]

it seems quite inefficient, and even if you did the most efficient solve in this method I bet it would still be slow

 

[Nikcubing13]

Yea I bet

 

[K3n]

The ring method is just the belt method right?

 

[Pyjam]

You didn’t ask it an "efficient" method.

 

[Nikcubing13]

um ok ill ask for an efficient method
"
This is actually a really fun challenge . Let's invent a new speedsolving method from scratch. I'll call it:

PRISM Method (Progressive Reduction + Intelligent Slot Management)​

The idea: combine the best parts of CFOP, Roux, and ZZ, but remove their biggest weaknesses.

Main philosophy:​

Instead of solving layer-by-layer, PRISM solves structures:

1. Create a skeleton
2. Build connected blocks
3. Finish edges/corners together
4. Use a final "orientation + permutation" phase

Target:

• ~45–50 moves
• Low pauses
• Easier transition than Roux
• Better lookahead than CFOP

---

Step 1: Prism Cross (PC) — ~8 moves​

Instead of normal cross, you make:

• Cross edges solved
• One adjacent corner solved
• One F2L pair partially created

Example goal:


<span><br><br></span>

but with one corner already attached.

Why?

CFOP cross wastes time because after cross you have to restart thinking.

PRISM gives your brain a "starting block".

Average:
8 moves

---

Step 2: Twin Blocks — ~20 moves​

Instead of solving four F2L pairs separately, create two giant blocks:

Left block:



<span>Corner + Edge + Corner + Edge</span>

Right block:



<span>Corner + Edge + Corner + Edge</span>

Like Roux blocks, but they are connected to the cross.

You are basically making:



<span>[BLOCK]

[BLOCK]</span>

The middle layer is still messy.

Benefits:

• fewer pauses
• more intuitive
• less algorithm dependence

---

Step 3: Edge Bridge (EB) — ~10 moves​

Now solve all four remaining edges.

But here's the trick:

You don't insert them normally.

You use a special idea:

orient first, then place.

Similar to ZZ.

Example:

1. Flip all bad edges
2. Insert them with short triggers

Average:
8–12 moves.

---

Step 4: Prism Last Layer (PLL + OLL Combined)​

Instead of:

OLL (57 algs)
+
PLL (21 algs)

PRISM uses:

24 "finish cases"

Each algorithm does:

• orientation
• permutation

at the same time.

Like COLL + EPLL, but expanded.

Example:

Case:



<span>Yellow cross done<br>Corners twisted<br>Edges swapped</span>

One algorithm fixes everything.

---

Full Method:​

PRISM​

1. Prism Cross​

Cross + corner

↓

2. Twin Blocks​

Two 2x2x3 structures

↓

3. Edge Bridge​

Solve remaining edges

↓

4. Prism Finish​

One-step last layer

---

Estimated Speed​

A good solver:

Beginner:

45 seconds

Intermediate:

20–25 seconds

Advanced:

8–12 seconds

World-class:

maybe sub-6 possible

---

Why it could beat CFOP​

CFOP:

• Cross → stop → F2L → stop → OLL → stop → PLL

PRISM:

• Every step creates information for the next step.

The goal is continuous solving, not separate phases.

---

The biggest problem would be the same as every new method: developing enough algorithms and training recognition. CFOP dominates because millions of people optimized it for decades.

But a method like this could genuinely become competitive if someone built the alg sets. It has a lot of ideas that modern methods already hint at: blockbuilding, EO, and fewer phase transitions. "
Here u go is this efficient wdyt​

 

[Nikcubing13]

The ring method is just the belt method right?

I guess I don't really know the belt method but probably