Hawaiian Kociemba 3x3x3 TUTORIAL - Method of Michael H.

[adimare]

hey guys ive almost completely transitioned over to this method because I can get insanely fast times with the petrus variant.
to get times about 8 seconds lower than my former main I do(im learning algs to make this faster):
1.petrus 2x2x2(plan during inspection)->2x2x3
2.solve final 2 pairs
3.2-look corners(OCLL+T and Y perm)
4.L5E EO(roux-like EO, but with only 5 edges)
5.insert the DF edge with M' U2 M' and EPLL

How would you finish this solve?

Scramble: x B2 L' D2 L2 U2 B2 L F2 L' U2 B' R' D' L2 U L' U' L' U2
D' F L' F2 D' L2 // 2x2x2
x U' R' U R2 U' R' y // 2x2x3
y U' R U' R' y' U R' U2 R' U' R // F2L pair

The piece you need to finish the F2L is trapped in DR, what to do? (You will run into this issue one out of every 6 solves using the method you described).

 

[Shiv3r]

I don't think that's really relevant to this. Sure, the solves were faked, but the point is that it gave people the (not so new) idea of leaving a part out of F2L to solve while solving LL.

Slightly irrelevant thought, I don't think it's good to put all of these very diverse methods under the umbrella term "Hawaiian Kociemba", considering how different some of the variations are.

I kind of agree. we could consider pikashit to be a hawaiian Kociemba method if we don't categorize them.
but then, they all include leaving one edge out of the F2L and solving it at the end.

 

[Shiv3r]

How would you finish this solve?

Scramble: x B2 L' D2 L2 U2 B2 L F2 L' U2 B' R' D' L2 U L' U' L' U2
D' F L' F2 D' L2 // 2x2x2
x U' R' U R2 U' R' y // 2x2x3
y U' R U' R' y' U R' U2 R' U' R // F2L pair

The piece you need to finish the F2L is trapped in DL, what to do? (You will run into this issue one out of every 6 solves using the method you described).

You gotta think like a roux solver. Try M' (U or U' or U2) M.

If you set up the corner right then you can often pair them up by executing this.

 

[JTWong71]

If I were doing it, I would do:
y U M' U2 L' U' l

 

[Shiv3r]

If I were doing it, I would do:
y U M' U2 L' U' l

I didn't actually put the solve in, i just answered the question of the trapped f2L edge, but yeah that works.

 

[wir3sandfir3s]

What you described is almost exactly the T-CELL method, except I think in that method, you leave out a corner out of F2L as well, and do TCLL to solve the remaining 5 corners. My point is, that method isn't exactly new or revolutionary.

In fact, I think I can improve it a bit. So I'm fairly certain that doing L5E the way your doing, and doing LSE Roux style is about the same amount of moves, and IMO, LSE Roux style has better lookahead, so the method could be revised as:

1. Build 2 2x2x1s (At least 1 planned in inspection)
2. Solve final 2 pairs
3. CMLL
4. LSE

At this point, this method is Roux with a different way of building the blocks. And in my opinion, building one block at a time is better anyway, due to no rotations and the moveset quickly being limited to [R,r,M,U].

But if you're better with the method that you proposed, good for you .

TCELL isn't very original either, then, because there are literally dozens of methods like that...
The L5E is a lot better IMO because you can actually do the EO while inserting the edge, and at that point you are left with an EPLL which all have awesome recog and execution. L5E is definitely better here, but L6E is better for a roux setup.

 

[wir3sandfir3s]

How would you finish this solve?

Scramble: x B2 L' D2 L2 U2 B2 L F2 L' U2 B' R' D' L2 U L' U' L' U2
D' F L' F2 D' L2 // 2x2x2
x U' R' U R2 U' R' y // 2x2x3
y U' R U' R' y' U R' U2 R' U' R // F2L pair

The piece you need to finish the F2L is trapped in DL, what to do? (You will run into this issue one out of every 6 solves using the method you described).

You wouldn't even solve the pair if it is there, insert it so that the edge doesn't end up in DL. Simple... Unless I completely misinterpreted you which I am very good at.

 

[AlexMaass]

https://docs.google.com/spreadsheets/d/1zSHsnNOpfEQP7qYrX_plR0YIqtEnx9Q9rvjCJe8URco/edit?usp=sharing
updated link, broke for some reason oops

 

[wir3sandfir3s]

https://docs.google.com/spreadsheets/d/1zSHsnNOpfEQP7qYrX_plR0YIqtEnx9Q9rvjCJe8URco/edit?usp=sharing
updated link, broke for some reason oops

Cool thx

 

[adimare]

You gotta think like a roux solver. Try M' (U or U' or U2) M.

If you set up the corner right then you can often pair them up by executing this.

You wouldn't even solve the pair if it is there, insert it so that the edge doesn't end up in DL. Simple... Unless I completely misinterpreted you which I am very good at.

That makes no sense. You guys should try opening the link before responding next time.

If I were doing it, I would do:
y U M' U2 L' U' l

Exactly, you're forced to do a y rotation and use M' to take the piece out. In a speedsolve that is terribly inconvenient. What's the advantage of not solving the piece again? All I notice is it leads to worse lookeahead, less chances of LL skips, and a ridiculous amount of algs needed to solve the LL in 2 algs... Have you seen the Mike H expose video? Rowe even says at some point "I solved F2L but left an edge out, which you should never do".

 

[Shiv3r]

It is efficient, and lookahead isnt that bad.
If you can get lower TPS but also lower movecount and can get decentlyfast(sub-20), then all you need to do to get sub-10 is to speed up TPS slowly

 

[gyroninja]

It is efficient, and lookahead isnt that bad.
If you can get lower TPS but also lower movecount and can get decentlyfast(sub-20), then all you need to do to get sub-10 is to speed up TPS slowly

It's not that efficient. You are only saving like 3 moves during f2l (assuming the edge doesn't get trapped). Doing all the complicated ll stuff isn't really worth it.

 

[wir3sandfir3s]

It's not that efficient. You are only saving like 3 moves during f2l (assuming the edge doesn't get trapped). Doing all the complicated ll stuff isn't really worth it.

Not if you do it correctly, you can easily avoid getting the edge trapped...

 

[Shiv3r]

It's not that efficient. You are only saving like 3 moves during f2l (assuming the edge doesn't get trapped). Doing all the complicated ll stuff isn't really worth it.

explain the insanely fast times then.

 

[adimare]

explain the insanely fast times then.

Did I miss something?

 

[wir3sandfir3s]

Did I miss something?

The times on this are really good...
That kind of outclasses all evidence of it being bad, but correct me if I'm wrong.

 

[adimare]

The times on this are really good...
That kind of outclasses all evidence of it being bad, but correct me if I'm wrong.

Are you talking about Shiv3r's alleged 32 second averages? What times are you getting with this wir3sandfir3s? I must have missed evidence of these "insanely fast times".

 

[wir3sandfir3s]

Are you talking about Shiv3r's alleged 32 second averages? What times are you getting with this wir3sandfir3s? I must have missed evidence of these "insanely fast times".

Just because they aren't fast for you doesn't mean they aren't fast for other people...
He claimed that he dropped about 10 seconds off his times with Roux, and Alex Maass made a video on his YouTube channel using the SAME variant and got some good times as well.

 

[adimare]

Just because they aren't fast for you doesn't mean they aren't fast for other people...

Ok then, let's look at the two examples you used.

He claimed that he dropped about 10 seconds off his times with Roux

Was Roux his main? Don't you find it hard to believe that someone would drop 10 full seconds with a new method regardless of how good the method is? This claim is also ridiculous because the Petrus variant heavily relies on being decent at Roux (which he's not).

Alex Maass made a video on his YouTube channel using the SAME variant and got some good times as well.

That's because of good TPS. Take a look at a reconstruction of his 16.43 second solve from that video:

Scramble: y' z U L U2 B2 D' F2 B L D' R F R2 D2 L2 B' R2 F' B2 U2 D2
z' // Inspection
R F R' U' R L' U L // 2x2x2
l R M' U R' U' R U r U' L' U r' // 2x2x3
U2 R U R' U y R U' R' // Pair
y' d R' U' R // Pair
y R U2 R' U' R U' R' U R U R' F' R U R' U' R' F R2 U' R'// CMLL
y U' M' U M // EO
y' U2 R2 U2 M2 U M2 U M' U2 M2 U2 M' R2 U2 // L5E
(72 moves)

Now, had he solved the cross piece after the 2x2x3 he would have had to execute 2 algs to finish the LL (OLL and PLL). Instead of that, he had to do CMLL, then Roux style EO, then a conjugated Z-Perm to solve the 4 unsolved edges... do you really think not having to solve one piece is worth all that?

 

[Kit Clement]

Tangentially relevant:

It's totally possible to make an absolutely garbage method and still get sub-20 on it. This doesn't obscure the fact that this said method would be completely outclassed by any real speedsolving method, just like this method is.

I'm still amazed people are arguing over the validity and usefulness of a method that fabricated itself out of an April Fools' Day joke. The solves that inspired this whole discussion were made by undoing a solved cube.