The BEST Method You’ve NEVER Heard of (APB)

[@nathanaelmorgan8604]

I think this actually has potential, the main thing holding it back is the lack of a clear progression from beginner to advanced but I could see this really taking off in a few years. I'm definitely going to learn it.

[@paper2222]

it's like that method where you must already have previous knowledge of cubing in order to learn the method

[@rujon288]

@@paper2222 but it is the natural progression of advanced CFOP tymon as an example moving towards xx cross and ZBLL and doing EO mid F2L very similar to Petrus/APB

[@bruce4139]

It also seems like it would be those things with a super steep learning curve but once you learn it your progression will be bigger than ever

[@AttackOnTyler]

​@rujon288 bruh, you just wrote a bunch of letters. How would you go from cross on the 1st layer, 1st layer corners, 2nd layer, 3rd layer cross, corners fixed and then rotated to learning the beginning of what you said?

[@paper2222]

@@AttackOnTyler i think that's on you for not knowing cfop yet "2nd layer" is not a thing in advanced cfop

[@Sanseru2702]

We will soon have a solving method like "Identify which one of the 43,252,003,274,489,856,000 scrambles you have, and use the corresponding algorithm (20 steps at most)".

[@JoMama-np3og]

Aka one look

[@SelviPandian-f3x]

There's a way faster method (5 style )also could someone explain some othe advanced bld methods like zbld or ftft

[@Jakethecuberofficial]

Ye hang ( I can’t spell his name)

[@mohammedpascal]

That might be true

[@TheBookDoctor]

This is very interesting. As an older cuber whose fingers just don't go as fast, this is appealing as efficiency is a good way for me to improve. A more efficient path through the solve with minimal new algs to learn (i.e. just the EO algs) is very enticing. I'd love to see a tutorial video or series.

[@chrisr6950]

learning 700 algs isnt efficient

[@yclee-jy4xn]

​@@chrisr6950huh? it is? oh nvm I think I get what you misunderstood. he meant that his solve can become more (move) efficient. which means less moves and less moves mean less turns and less turns means a lower time.

[@Silverback947]

I'm an older cuber too. I would consider myself pretty experienced, but slow. I consistently average sub 5. Sub 5 TPS that is. In my 10+ years cubing journey I've looked into all major methods, including ZZ , Petrus and Roux, but never really switched from cfop. I started to learn coll and wv, but never completed them. This is mainly because the cases come up too rarely. APB seems to be the perfect method for me. With my low TPS, but good understanding of the cube, I might improve beyond what was doable for me before. And I might finally learn wv and coll. Thanks for sharing. And keep the tutorials flowing. Make one for the 11 EO-cases.

[@nahometesfay1112]

​@@chrisr6950That's the most advanced version

[@_wetmath_]

i didn't want to learn zz or petrus because it seemed too different from cfop and layer by layer. but this method seems fun, has potential to be really fast, and combines parts of different solving methods all into one. when i have time, i'll definitely try learning this! your video really convinced me.

[@shahar78985]

Cfop is faster its sadly not worth it to learn apb

[@idk-qt4jy]

zz and petrus are a lot easier to learn than this lol, zz is very similar to cfop

[@rebucato3142]

I've been messing around with the method and LXS is such a fun step to do. I don't know the algorithms, but there are so many times when I just cancel into an F2L case while inserting the edge, which is super satisfying. The beginner EO algs are also pretty interesting, and you could kind of think about them intuitively if you borrow some ideas from Roux and ZZ EO. I think this is a pretty good method, and I hope there are more people trying to get faster with this method and discover its full potential.

[@heathero8980]

Very interesting! Though I do feel like seven hundred something Algs is a crazy amount, if you put in the work and mastered this technique they’d be getting some insane times!

[@iskandarrakhimshaikhov2121]

The mext method: 1-look 3 layers☠️

[@mrigankakalita1943]

43 quin trillion algs

[@lc846.]

And it’s just blind solving

[@CubingEmperor313]

1l3l

[@LukeLane1984]

​@@mrigankakalita1943*quintillion (million, billion, trillion, quadrillion, quintillion)

[@rodrigoqteixeira]

​@@LukeLane1984here (europe) we'd say A million A thousand millions A billion A thousand billions A million billions Million of billions

[@archiehandler1292]

as the 3x3 world record holder, I can say with certainty that this video is true

[@rax1899]

wow rlly omg that’s so cool, 你会说中文吗？

[@archiehandler1292]

thank you I worked really hard to get there and it means a lot to see people recognise me :)

[@tperm99a]

oh m gee its max park

[@raymok5644]

@@archiehandler1292Bro Max Park holds the world record

[@archiehandler1292]

this is a common misconception so i don’t blame you for making this mistake

[@curiousNic]

Sounds cool, can't hurt to give the BEST method a try. Plus I have time to learn a few algs during the summer.

[@wolfonic]

Wow! This method seems like it has a lot of potential. However, I'm refraining from learning it (Partly because it has so few resources and mainly because of the obvious difficulty- I mean, 700 algs, I've just barely learned full OLL and advanced F2L.) I'm glad you've made a video on it, though; You've exposed the method to many more people, and it truly does seem like it could become mainstream in the future. I think that you should make a more in-depth tutorial, as you suggested! It's always best for the cubing community to keep innovating and changing. (But are there any noticeable flaws to the method- or in other words, aspects or benefits that other methods have that APB doesn't? Maybe you could mention those if you made another video.) I'd also like to say that this video's quality is immaculate for under 1K subs! I was shocked when I saw your sub count. You did a great job on this, I'm looking forward to another video (and I'm sure other people are too.)

[@LiamHighducheck]

Thank you so much! Fear not for I’m already working on the beginner tutorial. There are of course some flaws with the method but I believe they can be fixed. For example one problem is alg recognition, especially ZBLL but we have someone who created an alternate ZBLL recognition system which is easier to do ZBLL prediction with. Although it’s obviously quite hard to test this stuff when no one’s at the level where it becomes necessary yet.

[@OfficerDonut0]

It is crazy to the of the possibility’s though because since it’s so alg based and good move turning it could make people like max park turn 16+ tps and get a 3 second avrg

[@amirPenton]

Awesome video and it’s a very cool method. Highly algorithmic methods are just not for me, but for someone with the time & dedication to learn the algs this seems really worthwhile & a very cool method. Thanks for sharing

[@danielstoica647]

Bro your chanel is extremly underrated, you put so much effort in your videos and they are actually really good videos, big respect👍👍👍

[@LiamHighducheck]

Thank you! I’m glad you like them

[@janmagtoast]

Last layer can still be solved with coll and really fast edge pll cases which is pretty nice and only 50-ish algs

[@PureRockerHK]

In some ways, I feel like APB is quite similar to Mehta in its heay reliance on algs. In the future, I can think of one possiblity is that we might change and adjust more to heavy algs set, the only thing left to improve is how we should develop more beginner-friendly approaches for new solvers and most importantly the transition from basic LBL method to advanced methods like the aforementioned example, or from CFOP and Roux to those methods. I feel like in some way this method doesn't have a very good look ahead compare to CFOP because of the blind spots at the first step and doing it like Roux maybe is the only optimal way to do so. The only thing I would like to point out is the look ahead aspect since you have already covered the ergonomics and move counts. It is really nice to see some small groups this community who pioneer research to construct new and possibly better methods to replace and improve CFOP. I will start trying this method soon.

[@rujon288]

There are no blind spots in any parts of apb if you plan 2x2x3 in inspection which you really should if you are looking to seriously do APB

[@LiamHighducheck]

Thank you! It’s funny you mention Mehta because APB was originally designed to be an objectively better version of Mehta-TDR and it was proven to be better in almost every aspect (I believe there is a forums post which compares 2x2x3 systems you can check out) which lead a lot of people who were learning Mehta (myself included) to drop it and switch to APB. The lookahead isn’t really a problem in the first steps because planning 2x2x3 is more than doable, it’s more so in the later step with alg lookahead which can become difficult and some new prediction methods may need to be developed (particularly the LXS to ZBLL transition)

[@PureRockerHK]

@@LiamHighducheck Thank you for your respond, i can't believe this has been designed to be better than Mehta lol.

[@rujon288]

@@PureRockerHK ye the original mehta method promo vid actually links APB as the methods successor in the des.

[@PureRockerHK]

@@rujon288 no wonder why they are so similar in certain ways XD

[@rujon288]

I already knew about and love APB and plan to switch to it when i finally finish ZB because there's no point switching before full ZB

[@OneDerscoreOneder]

Before full zbll you mean right?

[@milomaher1582]

​@@OneDerscoreOnederZBLL is part of the ZB method, along with ZBLS which is a way to solve last slot and edge orientation at the same time

[@OneDerscoreOneder]

@@milomaher1582 yeah... I know. I know all of zbll and half of zbls, you don't need to explain them to me. I said that because there's no reason to learn zbls (the other part of zb) before learning APB since APB doesn't use zbls. Hence why I was questioning why he would need to finish learning all of zb before APB since the zbls part of zb isn't useful for APB.

[@willlagergaming8089]

Full zbll is not necessary. So just learn eopair and LXS

[@rujon288]

@@OneDerscoreOneder Ye i call ZBLS EOLS so i can just call ZBLL ZB so ye i meant just ZBLL

[@disaus]

3:36 actually if you can just do the R move before the algorith to bring the pair on top the created pair will allways end up in 2 locations. That reduces the number of cases.

[@cyrildubois1815]

For cfop it also 50 coups. If you learn Xcross and neutral face, 7-9 can make a pair. Also you made two pairs ( F2L) and it's like 10-12 moov. After you learn winter and summer variation, it's like 7-9 moov, OLL easy case, 8-9 moov and pll 13-14 It's 50 but without learn 1000 alg

[@LiamHighducheck]

Bro where did you pull those numbers from. Also you can’t use summer and winter variation every solve cause they require EO to be done and if you do use them you don’t also do OLL the same solve cause the whole point of summer and winter variation is that it skips OLL. F2L is up for debate in terms of movecount but if you just take Yiheng’s average movecount for F2L in his reconstructed ao100 it’s 32.5 moves. Then you forgot about AUF while calculating OLL and PLL so OLL is actually about 10+0.75 and PLL is 13.5+1.5 so the total for CFOP ends up being 58.25 moves on average. Even taking Kian Mansour’s theoretical estimate of perfect CFOP where he estimates F2L as low as 29 moves (8 move Xcross+7 moves per pair) that still only gets you to about 55 moves on average.

[@cyrildubois1815]

@@LiamHighducheck yeah but some alg (7 only) can help you. (W variation or summer variation + one of then ) and you had done the yellow face

[@LiamHighducheck]

@@cyrildubois1815 I don’t think you know what winter and summer variation are. Winter variation is a set of 27 algs that inserts a built pair into your last slot and solves OLL when and only when you have all the edges oriented. Summer variation also has 27 algs and does the same thing but with a split pair. I have a theory that what you actually mean by winter and summer variation is ZBLS (302 algs) or rather it’s simpler version, VHLS (32 algs) which inserts either a built or split pair while orienting the edges. This would also make sense as you’re saying there are 7 algs you can do after that step which fits with OCLL (OLL with edges oriented (has 7 algs)). So the overall method I think you’re talking about is F2L-1 > set up pair > VHLS > OCLL > PLL Which using perfect F2L numbers gives a movecount of 22 + ~4 + 9.5 + (7.9 + 0.75) + (13.5+1.5) = 59.15 which is actually worse than standard CFOP. I think this must also be what you meant because in your original comment you said OLL was 8-9 which fits with OCLL. Either way still not close to 50 moves.

[@ErenMichelet]

As a roux solver I’m curious about the planning part of this method? For roux planning First Block + DR consistently is something recommended for people in the sub-12 to sub-10 range, so adding another piece seems really difficult, and I feel like the probably slightly wonky finger tricks don’t help? Also, is the beginner method version of this any different than just doing eo then petrus? Is there a meaningful difference between making the pair before or after eo? Overall it looks cool, and seems around equal to CFOP and Roux. Subjectively is there difference in fun/style you see in this method that would make you recommend it? The differences in style between CFOP and Roux are obvious, but this seems like CFOP for someone who loves algs, which I can see being popular in ZMS server lol. I don’t mean this as an insult btw, but just how does the solve feel to do?

[@LiamHighducheck]

It’s hard to say since I came into APB with a pretty strong foundation in CFOP and my inspection planning was pretty good. I don’t think planning DFDB will be too difficult for solvers around the sub 10 area but once again it’s hard to say since the method is quite new and we don’t have a lot of users. Even I didn’t have a background in Roux so my FB leaves much to be desired. Another thing to note is that FB>DFDB is only one way to solve the 223 and it’s possible to find some more efficient solutions that partially combine the steps, so it’s not strictly FB>DFDB which can make planning the entire 223 in inspection easier. I would like to do some analysis on DFDB finger tricks, including perhaps making some sort of intuitive way you can solve every case within 1 move of optimal. The other solution is just algs which is less elegant (and will probably only be learnt by the most dedicated individuals) but it gets the job done while guaranteeing the best ergonomics/fingertricks. In the beginner version solving the pair does make a large impact. The difference between recognising EO with 2 F2L edges vs 1 F2L edge is large and only having 1 F2L edge means it can be tracked and therefore EO recognition becomes instant. Also making the pair in the beginner method makes sense for transitioning into full APB. In terms of fun/style I can say it’s very fun to use. Since the entire method is optimised algs after the first few steps, you get some really satisfying solutions because of how good the ergonomics are, particularly in LXS which is an amazing alg set. In most methods with large semi-intuitive steps (like F2L or SB that are partially supported by algs) it usually takes a solver a lot of practice with the method to start finding the speed optimal solution and sometimes even the most advanced solvers will miss something good, but having alg steps means a beginner and the most advanced solver will be doing the same most ergonomic solution.

[@BenTheCuber]

Amazing video!!

[@SparshW]

3 LIKES?!

[@user-ed4fv9nd3b]

To increase your TPS, make sure you're using the new cover sheet.

[@jroot1279]

Sounds like somebody’s got a case of the mundays.

[@TriCubing]

cool method! thinking of trying it but i think i’ll just use oll pll for last layer

[@ugwuanyicollins6136]

CDRLL and L5EP is better

[@ugwuanyicollins6136]

APB CDRLL

[@ugwuanyicollins6136]

After EOpair, you solve the front F2L pair (DFR+FR) You solve the last layer with CDRLL (42 algs) Then solve the last 5 edges with L5EP (16 algs)

[@ugwuanyicollins6136]

Solving the last layer with OLL and PLL defeats the purpose of the method.

[@Drizzz32]

@@ugwuanyicollins6136its easier 😂

[@HolmesHobbies]

I originally learned the petras method, not knowing any other choices (and not having instructions) around 2003. This is a fantastic tutorial !! Now that im learning other cubes ive been learning other styles and moves with different operation orders

[@Xzy_158]

This sounds like a slightly altered, more algorithm oriented version of petrus. I can see this being popular but it seems intimidating to learn with the large algorithm sets. Will definitely give this a shot, thanks for the vid

[@Andlekin]

I know Summoning Salt has a great choice in music, but there are a lot of synth lofi tracks to choose from.

[@LiamHighducheck]

Can you give me some suggestions?

[@willlagergaming8089]

Great video. Apb user right here, been using it for 6 months

[@LiamHighducheck]

Thank you! Good luck with your future APB endeavours

[@cariyaputta]

2x2x3: 15 FBEO: 5+7=12 LXS: 9 ZBLL: 14 Average 50 moves, which comparable to CMLL+intuitive Roux or optimized CFOP. The planing during inspection is equal to CFOP Cross+FP and harder than Roux FB+DR. While Roux M slice is problematic, this method is just a worse Roux with S slice abuse. But still I'm looking for a global sub 6 stackmat APB user.

[@forgotten.mp3]

Then I'm your guy, was sub-10 global with CFOP after 2 years, now after four days of beginner APB, I already have this: 7.58 F' L' U2 R' B2 U2 B2 R D2 F2 R' F' R D' L2 B R' B2 D x' z // inspection D2 x' U2 R U' R' U R U' R' U F' U' r' U r // 2X2X3 U2 R U' R' U R' U2 R U' R U' R' U2 F R' F' R // EOPAIR U2 R U R' U' R U2 R' // LXS 40 moves, 7.58 seconds = 5.28 tps

[@GWAIHIRKV]

I think I’ll stick with - less than 2 minutes! I’m in my 70’s and just can’t be bothered. . . . .

[@Bmxmusikian]

JEEEZUS 😂, im on my 2nd week n got down to 1:32, thought I was doing well so went on youtube to look at how to get quicker, had no idea how advanced it could get!

[@EliTheFri]

Yo how’s it going bro ru still cubing

[@fallofshadows2209]

As a Petrus main, this sounds like torture. But then again, I don’t like learning algs haha. Your comment about people who lack finger tricks hit me though. My turning style hasn’t evolved since 2007, and my wrists hate me for me 😂

[@The4gotNdeath]

As someone who has never learned anything more advanced than the beginner method, not sure how the yt alg placed this in my feed. I was impressed nonetheless.

[@KefashWhite]

Just lean to solve a 3x3 cube two days ago. Now I feel it is going to take me 10 years to become even half as good as what I just watched.

[@pessimistic5579]

as a cfop user who averages around 25 seconds, im going to give this a shot. if i can match my times with the beginner or an intermediate variant, i might go for it

[@bl00dy-ca76]

how'd it go?

[@pessimistic5579]

@@bl00dy-ca76 i didnt end up giving it a shot and i still avg 20-25

[@Arcticgdnmore]

Good video and very persuasive, but I think I'll stick with cfop, because I'm bad at doing the 2x2x3 and Eo seems difficult. These are also the same reasons why I don't use roux

[@zakattube]

@LiamHighducheck What's the reason you specify the counts of OLL and PLL to be only 7 and 21 respectively at the end of the video while more algorithms exist?

[@LiamHighducheck]

Do you mean for the beginner variant?

[@ekhau1977]

I think you meant 2 look OLL and PLL. I use 17 PLL only, breaking down G perms to 2 PLL 😂 Old man can't remember much algs

[@janaki3829]

ZB's average movecount is 45.43, and its alg count is 772 I don't trust myself to analyze too much beyond that, but I think it's worth throwing out there

[@ugwuanyicollins6136]

ZBLS is bad

[@ugwuanyicollins6136]

CFOP with ZBLL every solve is called ZB and orients the edges during LS in a step called ZBLS. ZBLS is not a very good step, not to mention people often forget it has to be done from more than just the FR slot. APB will also offer more consistency. Due to being mostly algs, you’re always going to be doing the best solution for every case and each of those solutions will be algorithms that you have practiced and drilled to provide high tps. Credit: LIAM

[@rujon288]

@@ugwuanyicollins6136absolutely. I hate FR dependant methods just cuz doing wide d in between Steps is the worst for lookahead

[@LiamHighducheck]

Certainly a worthwhile comparison as at first glance the methods seem similar. A lot of stuff has already been mentioned about ZBLS and its flaws but I’d also like to point out something about that movecount and why you should be careful taking the wiki at face value. That 45.43 assumed an average ZBLL movecount of 12.85 which is close to if not move optimal. The ZBLL algs used to analyse APB are about 3 moves higher, at 15.7. For ZB, the reason move counts like that were used of course was because it was analysed in practice, by hand over a slow turning Ao100, in 2005, where the efficiency/TPS balance was wildly different than it is today (not exactly the height of objectivity).

[@jura3760]

Bro i used to cube every day like 3-4 years ago then it stopped being intressting for me and i stopped but once in while i come across a speedcubing video and after watching this i was asking myself wtf has happend to speedcubing

[@LiamHighducheck]

Honestly this video was supposed to be really niche and had no business getting as many views as it did

[@InfinityCuberRS3M]

please make tutorial, this might not be great for me for 2h, but for OH i am willing to look into this

[@nathanaelmorgan8604]

Roux is best for OH, there's no way you're going to learn to optimize all those algs for OH

[@Arcticgdnmore]

Wouldn't this be pretty difficult for OH because you have to do algorithms that use middle slices?

[@lmcubing]

@@Arcticgdnmoreyou use the table to do M moves, just look at good roux OH solvers

[@PiercingSight]

I only have 5 algs memorized, and that's for the last layer, lol. Even doing 20-30 algs would already be a stretch for me.

[@ekhau1977]

Just less than 30 algs for last layer, and another 12 for 2x2 Ortega method, easy peasy lemon squeezy 🎉

[@AmbisinisterSSBM]

super cool video, well structured and interesting. Hope someone takes this method and runs with it

[@Stab.a_Saurus]

been cubing a while but then i just stopped some day, this vid has inspired me to keep going

[@Stab.a_Saurus]

(i subbed)

[@LiamHighducheck]

Good luck!

[@WangYiheng347]

Didn’t expect the video to be this good

[@LiamHighducheck]

Edited on mobile 💪💪

[@WangYiheng347]

@@LiamHighducheck wtf actually?? I would not have the patience for that holy

[@KotegawaYui]

heyrrison

[@FlummoxTheMagnificent]

You use the same alg generator as me lol. I’m also working on a method that should rival this (just as algorithmic but lower movecount and less than 200 algs)

[@norsitoang7714]

I have 2questions, what other things can we improve/optimize on this method and what do you think about doing CDRLL and L5EP instead of LXS and Zbll.

[@rujon288]

whats wrong with LxS and ZB not only are they developed highly but they are also proven to be pretty optimal

[@LiamHighducheck]

So we’ve actually put hours into brainstorming new variants or ideas then genning and testing alg sets for different ways to solve things or different sub steps but most of them just can’t beat standard APB. For an extremely advanced solver their main improvements will come from planning further in inspection, solving pseudo blocks, drilling algs for faster execution, and learning alg prediction. Alg prediction is probably the biggest thing we are currently looking into, specifically with ZBLL as that is the hardest pause to overcome. One of our members developed a new ZBLL recog system (Tv2) that could allow us to track some useful information during steps like LS or LXS. But obviously no one has gotten near the level where that stuff actually matters so it’s hard to test ideas. We are still working on finding other optimisations where possible. For CDRLL and L5EP it is just a worse way to solve things when compared to LXS and ZBLL and is kinda just there for fun. The main reason it’s a variant is because it was a variant in Mehta and when all the Mehta users switched to APB and development started occurring in APB the CDRLL variant was bound to occur.

[@norsitoang7714]

Ok thanks

[@bruce4139]

Ngl when I saw this video thought it was a jperm one, glad to see we have someone new in the mix, I'm kinda wanting to come back to cubing as well as learn this, I'm curious to how popular this could become as well

[@qutubjed1845]

what's the software used for the cube algo move counter mentioned in the video?

[@LiamHighducheck]

Batch solver which was created by Trangium. Ryan Hudgens has a tutorial on how to use it (has a bit of a learning curve but after you gen your first alg set you should be fine)

[@rujon288]

do you know if Trangium was the first person to come up with MCC or was it something they just incorporated into the batch solver?

[@LiamHighducheck]

Trangium made the MCC then later on he made batch solver and integrated the MCC in. I would say he’s not the first person to think of the idea since Jperm had this one video where he did a simplified version of it but I’d say he’s almost certainly the first person to actually make a good software version optimised for a large amount of algs

[@rujon288]

@@LiamHighducheck ye although I’ve found it to be a bit weird in its ranking when it comes to things greater than 3-gen

[@haydenlesser9725]

I’m not sure this is the best method, but I’m very interested in learning a completely new thing and I think it would be interesting to be along for the journey of this

[@abdelilhmanflores]

I’ve been solving CFOP since I was 15. I’m now 19. This honestly looks like a lot of fun and I really want to work on efficiency. Also what cube are you using in this vid? It has a very nice sound.

[@stevefromminecraft2466]

What is the alg if all edges are good and your pair is not inserted yet? i somehow cannot find it in the EOPair Doc

[@LiamHighducheck]

There are 4 sets of EOPair, Oriented Pair U, Oriented Pair R, Misoriented Pair U, Misoriented Pair R. The 2 sets where all edges can be oriented are Oriented Pair U and R, and the algs for both of those will just be U’ R and R2 respectively.

[@salomille4709]

was about to send this to you then saw you posted it. shocked.

[@LiamHighducheck]

No shot KZbin raw recommended you an advanced underground method overview

[@willlagergaming8089]

Can you do lxs tutorial video?

[@LiamHighducheck]

Potentially. I’ve got a few ideas for the some videos but in the mean time I made a sheet that groups the cases by what the alg does so it makes them easier to learn. It’s linked in the mega doc (LXS learning guide).

[@hideralduhami]

Alg methods are faster than intuitive methods

[@ugwuanyicollins6136]

Alg heavy*

[@squirrelcubed31415]

I will potentially switch, I do quite well with algs

[@GünMeteUysal]

Time is not movecount/tps, you should include pauses too!

[@LiamHighducheck]

It’s still movecount/tps, pausing just means your tps is lower. Also all pauses are skill issue except ZBLL and even then you can do partial prediction with tv2

[@GünMeteUysal]

@@LiamHighducheck Thanks! Now I got it.

[@disaus]

Finally someone who uses Batch solver instead of cube explorer

[@omarbarbar1078]

As a GD player i can confirm abp is hard to memorize

[@fantasticfuze646]

Good one

[@rujon288]

It’s ok it’s not may so no compulsory maymory levels for me

[@molybd3num823]

last wave flashbacks

[@pro.cuber_0]

Dude... nice editing and background music

[@purplefurture6960]

How many methods to solve 3x3 ? I know 1 method and memorized it but take 5 mins First do the white + Then white corners Then mid edge Then yellow + Then yellow edge Then 2 methods to fix the last layer forgot its name

[@guiwarr]

There are so many methods methods are being created every day

[@iliyashapirov7720]

Do you know of any alg set that does beginner eo (solved pair eo) and cp simultaneously to make the rest of the solve 2gen?

[@LiamHighducheck]

I’m not sure if we genned it, I think we were having problems with batch solver since you can’t really input a way to do 2-gen reduction. But the algset itself would be questionable since mid solve cp recog with more than 4 corners is brutal

[@iliyashapirov7720]

@LiamHighducheck I think learning some of the best cases may be worthwhile like winter variation in cfop. Recog seems at least possible in easier cases since there are no blind spots. If you find out anything, lmk

[@ArthurCooper1]

Well i cannot find any videos related to this method and idk where to get those algs cuz im not sush an intuitive person i was never

[@LiamHighducheck]

If you check the description there’s a link to a mega doc with all the alg sheets.

[@ArthurCooper1]

Thanks for the docs and also thank you cuz you replied thias is the first time a content creator replies to me i really aprrciate that

[@Kyungo_]

Step three for those who want algoritgmless,insert the pair and do eo

[@ugwuanyicollins6136]

So if yiheng or max or ruihang masters this method sub-4 avg. Is guaranteed 🤯

[@ugwuanyicollins6136]

Few years ago, i would have said sub-5 avg. is impossible

[@rax1899]

Well, 48 moves means u need over 12 tps per solve, it’s not impossible but still hard

[@TMBDucks]

Solving style is another thing to consider. Just because it is theoretically the best method doesn't mean it will be the fastest for everyone.

[@CubingMania]

i might learn intuitive apb so that i know it. idk if ill switch, but i think that this is a great video.

[@joshinon6980]

I’m 1:52 ao12 with beginner method, learnt cubing 4 days ago. Should I switch to this, or should I learn some cfop algos first

[@MarcuustheMoose]

I'm barely ahead of you in our journey, I just switched from beginner to cfop yesterday. I suggest getting good at beginner, then learning cfop. After that, you have a good grasp on how the cube works and what does and doesn't work. I'd finish learning cfop before you move on, but again, I'm still a beginner, only having started cubing 20 days ago.

[@LiamHighducheck]

I’d say it’s not a bad idea to try and learn CFOP first because right now this method is severely lacking beginner resources

[@rax1899]

This method probably isn’t that great for a beginner

[@joshinon6980]

I figured. Tks everyone for replying

[@mariacristinamagtibay9918]

Here's the thing I'm trying to do It's crazy I know but I wanna 1 look f2l With that I can develope an alg set that turns any paired case to an oll cross in any angle And with that I can also see the ZBLL which means I can 1 look which is possible I know it I just don't know how to yet

[@LiamHighducheck]

Lmao me too. I’ve already started work on systems to memorise the edge cycles that occur during a LXS alg to predict ZBLL. I’ve also started learning Tv2 recognition for ZBLL which is a system that I think is much more suited to prediction/1 looking. Also I think the alg set you’re thinking of is VHLS which is a sub set of ZBLS which is one of the 2 big alg sets in the ZB method.

[@channelnumber52]

Do you think Lexus could be useful for someone who solves with Petrus? I think EOPair looks awful. But, it looks like learning some nice Lexus cases could help certain Petrus solves. I am trying to get sub-15 with as many methods as possible(I currently have Petrus, CFOP, CFCE, and ZZ. My Roux average is like 17). I like Petrus and think it can be faster.

[@rujon288]

EO pair is limited but i think the ideal advanced Petrus solver would combine steps like EO and 2 gen block-building to what is intuitively optimal. this would sometimes manifest with EO pair + LxS but it wouldn't be constricted to that and i know that kinda defeats the whole 'algorithmic aspect of APB but the algs in APb are extremely intuitive anyway.

[@LiamHighducheck]

I think it could because if you’re in the EO223 state, solving the back right pair then doing LXS is the most efficient 2 step way to solve the right block (we did some tests on this a while back). EOPair is a great alg set though, but of course recognition is a massive learning curve and the finger tricks can be quite complex. Good luck with your sub-15 journey!

[@davidchapper6120]

Hey @LiamHighducheck , great video, inspired me to try to create my own method. What is the algorithm analysis software mentioned at 6:22? Thanks in advance!

[@LiamHighducheck]

It’s called Batch Solver, it’s made by Trangium and Ryan Hudgens has a tutorial for it. It’s very useful

[@davidchapper6120]

@@LiamHighducheck thank youu ❤️

[@glebnekrasov1]

There are so few tutorials on this method but I really want to learn it.

[@BalthazarMaignan]

I will definitly look into this method, this seems very good 😎

[@ur_local_goomba]

Just wondering, how do you get the cubes in the thumbnail?

[@LiamHighducheck]

Images were made with visual cube. If you look up “cube rider visual cube” it should come up

[@ThamosProductions]

what would be the advantages of this over full zb? seems like the same number of algs but slightly less optimal due to a more specific solving order if that makes sense

[@muyou0107]

This is better than ZB in every aspect.

[@LiamHighducheck]

ZBLS requires always forcing the FR slot unless someone wants to learn a massive number of algs for all four slots, or rotate. APB's specific solving order doesn't hinder the solve, it actually improves it since the intent is to solve the back pair to have good LXS visibility. To have a set solving order for ZBLS to always end in FR would be a negative impact. You would be losing the important freedom in F2L, making it less efficient. So the specific solving order creates optimal solutions in APB and creates less optimal solutions in ZBLS. APB has fewer algs. EOPair + LXS is 244 algs. ZBLS is 305 just for the FR slot. APB has better algs. EOPair is under 7 moves on average and the algs are great. LXS is amazing. We just did an analysis on the ZBLS algs movecount and ergonomics and they are worse than both EOPair and LXS.

[@rujon288]

2x2x3 is the most efficient way of blockbuilding and way >>>> than F2L. F2L is really fairly restrictive in what it allows you to do in the intuitive stage of a solve.

[@thegappleapple7763]

Ah yes. Time to learn 600 algorithms

[@thetwistedcube]

Very high quality video! For this i subscribed!😃But I will probably stick with cfop.

[@fmga]

Or you could just use petrus (But fr good video, interesting idea)

[@Miner13-rk8qx]

at first i thought it was click bate. I will learn it. thanks!

[@mid_bet0cubing430]

Did anyone get a pb with apb?

[@DavidThomas005]

I’m at the point where I know I’ll never be exceptional at speed solving so I’m gonna be the guy who knows the random ass method nobody uses 😎

[@idratherbecubing]

Here’s an interesting idea. Instead of doing LXS and ZBLL, it might be faster to do CPLS (Corner Permutation + Last Slot) and 2GLL+1. Not sure how good this is for recog, but it might be fast.

[@LiamHighducheck]

Generally doing CP anytime before last layer has very bad recognition and I have my doubts about the quality of 2GLL+1, but I like the brainstorming

[@OreKehStrah]

@@LiamHighducheck CPLS recog is just being able to do COLL recog from any angle. IIRC someone in ZMS looked into 2GLL+1 and it wasn't very good/useful

[@NOONELIKESTHISFEATURE]

Ah yes, the "A Bizzare Phantasm" method

[@archiehandler1292]

I’m already in the 95th percentile for viewers subscribing let’s goo barely even had to try

[@gewinnste]

If you' planning to learn that many algs, I'd suggest to instead learn ZB (CFOP-start (cross+ 3 F2L-pairs)-->ZBLS-->ZBLL). It's just 5% more algs, but has an even lower move count (a bit higher than 40) and arguably the best inspection-planned portion of all methods (x-cross). But anyways, any method with ZBLL won't be superior to CFOP, at least for averages, because the case recognition takes so damn long and most of the algs are a disaster. Most people think ZBLL is equivalent to getting a PLL-skip every time, but it's not - AT ALL. Instead, ZBLL is equivalent to getting an OCLL-skip, e.g. skipping a Sune or T-OLL, and then getting one of your *worst PLLs every time* ! (Avg. ZBLL is 2-3 moves longer than avg. PLL, plus, less streamlined algs). It saves just ~ 5 moves (not 12, like a PLL-skip) but the longer recognition time kills even that advantage. Plus, there's the what I call the "upkeep": You have to invest several hours a day just to keep your ZBLL-recognition and execution at its best, otherwise you're making mistakes (see Tymon's failed Ao5 sub-5 attempt and he's one of the best pro-cubers, so if that happens to him ...) or you mis-recognize cases or stare at the cube for 2 full seconds to finally recognize the case, etc.

[@LiamHighducheck]

ZB is not very good and definitely does not average just over 40 moves. The first problem with ZB is ZBLS. ZBLS algs are not very good, we’ve done analysis on them and regenned speed-optimal algs for the FR slot and ZBLS is worse in movecount and ergonomics than LXS and EOPair. Furthermore ZB requires you to end with the front right slot every solve, so a ZB user will either have to learn a large amount of algorithms to use ZBLS in every slot or rotate. Forcing the FR slot removes the important freedom F2L offers and will reduce efficiency. Taking numbers from Kian Mansours seminar where he assumes 8 moves for an Xcross then 7 moves per F2L pair we would get to F2L-1 at 22 moves. Then 9.16+0.75 for ZBLS and 15.7 for ZBLL gives 47.61 at best assuming you have the world class knowledge and skill to make an ergonomic 22 move F2L-1 vs APB’s only non-alg step requiring a 12 move start to average the same thing. To further add insult to injury, notable users of ZBLS such as Tymon and Tao Yu have begun unlearning cases they have found aren’t worth it. ZBLL recognition is naturally harder than something like PLL, but not by as much as you would think. Tymon averages around 0.5-0.6 for recognition. The main reason PLL recognition of top solvers seems so much faster is because of PLL prediction. ZBLL prediction is still in relatively new territory and is something that hasn’t been looked into a lot due to no cubers that use ZBLL every solve getting to a high enough level where it matters, but this is absolutely something we’ve thought about. One of our members developed a new ZBLL recognition technique (Tv2) which allows applications for prediction during a step such as LS or LXS. Still of course recognition will take longer, but ZBLL will still be faster overall because, OCLL averages 8 moves and PLL averages 13 moves, + 0.75x3 for AUF gives you 23.25 moves compared to ZBLLs 15.7 which means ZBLL saves around 7.5 moves which is more than enough to make up for any losses in recognition. Furthermore, characterising ZBLL as getting your worst PLL every time is inaccurate as depending on what algs you use, half of PLL is over 14.2 moves. Furthermore the best ZBLL algorithms are better than the best PLLs. But obviously movecount isn’t everything, it’s about speed, and in terms of speed we have people like Daniel Egdal who has timed 98% of his ZBLLs to sub 1, and 46% to sub 0.9 and even to my personal experience I would say the algs are far from a disaster. Regarding upkeep, it takes nowhere near several hours a day just to keep ZBLL recognition and execution at its best, in fact you don’t even have to train it daily or even weekly. Like all algs sets it gets to a point where you just know it, there are just simply more algs so it takes longer to get to that point which is why people dedicate time in the beginning to drilling cases and using alg trainers because that way you can master the alg set faster than just getting better at the cases through solves. Mis-recognising cases or pausing to stare at the cube for 2 seconds is what happens when you’ve just recently learnt the algs. After you get comfortable with them it doesn’t. People messing up while using ZBLL isn’t as a result of ZBLL being a bad algset, it's a result of not having practised it enough, which keep in mind if you’re trying to be one of the best cubers in the world, putting that much time into practising shouldn’t be a problem and if anything should be expected. Recognition also gets better when you use a method that always ends in ZBLL because of priming which is why it can be worth it to learn sunes and anti-sunes in a ZBLL ending method but not in CFOP.

[@superdupersir7054]

@@LiamHighducheckI ain’t readin allat 💀💀 But yes true

[@gewinnste]

@@LiamHighducheck ​ Sorry for the delay. I think I remember ~40 moves for ZB from Zbigniew's website. He obviously underestimated the number to advertise their method. I had a feeling that was too low, but arrived at an estimated ~42-44 back then. And I forgot that all that came from a time in which they used one of the first ZBLL alg sets, which were focussed on low movecount vs. low execution time (which turned out to be a bad idea) so I think the avg. ZBLL movecount was ~12 moves. Also, back then, they didn't include the 0.75 moves for each AUF. So you're right, the ZB movecount is apparently just a bit lower than that of APB. Regarding the move count savings of ZBLL: PLL has 12.8 moves on average, accounting for alg frequency, and I found ZBLL to have 14.55 moves on average (I saw this on Tao Yu's alg trainer page, but now the stats seem to be gone. Anyways, I wrote them down in a file with my alg sets), so it's 23.05 (OCLL, PLL) vs. 16.05, i.e. 7.0 moves less. But PLL is much more streamlined than ZBLL - only 21 algs were optimized for over 20 years now and the global optimum is fairly surely reached. That can't be said at all for ZBLL. It will take a very long time until ZBLL will be at the state of current PLL. And until then, ZBLL algs of move count x are just worse/take longer to execute than PLL-algs of move count x. So it's actually a bit less than a 7.0 moves advantage. So now we have to -compare- consider the execution time of a bit less than 7 moves of OCLL/PLL, which for pro cubers -is roughly at 15 TPS (or a bit higher), so- , for which I estimate an algorithm-TPS of 15 (which is a conservative estimation), is 2gen-algs) like adv./exp./rotation-less F2L or VHLS to even work on the fly (F/B-mirroring obviously doesn't work on the fly), i.e. without ever having practiced them for the FL-slot. I always thought they're using both front slots for ZBLS, in which case you're actually as flexible as it gets because the backslots should be solved first anyways in CFOP. Another issue, and an important one, is color-neutrality and that is much, much more difficult with ZBLL than with PLL (ZZLL is already much harder, but a breeze compared to ZBLL, because you only have 4 cases per COLL, vs. 12 for ZBLL). Tao Yu in his video was apparently dual neutral in 2022, and with a strong preferrence for yellow cross over white cross (10 y, 2 w). Are there any CN-ZBLL cubers out there and what are their recognition times? Edit: Sorry, I messed up two phrasings - I crossed them out, followed by the correction, so the (wrong) original is still available.

[@theonlybass8947]

@@gewinnste feliks, tymon, and juliette are all comfortable with most ll colors if not full cn from what i've seen, and as far as I know feliks knows TUL, tymon currently knows TULHPi, and juliette knows full. Recognition is not an issue if you use a regulated system like tranv2 or baum-harris since all you need to recognize are the color relationships between a corner and 2 edges after recognizing corner permutaion, and methods like tranv2 make multi-angle recognition significantly easier than blocks or bh. Viability of zbll for not just cfop but other methods has still yet to be proven given how few top solvers are actually committed to learning and applying it. Also I am inclined to believe that your experience with zbll might be a tad skewed given you skipped learning literally any good zbll cases in favor of zzll, which is categorically worse than unphased and can be avoided with antiphasing lol.

[@gewinnste]

@@LiamHighducheck Should I expect an answer or should we leave the discussion as is?

[@litcuber1854]

Can you make a tutorial on EO, since it seems to be to most important and tricky step coming from a 13 second CFOP solver.

[@alika3651]

this sounds so promising ,makes me wanna pick up the 3x3 once again.

[@skellq4385]

Feel like the lookahead and inspection for this method must suck though

[@Lemueladriano]

When i saw the thumbnail i thought it was j perms video lol. Btw great vid!

[@hyperjanny1510]

bro i use a bit of this

[@mrnimblefingersgaming]

Honestly just learn every zbll and winter variation and you’re good

[@HBblazas]

Can you do a video on the 11 EO algs? I want to learn this

[@ferguson1376]

Do vids for fingertricks please Liam

[@cwcoc170]

I am imterested, but I would need more input to decide, if this really makes sense. So I would love, if you make more vids about this method

[@SavciSV]

I was holding the cube before video starting and when it started i found u holding it from the same position

[@rakib_hoque]

I like to call APB "absolute personal best" like every time you do a solve you're gonna get your pb 😊

[@tomdumon9189]

does anyone know if there would be a kind of mathematical formula that can 'calculate' the absolute minimal moves to get to the solution?

[@ironclickshorts]

Yeah but it takes some time to calculate the best possible solution

[@cubilatic3]

Cubing With Alex proves you wrong...

[@guiwarr]

He didnt even learn a single LXS case, thats why he didnt achieve sub 15 in a month

[@sethzero47]

idk, i tried learning beginner abp and is quite good with barely any practise but it takes the fun out of it somehow and idk why, so as good as it is, im not that serious about this and im probably gonna stay with cfop

[@disaus]

Here before APB method becomes very popular

[@Inti-g4u]

I need a tutorial

[@dume85]

is this a rubik's cube video or a summoning salt speed run video?

[@SwagridCubing]

baller method tbh

[@goldenwarrior1186]

I like how the method is literally called “a PB”

[@EaglesFanCTC-d6b]

Using ZZ wil have the same movecount with less algs