Does anyone else do Fridrich F2L this way?

[blah]

I don't quite know why, but I seem to have a habit of doing F2L in the following manner about 90% (rough estimate) of the time:

(0. Do cross on D)
1. Spot an E-slice edge on U (occurs 98.6%, 97.1%, 93.3% and 80.0% of the time for the first to fourth corner-edge pairs respectively, you'd have to be pretty unlucky not to spot one )
2a. Note the orientation of that edge
2b. If necessary, do a y/y'/Dw/Dw' such that the edge can be solved using only <U, R, L>
3. Locate the corner
4. Solve the corner-edge pair using only <U, R, L> (most of the time it's only <U, R> or <U, L> though)

I know others who spot corners first when solving corner-edge pairs, mainly because the white (most common cross color) stands out among the other colors. So I was wondering, how many people 1. spot edges first, and 2. orient the edge before solving the pair?

I guess I developed this habit because I don't like rotating the cube while solving a pair, kinda breaks my momentum and stuff.

The only time when this approach leads to a longer solution is when you get these cases:

I do R U R U R U' R' U' R' (and its mirrors) or Dw' L' U L Dw R U' R' (and its mirrors) to get at least two LL edges oriented after inserting the corner-edge pair.

I do R U' R U2' R2' U' R2 U' R2' (and its mirrors) or leave the edge misoriented if it already was, and do R U' R' U2 [u'] R' U' R (and its mirrors) to get at least two LL edges oriented.

If the back slot is empty, I just do R' U2 R2 U R (and its mirrors); if it's not, I'd re-insert the back slot with R' U2 R2 U R2' U R (and its mirrors). I also do Dw R' U2 R Dw' R U R' (and its mirrors) if there are no LL edges oriented.

I generally avoid cases in which an E-slice edge is in the E-slice but misoriented (i.e. not solvable using only <U, R, L>), because that means a cube rotation (or Dw/Dw') is necessary between the setup and the insertion of the corner-edge pair.

----------

In case you were wondering how I obtained 98.6%:

1 - ( (4/8) * (3/7) * (2/6) * (1/5) ) = 0.986

You should be able to figure out the rest yourself

 

[emu]

2b. If necessary, do a y/y'/Dw/Dw' such that the edge can be solved using only <U, R, L>

What if the edge piece is stuck in a bad orientation (so that no matter how you rotate it/do Dw moves, it is still not oriented)? That's one reason why I generally try to keep my rotations for the insertion, if possible. Also, I do this because rotating the cube for insertions helps me to lookahead using two angles, sparing me a cube rotation or U move if I hadn't seen a corner and/or its edge piece.

That said, I look for corners first, then its edge, for the exact reasons you stated.

 

[blah]

2b. If necessary, do a y/y'/Dw/Dw' such that the edge can be solved using only <U, R, L>

What if the edge piece is stuck in a bad orientation (so that no matter how you rotate it/do Dw moves, it is still not oriented)? That's one reason why I generally try to keep my rotations for the insertion, if possible. Also, I do this because rotating the cube for insertions helps me to lookahead using two angles, sparing me a cube rotation or U move if I hadn't seen a corner and/or its edge piece.

1. Spot an E-slice edge on U (occurs 98.6%, 97.1%, 93.3% and 80.0% of the time for the first to fourth corner-edge pairs respectively, you'd have to be pretty unlucky not to spot one )

...

I generally avoid cases in which an E-slice edge is in the E-slice but misoriented (i.e. not solvable using only <U, R, L>), because that means a cube rotation (or Dw/Dw') is necessary between the setup and the insertion of the corner-edge pair.

 

[emu]

Oh, sorry. :fp

 

[DAE_JA_VOO]

2a. Note the orientation of that edge
2b. If necessary, do a y/y'/Dw/Dw' such that the edge can be solved using only <U, R, L>

I personally tend not to do this. I only learned how to do it about 6 weeks ago, but I blew it off. I was very slow at it at first, and as such, decided not to go with it. When I start struggling at, say, 15 seconds or so, I'll consider it.

Do you do this for ALL your pairs? Wouldn't you end up doing lots of cube rotations (well, 4, max) if you did it with each pair? Wouldn't it be easier to identify which edges are misoriented BEFORE you start solving F2L, solve the correctly oriented edges, do one cube rotation (or d/d') and then solve the rest? That would guarantee you only needing a single rotation for the whole of F2L, not so?

 

[blah]

Do you do this for ALL your pairs? Wouldn't you end up doing lots of cube rotations (well, 4, max) if you did it with each pair? Wouldn't it be easier to identify which edges are misoriented BEFORE you start solving F2L, solve the correctly oriented edges, do one cube rotation (or d/d') and then solve the rest? That would guarantee you only needing a single rotation for the whole of F2L, not so?

Exactly! You got the gist of (the ideal situation of) my whole F2L approach!

I'm nowhere near that efficient in terms of cube rotations yet, but yes I am working towards it, pretty slowly though I've sorta "retired" from speedsolving for a while now, but fortunately, this doesn't mean I've retired from researching speedsolving approaches

That would guarantee you only needing a single rotation for the whole of F2L, not so?

Not quite. Misoriented edges trapped in the E-slice. That's the problem that's been giving me a very big headache. The best solution I can think of now (which isn't a good solution at all!) is, after solving all correctly oriented edges (together with their corners), quickly glance the E slice for any misoriented edges, take them out, do a quarter rotation about the y-axis, then solve the rest. Of course, this would add to the move count, so I've been trying to learn all special F2L cases with misoriented edges trapped in other slots, haven't progressed far though, there are way too many cases and almost none are intuitive

 

[DAE_JA_VOO]

That would guarantee you only needing a single rotation for the whole of F2L, not so?

Not quite. Misoriented edges trapped in the E-slice. That's the problem that's been giving me a very big headache.

Yeah, I saw this just after making my post. Hmmm... Headache indeed.

You might want to speak to Escher about this. He's actually the guy who introduced me to this whole concept, and he really knows what he's talking about on the subject

 

[AvGalen]

Do you do this for ALL your pairs? Wouldn't you end up doing lots of cube rotations (well, 4, max) if you did it with each pair? Wouldn't it be easier to identify which edges are misoriented BEFORE you start solving F2L, solve the correctly oriented edges, do one cube rotation (or d/d') and then solve the rest? That would guarantee you only needing a single rotation for the whole of F2L, not so?

Exactly! You got the gist of (the ideal situation of) my whole F2L approach!

I'm nowhere near that efficient in terms of cube rotations yet, but yes I am working towards it, pretty slowly though I've sorta "retired" from speedsolving for a while now, but fortunately, this doesn't mean I've retired from researching speedsolving approaches

That would guarantee you only needing a single rotation for the whole of F2L, not so?

Not quite. Misoriented edges trapped in the E-slice. That's the problem that's been giving me a very big headache. The best solution I can think of now (which isn't a good solution at all!) is, after solving all correctly oriented edges (together with their corners), quickly glance the E slice for any misoriented edges, take them out, do a quarter rotation about the y-axis, then solve the rest. Of course, this would add to the move count, so I've been trying to learn all special F2L cases with misoriented edges trapped in other slots, haven't progressed far though, there are way too many cases and almost none are intuitive

Just take a different approach and learn all OLL's with 1 edge flipped in FR. Requires quite some algs but turns all those nasty cases into OLL-skips

 

[blah]

Just take a different approach and learn all OLL's with 1 edge flipped in FR. Requires quite some algs but turns all those nasty cases into OLL-skips

No. There's no skip. You STILL have to solve the F2L pair to have the edge misoriented and the corner solved.

 

[Escher]

Have you considered <R, L, U, u>?
There are plenty of tricks you can use to solve pairs that have misoriented edges without cube rotations.
I know that Breandan Vallance knows a load of tricks to avoid cube rotations and he averages just under 1 per solve (plus he solves f2l pairs in the order BL, FL/BR, BR/FL, FR).

Couple of examples:
R u R' u' R' (and reflection and inverse)
R u R u' R'
R' U R u' L U L' u (as above)
R U' u R u' R'

When preinspecting I normally just look for two obvious edges and track their orientation during the cross, rather than actually manipulating them so that a cube rotation solves their orientation.

Btw, if you have the edge in the correct place but flipped (and an empty slot) you can just use variations on:
R U' R' U u R U' R' u'
you can execute that without cube rotations or regrips even

 

[AvGalen]

Just take a different approach and learn all OLL's with 1 edge flipped in FR. Requires quite some algs but turns all those nasty cases into OLL-skips

No. There's no skip. You STILL have to solve the F2L pair to have the edge misoriented and the corner solved.

No, you would solve the OLL while fixing the misoriented edge. I am talking about the case where the corner is already solved.

 

[xXdaveXsuperstarXx]

I go for corners in stead of edges. I don't orient the edges. I mean, isn't that what full OLL if for? To do that for you? Oh, BTW, I love edges in the E layer, they're easy to spot and very quick to pair up.

 

[blah]

Just take a different approach and learn all OLL's with 1 edge flipped in FR. Requires quite some algs but turns all those nasty cases into OLL-skips

No. There's no skip. You STILL have to solve the F2L pair to have the edge misoriented and the corner solved.

No, you would solve the OLL while fixing the misoriented edge. I am talking about the case where the corner is already solved.

Which is only one of the many cases with an edge misoriented in the E slice. I don't think you got my point...

I go for corners in stead of edges. I don't orient the edges. I mean, isn't that what full OLL if for? To do that for you? Oh, BTW, I love edges in the E layer, they're easy to spot and very quick to pair up.

You missed it too :fp

Have you considered <R, L, U, u>?
There are plenty of tricks you can use to solve pairs that have misoriented edges without cube rotations.
I know that Breandan Vallance knows a load of tricks to avoid cube rotations and he averages just under 1 per solve (plus he solves f2l pairs in the order BL, FL/BR, BR/FL, FR).

Couple of examples:
R u R' u' R' (and reflection and inverse)
R u R u' R'
R' U R u' L U L' u (as above)
R U' u R u' R'

When preinspecting I normally just look for two obvious edges and track their orientation during the cross, rather than actually manipulating them so that a cube rotation solves their orientation.

Btw, if you have the edge in the correct place but flipped (and an empty slot) you can just use variations on:
R U' R' U u R U' R' u'
you can execute that without cube rotations or regrips even

Yes yes yes! THANK YOU!

 

[xXdaveXsuperstarXx]

You missed it too:fp

You missed my point.

 

[DAE_JA_VOO]

You missed it too:fp

You missed my point.

Nope, YOU missed HIS.

 

[fanwuq]

Blah, I do this sometime.

I remember Chris Hardwick posting that he thinks it is better to search for corners first because it is easier to see them in the U layer.
Usually when I make the cross, I try to solve 1 corner with it. Then I keyhole the edge then try to recognize the orientation of my next 2 F2L edges.

 

[blah]

Let me clarify something:

While solving an F2L pair (or the cross), I spot an E-slice edge which I know will still remain in the U layer after I've inserted the current pair. Which means, while solving that F2L pair, I will know whether or not my next pair requires a rotation/Dw. So if necessary, I rotate while doing my final R U R' (or whatever corner-edge pair insertion moves), so for example, an R U R' insertion would be something in between R U R' and R U y F'. By the time I've inserted my current pair, I would have spotted the corresponding corner for the next pair (well, ideally ), so I would be ready for my next slot.

Personally, I feel that such rotations cost less time than "normal rotations", but that's just me

@Wuqiong: I'm not a big fan of keyhole because the Uw/Uw' moves mess up my bearings But I tend to do exactly what you do when solving the cross. But I solve that corner-edge pair differently, it's just an intuitive thing that comes from my beginnings as a Petrus user: If the solved corner is at DRF and the corresponding edge is at UR, I do R2 U' R' U R2', which IMHO, is MUCH faster than keyhole.