Let's talk about Petrus

[mDiPalma]

You make the 222 in DBL. Then you expand to 223 in DBR (<R,U,F> fingertricks). Then you orient the edges (<R,U,F> fingertricks). Then you do ONE rotation (y'). Then you finish the F2L (<R,U> fingertricks). Then you do the LL (in your favourite way).

Weird rotations ? Bad fingertricks ? I don't understand why you say that. There is often only one rotation and blockbuilding is RUF (then RU).

Yeah, that's exactly how I do it too. The disadvantages with Petrus are certainly NOT lookahead and rotations/ergonomics.

Please do post speedsolving reconstructions/videos!

 

[GuRoux]

i think disadvantage of petrus is too much freedom and a high stress on efficiency. This leads to bad lookahead and rotations. i think if you formalize the block building depending on what you see in inspection, it will be much more manageable, and still retain good movecount and few rotations.

 

[ion]

i think disadvantage of petrus is too much freedom and a high stress on efficiency. This leads to bad lookahead and rotations. i think if you formalize the block building depending on what you see in inspection, it will be much more manageable, and still retain good movecount and few rotations.

Well, a lot of Petrus users think they must have a very low movecount, just because they use Petrus. It's a wrong approach. In speedsolving, movecount is not THAT important — we have to focus on ergonomy. Like I said, we can solve the cube with very few rotations and very good fingertricks, if we consent to do ~55/60 moves.

Unfortunely I have not the hardware to record me, but I will post some reconstructions later.

 

[GuRoux]

Well, a lot of Petrus users think they must have a very low movecount, just because they use Petrus. It's a wrong approach. In speedsolving, movecount is not THAT important — we have to focus on ergonomy. Like I said, we can solve the cube with very few rotations and very good fingertricks, if we consent to do ~55/60 moves.

Unfortunely I have not the hardware to record me, but I will post some reconstructions later.

i can sub 55 moves in a slow solve pretty consistently just with 1x2x3 ->2x2x3, so someone better at petrus could probably do around 50 in a speedsolve and have little rotations, easy fingertricks, and good movecount.

 

[porkynator]

I think I average about 12 with Petrus, yesterday I got a sub-8 single (I'm not sure if it is the first I get with Petrus); at the moment it's my second fastest method (I average 11 with ZZ).
I agree that being forced to solve EO 2 pieces at a time with 3 moves conjugates isn't nice, but it can be pretty fluent with good look ahead.
My approach is: start with any 2x2x2 block, then expand it in order to have W/Y on D and O/R in F (my usual ZZ orientation); it's easier for me to recognise EO this way, but I can also solve with B/G on F.
My thoughts: maybe Petrus is a bit underrated at the moment; I don't know if it has the right to be part of the "big 4" due to lack of evidence, but I see the potential for (at least) sub-10 global average with it (not by me).
A side note: I think it is really nice as an intermediate method, and I have succesfully taught it to a beginner in just a couple of "lessons". I wish I had started with Petrus and then moved on to something else (or even stuck to it, why not).

 

[ion]

I agree, it's a very pedagogical method. You can learn a lot of things about how the cube works, and it's quite natural (as Lars say, the first layer is not natural because you can't do anything without breaking it). I think it's not a very good idea to teach a layer-by-layer method to a newbie, because it's just... well, stupid.

 

[TDM]

You make the 222 in DBL. Then you expand to 223 in DBR (<R,U,F> fingertricks). Then you orient the edges (<R,U,F> fingertricks). Then you do ONE rotation (y'). Then you finish the F2L (<R,U> fingertricks). Then you do the LL (in your favourite way).

Weird rotations ? Bad fingertricks ? I don't understand why you say that. There is often only one rotation and blockbuilding is RUF (then RU).

Also, look-ahead. Once you made the 222, you can see ALL the unsolved cubies. All the hidden cubies are the solved ones.
And block-building don't require a lot of thought when you are used to it.

I'm not saying Petrus is the best method. I even think CFOP and Roux are slighty better. But Petrus is underrated, and is as good as ZZ in my opinion. ZZ has his advantages, but Petrus too (we look at only one side of the cube, we utilize only one hand...)
I can post speedsolves reconstructions if you are interested in.

(Also, sorry for my bad english.)

RUF =/= good fingertricks

 

[obelisk477]

RUF =/= good fingertricks

Well which 3-gen set do you prefer then? I personally find RUL awkward. Maybe RUD, but until recently, not too many algs have been generated using it.

 

[TDM]

Well which 3-gen set do you prefer then? I personally find RUL awkward. Maybe RUD, but until recently, not too many algs have been generated using it.

I prefer RUF, but the point I'm trying to make is that just because it's RUF, it doesn't necessarily mean the fingertricks are good.

 

[ion]

You're right, but they are.

 

[TDM]

You're right, but they are.

Do you honestly think they're as good as Roux, CFOP or ZZ?

 

[Smiles]

You make the 222 in DBL. Then you expand to 223 in DBR (<R,U,F> fingertricks). Then you orient the edges (<R,U,F> fingertricks). Then you do ONE rotation (y'). Then you finish the F2L (<R,U> fingertricks). Then you do the LL (in your favourite way).

Weird rotations ? Bad fingertricks ? I don't understand why you say that. There is often only one rotation and blockbuilding is RUF (then RU).

Also, look-ahead. Once you made the 222, you can see ALL the unsolved cubies. All the hidden cubies are the solved ones.
And block-building don't require a lot of thought when you are used to it.

I'm not saying Petrus is the best method. I even think CFOP and Roux are slighty better. But Petrus is underrated, and is as good as ZZ in my opinion. ZZ has his advantages, but Petrus too (we look at only one side of the cube, we utilize only one hand...)
I can post speedsolves reconstructions if you are interested in.

(Also, sorry for my bad english.)

first of all, at least in my opinion, weird rotations and bad look ahead are not Petrus problems so i will not address them.

Relatively difficult block building:
- because many pieces aren't oriented. In Roux you use M/M' to orient, in CFOP you use y/y', and in ZZ they're all done. F moves are slow.
- because you have the freedom to expand 3 ways (more thinking). if you limit yourself to 1 (less thinking) then you might as well do an xcross.

Annoying to fix bad edges:
- because there are some bad cases and sometimes it's just not worth it.
- because you have to recognize mid-solve, which may require a pause if there are a lot (ZZ uses inspection for this which is better)
- because 3 moves = 2 edges

Bad finger tricks:
- relative to CFOP (F2L algs) / ZZ <R,U,L> / Roux <R,r,U,M> or moo.
- RUF is not better than RUL because RUL means "2gen on the left side and then 2gen on the right side". RUL is not the same as an RUL V perm or RUL N perm, because those are rapidly alternating R/L and thats why they suck. If you have to do a lot of F moves at once, it's faster to just rotate. that's the premise of CFOP, at least.

F turns are not good. the other methods try and avoid F turns. they're fine for specific algorithms where they integrate nicely (like J perm).

But there are good things about Petrus:
- low move count
- EO done at LS/LL (slight advantage)
- 2-gen for last block (not sure if this is advantageous because the EO step is required for this)

I'm not saying petrus sucks. it doesn't. it's big 4. but there are many reasons why it's not as good as the big 3.

There is often only one rotation

ZZ = no rotations
Roux = no rotations
CFOP = usually more rotations, but no matter how many bad edges, i can force there to only be 1 rotation. if you let me use F turns a lot (like petrus), i can guarantee you more than 50% of my speedsolves will have no rotations. but F turns are slow, so i rotate instead.

Petrus = build a block on DBL without ever using a B move? i sense that there are often 2 cube rotations.

out of curiosity, is petrus your main method?

 

[ion]

difficult block building = relative to ZZ because your pieces aren't oriented yet.

Slighty more difficult, but still okay. Rouxers use M to deal with bad edges, Petrus use F. Of course M is more comfortable, but not really easier. Just a matter of experience. In most cases it's the same strategy.

difficult block building = because you have the freedom to expand 3 ways (more thinking). if you limit yourself to 1 (less thinking) then you might as well do an xcross.

I tend to limit to one, less thinking and less rotations. XCross is cool but I still like the 223, it's very ergonomic and you have a good view of the cube.

hard to fix bad edges = because there are some bad cases and sometimes it's just not worth it.
hard to fix bad edges = because you have to recognize mid-solve, which may require a pause if there are a lot (ZZ uses inspection for this which is better)

Erik Johnson used to solve bad edges in one second. Recognition and moves are very fast most of the time. Bad cases happen but actually you can say that for every step in every method.

bad finger tricks = relative to CFOP (F2L algs) / ZZ <R,U,L> / Roux <R,r,U,M> or moo.

I really don't see why they're bad relative to CFOP or ZZ. It's RU most time, with some F (not a lot though). Mixing two hands have disadvantages too.

F turns are not really good btw. all the other methods avoid F turns. they're fine for specific algorithms where they integrate nicely (like J perm).

F turns integrate nicely in EO. Not so much in 223 though, I admit it.

I'm not saying petrus sucks. it doesn't. it's big 4. but there are many reasons why it's not as good as the big 3.

Maybe you're right. I'm not sure. It's impossible to be sure about that. The gap is probably very short (at least relative to ZZ).

if you let me use F turns a lot (like petrus), i can guarantee you more than 50% of my speedsolves will have no rotations. but F turns are slow, so i rotate instead.

You rotate to do F insertions (like F'UF) or similar things. You have not things like that in Petrus. F turns are typically alone, and most of the time they combine well with R turns.

Petrus = build a block on DBL without ever using a B move? i sense that there are often 2 cube rotations.

I didn't consider the first step, which is very situation-dependant (in every method).

out of curiosity, is petrus your main method?

Yes it is.

I'm not saying petrus sucks.

I know. Actually I find this discussion interesting.
Also, I don't intend to be stubborn. I just feel like there is a lot of misconceptions and clichés about Petrus method. I'm very aware that it's not a perfect method (not straightforward enough, I guess).

 

[TDM]

I tend to limit to one, less thinking and less rotations.

But more moves, meaning you're not taking advantage of Petrus' efficiency, which is, in my opinions, its most important advantage over other methods.

XCross is cool but I still like the 223, it's very ergonomic

I'm still not convinced it's anywhere close to building an EO223 in ZZ, F2B in Roux or CFOP F2L.

Erik Johnson used to solve bad edges in one second.

Really? Do you have any proof?

Recognition and moves are very fast most of the time. Bad cases happen but actually you can say that for every step in every method.

I wouldn't have thought recognition would be as easy as with the other three methods. Again, I don't think the moves are faster. ZZ/Roux/CFOP all have very easy to fingertrick steps for 90% of the solve, not 50%. And actually, the bad cases with Roux and CFOP aren't that bad at all. I think this is one of the limitations of ZZ though.

I really don't see why they're bad relative to CFOP or ZZ. It's RU most time, with some F (not a lot though). Mixing two hands have disadvantages too.

There are a lot more F moves in Petrus than CFOP. With ZZ they're all done in the start, planned in inspection and done quickly - definitely faster than Petrus. Petrus has a lot more F moves. Mixing two hands isn't bad with CFOP, since usually the regrip isn't noticable as you rotate - but again, this is one of the problems with ZZ.

Maybe you're right. I'm not sure. It's impossible to be sure about that. The gap is probably very short (at least relative to ZZ).

ZZ isn't that far behind the other methods, since although it has the disadvantages compared to CFOP I said above, it has other advantages over CFOP. The gap between Petrus and the other methods is much bigger.

You rotate to do F insertions (like F'UF) or similar things. You have not things like that in Petrus. F turns are typically alone, and most of the time they combine well with R turns.

What do you mean by this?

 

[Smiles]

i think it means that instead of using F for CFOP-style inserts (pairs of F moves like the example), they occur less frequently and not in pairs.
i'm not sure if this is actually true though.

 

[Kirjava]

I'm of the conviction that it lends itself to high movecounts and more difficult recognition when used as a speed method - and the EO step really isn't helping.

Do we have a decent amount of recons for a 'fast' petrus user?