What do you think of the APB method?

[OtaMota]

You guys may have recently heard of the APB method, and is catching on with a lot of people. What are your thoughts about it, and do you think we will see someone in the top 10-20 using in the next 5 years, do you think it could compete with methods like CFOP or Roux?

 

[Caden Fisher]

You guys may have recently heard of the APB method, and is catching on with a lot of people. What are your thoughts about it, and do you think we will see someone in the top 10-20 using in the next 5 years, do you think it could compete with methods like CFOP or Roux?

I think it has a lot of potential. Since it’s so similar to CFOP, I think it’s going to be really big in a few years when more people know ZBLL. I was able to get to about 22s average with it in 2 days. I’d be very interested to hear Tymon’s thoughts on it. Or anyone on the forums who uses full ZBLL.

 

[OtaMota]

I think it could be pretty good, but I have a bit of trouble thinking it could be better than CFOP, at least in current times. APB is quite similar to Square-1 Methods in the fact that they are mostly algorithmic, and you do an algorithm, pause to recognize the next one, and so one. A problem I have with APB is even though it has low move count, and you can do those moves ergonomically, I am not sure how easy it is to recognize cases, like in Sq-1, it takes a bit of time to recognize the OBL or PBL. If it is easy to recognize cases in APB, I think it could be really good, but if it takes a long time to recongnize, it could take longer for it to catch on with more people.

 

[Swagrid]

 

[BradleyGallella]

Simply put it is an algorithmic, and slightly more ergonomic way to do Petrus. Through a lot of discussion on the discord server a lot of people have come to the conclusion that we'll just have to wait and see as more people learn it and become proficient at it to gain more insight to actually see if its any better then Petrus, however It seems as if most people on the discord server are leaning on the more optimistic side of that prediction in favor of APB. There are arguments from both sides. A big pro over Petrus is that it is very algorithmic, allowing for higher TPS spam, and allows for you to always have more ergonomic solutions for EO and finishing F2l, a con though is that it does require 128 EOPair algorithms and 116 LXS algorithms. LXS algs are very easy and can be learned intuitively, but EOPair is still mid solve EO recognition which is one of the main complaints with Petrus and is generally seen as very bad, however with APB there are more algs for EO. Both of these alg sets do bring up the problem of how much this will increase recognition time and lookahead. Then of course there are the 493 ZBLL algs which are really required to make both APB and Petrus viable in any way, and have their own problems with recognition/recognition time. In terms of efficiency of APB vs Petrus, its pretty much the same from as far as we can tell, there are some oversights and unsolvable problems when it comes to the calculation of the movecounts of each method, for example not being able to accurately calculate the move counts of human block building, that make it hard to reach an objective answer. However, the movecounts are most likely so close that it will not make any meaningful difference. I guess this all depends on what your opinion of Petrus is to begin with though. I have my own opinions on the viability of the method, however that conclusion is something you should come to yourself. It definitely is fun though, absolutely give it a try if you are thinking about it. Here's a link to the discord server.

 

[edward_9x]

I don't understand how some people enjoy memorizing and drilling huge algorithm sets in methods like Mehta and APB. I tried learning a very small 1LLL subset this summer and gave up at like 30 algs because I couldn't stand it.

Yeah, from a cube theory point of view APB is good. It has nice ergo, movecount and moveset. But so is Roux (even more so than APB). If you just use Roux you actually get a lower movecount than APB, equally good ergo, way less algs to learn and you'll just have a better time overall.

 

[BradleyGallella]

I don't understand how some people enjoy memorizing and drilling huge algorithm sets in methods like Mehta and APB. I tried learning a very small 1LLL subset this summer and gave up at like 30 algs because I couldn't stand it.

Yeah, from a cube theory point of view APB is good. It has nice ergo, movecount and moveset. But so is Roux (even more so than APB). If you just use Roux you actually get a lower movecount than APB, equally good ergo, way less algs to learn and you'll just have a better time overall.

I would even argue that Roux has better then just "equally good ergo" as even though the dM step and EOPair arent that bad ergonomically, it's just not quite there when comparing with any and every step of Roux. There is just no part of Roux that isn't incredibly ergonomic, but I digress.

I do not believe that the main appeal of all of these methods to be to maximize theoretical limits over all to find the "best" method, as of course that leaves out probably the most important variable, and that's the individual solver. Everybody solves differently and everyone enjoys different things. If APB fits your solving style well, then you'll probably be faster with APB then with Roux. If you have more fun with APB then with Roux you will probably also be faster with it as you'll be more motivated to practice. You say you don't have fun memorizing algs, but some people do. For example, I know full ZBLL (many outdated algs ill admit) and I have fun just being able to say I did it! Although I do prefer the elegance of roux over just brute-forcing your way through a solve with algs. It's honesty for the most part personal preference (as long as there isn't too big of a difference in the quality of the method lol.)

 

[Topher_]

I think with how many steps there are in APB, the lookahead isnt great compared to CFOP

 

[BradleyGallella]

I think with how many steps there are in APB, the lookahead isnt great compared to CFOP

The number of steps in a method has nothing to do with lookahead?

 

[OtaMota]

The number of steps in a method has nothing to do with lookahead?

Yeah, I would argue it has not the best lookahead because you do an alg at max speed, then you recognize the next alg afterwards. Number of steps does not affect lookahead though.

 

[BradleyGallella]

Yeah, I would argue it has not the best lookahead because you do an alg at max speed, then you recognize the next alg afterwards. Number of steps does not affect lookahead though.

No I agree it doesn’t have the best lookahead. Just number of steps means nothing beyond the number of steps.

 

[Topher_]

I mean it's like reco -> alg -> reco -> alg etc., you cant really lookahead inbetween those, and with how many times you have to pause for reco, it wouldn't be great.

 

[Swagrid]

You can almost definitely lookahead at least partially into LXS, but into ZBLL is a different beast.

 

[BradleyGallella]

You can almost definitely lookahead at least partially into LXS, but into ZBLL is a different beast.

Lookahead into EOpair is also just not great. Like you can find the pair but.

 

[Sion]

Most new methods are "optimized for fast turning" when in reality they are:

-Some blockbuilding step that requires above average efficiency.
-At least one very bizarre algorithmic step that has difficult recognition.
-ZBLL.

If I wanna blockbuild and have lower move count I'll do roux. If I want fast tps and flow, I'll do CFOP.

 

[Swagrid]

Lookahead into EOpair is also just not great. Like you can find the pair but.

while building pair you can see eo.

 

[BradleyGallella]

while building pair you can see eo.

Like I said, it’s rough. First of all, multi sided EO lookahead is always pretty bad compared to other steps , but also looking ahead only 1 or 2 moves during pair creation is not great.

 

[Spencer131]

The steps of petrus are:
1. 2x2x3
2. EO
3. Square
4. Last pair
5. Last layer

Now swap steps 2, 3, 4 and you get APB