A Guide to F2L Look Ahead

If you average above 20 seconds using full CFOP, here's the single best speedcubing advice that anyone can ever give you: GO SLOW AND LOOK AHEAD. This guide contains information on F2L look ahead, from basic to advanced.

Part I: Basics

1. Introduction

Looking ahead (noun "lookahead") is the counterintuitive speedcubing concept, most often applied to F2L, that slowing down will improve your time. Executing an F2L algorithm is quickly is meaningless if one stops to look for the next pair. When looking ahead, one instead slows down while solving one pair to find the next pair. The goal is to have no pause, solving the entire F2L in one smooth motion.

Who this is for

You should start practicing to look ahead as soon as you know a simple layer-by-layer method.

2. First tests

Use these to convince yourself why looking ahead is a such a good idea.

2a. Average F2L tps

Calculate your tps (turns per second) for F2L. First find your F2L average. To do this correctly, you need to actually start from the cross. Inspect a scrambled cube as usual, then solve up to the end of F2L. Start the timer as soon as you finish the cross, not when you begin the first pair; the delay here must be included. Do another average to find your average F2L move count (no optimization!). Now, tps = (average time) / (average move count).

Find the corresponding speed in bpm (beats per minute). For example, if you average ~25 seconds with full CFOP, your speed is probably ~2 tps = ~120 bpm. Use a metronome and solve the F2L smoothly, doing one move every beat. To beat your current average, you only need to turn slightly faster than this. It probably feels slow!

As this should demonstrate, you don't need to turn very fast to be sub-20. The cross and F2L in 13 seconds corresponds, assuming 35 moves using full F2L, to 2.7tps = 162bpm. Try this speed on the metronome. With proper fingertricks, turning at speed is in itself not difficult. The challenge is to maintain this speed throughout F2L: looking ahead.

2b. Theoretical Average

Calculate your theoretical average, the sum of your averages for each step of your method, with inspection before each. For full CFOP, for example, this means allowing inspection before each of thep 7 steps: cross, 4 F2L pairs, OLL, PLL.

For top cubers, the difference between theoretical and real average can be as low as 1 second, coming from OLL/PLL recognition. Any additional difference is from pauses during F2L. Your theoretical average more or less represents your potential if you can turn at your top speed and still have 100% look-ahead, i.e. zero pause during F2L.

3. Practice Technique: Metronome

Use a metronome to cube at a fixed tps (turns per second), making one move per beat. Make sure to do this smoothly. Since you want to also practice the transition between the cross and the first pair, start from the cross.

Improving look-ahead while turning at top speed is almost impossible. The more effective path to reach your theoretical average is to start very slow with perfect look-ahead, then to maintain this look-ahead while slowly increasing the speed.

First try the speed for your current average. You may find even this difficult at first; if you are not used to looking ahead, you will inevitably stop between pairs. Use Part II of this guide to work on looking ahead, then test yourself again with a metronome. Once you are comfortable at one speed, try the next setting. For a solid sub-20 average, aim for 3 tps = 180 bpm. In addition to your target speed, once in a while (say every 10 solves) try your eventual goal speed, or even your top speed while still trying to look ahead. This should feel very frantic; the hope is to make your current target speed seem slow.

With newer cubes with a very light turn (e.g. type F or any Dayan), you may have trouble maintaining a slow speed. If this is the case, practice on an older cube with a heavy or gooey feel.

Part II: How to Look Ahead

Looking ahead involves two parts: locating the next pair and predicting its F2L case.

Locating the next pair: the Corner Bias

I first heard this from Harris Chan and Yu-Jeong Min in late 2006. Recently named and formalized by Chris Hardwick, the corner bias for F2L has been the most helpful lookahead technique for a number of cubers. Many fast cubers who never explicitly learned the corner bias have also noticed this tendency in their solves.

Solve the F2L with a corner bias: first look for a first-layer corner, then the matching edge. Chris's visibility/identifiability argument says that because corners can be identified from two stickers, more corners than edges can be identified without rotation, especially when the FR pair is solved.

Use Conrad Rider's Coracle to practice determining a corner from two stickers.

Tracking and Predicting the F2L case

Once you have located the next pair, you need to predict its F2L case.

Prediction practice

This is an exercise I've sometimes suggested though admitted never stressed: try to plan two pairs at once. Phillip Espinoza recently suggested it as a helpful exercise. After the cross, plan the first two pairs. At first, keep your eyes open to see if your plan works. Learn to do this blindfolded. This will be very difficult at first. Although you won't do this very often in actual speedsolves, the exercise will help you more easily look ahead.

Since predicting the location of the pair is relatively easy, the following tips focus on the orientation. Be sure to set your color scheme.

Common patterns

Many F2L algorithms work by making then inserting a pair. Learn how insertions affect another pair, especially the following harder cases.

: before

: after

: before

: after

: before

: after

: before

: after

Edge-Center matching

Consider the following F2L cases (U and V).

: RU2R'URU'R'

: RUR'URU'R'

: RUR'U'RUR'

: RU'R'URU'R'

: R'U2RU'RUR'

: R'URU'RUR'

: RU'R'U'RUR'

: RU2R'U'RUR'

In each case, the algorithm begins with the edge's side sticker matching a center, and begins by rotating that center to bring the edge to the middle layer. This can be used to help look ahead without completely tracking the next pair.

Example: Set-ups R'U'RUR'URU'L'U'L and R'U'RUR'U2RU'L'U'L
L'ULUR'...: This is especially useful for distinguishing the first two F2L pair patterns above (U and V) without completely identifying them. In both cases to the left, it's easy to see that inserting the FL pair leaves the next pair separated on U layer with the corner pointing up, so that the pair pattern is either U or V. We don't need to identify which one; edge-center matching tells us to insert with L'UL and match and continue with UR'. By this time, the appropriate ending becomes clear, allowing us to look ahead to the next pair.

Example: Set-ups RUR'U'RUR'U'L'U'L and RUR'U'FR'F'RU2L'U'L
L'ULU...: Here, the insertion L'UL should be followed by U and U2, respectively, for edge-center matching. In the second example, a standard solution is to rotate to y'R'U'RUR'U'R.

Edge orientation

Not all F2L algorithms start with edge-center matching. To understand why, we need to know about the orientation of F2L edges.

A (correctly) oriented or good F2L edge is one that can be solved in < U, R, L > (i.e. using only U, R, and L turns). An edge that is not oriented is said to be flipped or bad. An F2L edge in the U-layer is good (resp. bad) if its side sticker color matches the R or L face (resp. F or B face). An F2L edge in the middle layer is good (resp. bad) if its stickers have the same/opposing (resp. adjacent) colors as the centers they are next to. For more details, see my 3OP guide or Conrad Rider's ZZ guide.

: good

: bad

: bad

: good

An F2L case can be solved in < U, R, L > precisely when it has a good edge, which explains the terminology since R/L are prefered in speedcubing over F/B. For many cases with good edges to FR or BR slots, the standard algorithm is in < U, R >, which partly explains why edge-center matching occurs in many F2L algorithms.

Being aware of edge orientation allows us to understand and better recognize certain F2L algorithms and cases.

Example
R'URU2R'U'R: The basic solution (U2R'...) uses edge-center matching, but not the better solution (R'UR...). But because the solution is in < U, R >, the edge must be good; indeed, the side color matches the R face.

Example
RU'R'U'R'U'R: RU'R'U' visibly brings the pair to a position solvable by R'U'R, except perhaps with the edge flipped. But the edge is good to begin with and the pair lies in the U/R layers, so this must solve the edge.

Example
R'U2RUL'U'L: With considering edge orientation, it may be hard to distinguish this case from the one with flipped edge from this angle. The good edge tells us that we are in the case that can be solved in < U, R, L >.

It is helpful to know the following rules.

1. < U, R, L> moves preserve edge orientation.
2. A quarter F/B turn changes the orientation (good to bad, bad to good) of each edge piece it moves.
3. A quarter rotation y or y' interchanges R/L and F/B, hence changes the orientation of F2L edges on the U layer.

F2L cases with bad edges require eiter a rotation, reducing to a case with a good edge, or a special trick.

These rules let us understand center-edge matching in a different way.

Example: Set-up R'U'RUR'UR2U2R'
RU2'R' R'U'RU'R'UR: This is standard center-edge matching. The edge at UR is good. The insertion RU2'R', being in < U, R, L > (even < U, R >), preserves orientation, so we know that the resulting F2L case has a good edge and can be solved in < U, R, L >. Here, the standard algorithm is in fact in < U, R >.

This reasoning applies to cases more general than simple insertions.

Example: Set-up L'U'LUL'U2LU2RU'R'URU'R'
RUR'U'RUR' U2L'U2LU'L'UL: Suppose we identify the FL pair as we start solving the FR pair. The edge at FR will remain good after RUR'U'RUR'. This also visibly keeps the FL corner pointing up, so edge-center matching tells us to continue with U2L'.

Example: Set-up L'ULU'R'U2R2UR'U2RUR'
RU'R'U2RU'R' R'U2RUL'U'L: Here, we can note while solving the FR pair that the edge at BR is good. The next pair, which may be a bit hard to recognize without this consideration, can be solved in < U, R, L >.

Example: Set-up L'ULUBU'B'LU'L'U'LU2L'
LU2L'ULUL' bad edge: We find the edge at BR here as we start solving the BL pair. The edge is bad, so we need to rotate (y'zURU'R'F'R'F) or use a special trick (U'L2RB'R'BLU'L).

The Sledgehammer

The sledgehammer (R'FRF') changes the orientation of the edges at UF and UR and play a major role in fast algorithms for F2L cases with bad edges.

Example: Set-up L'U'L2U'L'FR'F'R
before after: On the left, edges at UF and UR are bad. After inserting the FR pair with a sledgehammer, they are good.

Knowing this effect helps us predict some complicated pairs.

Example: Set-up L'U'LU2L'U'LUL'U'LU'RUR'FR'F'R
R'FRF' RU'R' UL'ULU'L'ULU2'L'UL: A standard solution for the FR pair, with the bad edge at FR, begins with the sledgehammer. As we insert the FR pair (second figure), we can predict the pair position and corner orientation, but not necessarily the edge orientation during a speedsolve. But we know that the edge at UF turns from bad to good, so that the FL pair can be solved in < U, L >.