The New Method / Substep / Concept Idea Thread

[Nir1213]

Care to DM me about this, so we do not clutter the thread? I cannot seem to get the video to play on my phone at the moment. I will try again in a few minutes.

ok then.

 

[PapaSmurf]

also new idea:
do cross but its oriented so it looks like either a z or h perm
might make cross easier and faster if some of the edges are already oriented to be like that.
next we do f2l but we insert them accordingly with the centers, not accordingly with the edges. After f2l is done it would look something like this:View attachment 13911
the cross looks like an h perm and and all we have left is the top layer. Note: Im only using h perm cases for z perm its the same except for a dif alg for solving the z perm at the end.

For top layer we use a alg that orients the edges and corners, while also matching the edge colors with the f2l, like this.
Now it should look like this:
View attachment 13912
after that you just orient the edges with M and U and we solve the cube.

I would only use this type of method when the cross kinda looks like a h perm or z perm or is close to looking like one.
If it isn't just use normal cfop or any other method.

Example Solve:

Gofile — Cloud Storage Made Simple

Secure, fast and free cloud storage solution. Upload and share files instantly.

gofile.io

You save maybe 1 move on cross for about 8 extra moves at the end, not worth it. Also, use alg.cubing.net for example solves, works a lot better.

I have a new (I think) method for 3x3. Here it is. I call it F2LDiags. It is where you start with roux, and then do two diagonal F2L pairs, then solve the other 2 cross pieces with M and U moves, then solve the other two F2L pairs, then OLL and PLL. Here is a link to a example solve. alg.cubing.net example

What do you mean by "start with Roux"? If you mean something like this, that's worse than Roux and CFOP (and ZZ I guess) for multiple reasons:

1. Movecount is high 50s - low 60s compared to CFOP's high 50s and Roux's high 40s.
2. Lookahead is bad as you're forcing a back slot and you have DB as a blind spot.
3. Mix of movegroups. For example, Roux goes from RULDBF to RrUM pretty quickly, then MU after CMLL, wheras this goes from RULDBF ro RrUMFL to RrUMFL to CFOP F2L. Not really that great.

TL;DR: you're better off using normal Roux or CFOP. There's a reason Roux has barely changed since it was first proposed - it's already so good. Any optimisations involving non linear blocks and mixing with CFOP have all been discounted for that reason. But it's not all bad as you can learn from this.

[This applies to this thread in general atm]
If a method is similar to one that exists, think "why don't they do it like this already?" Chances are, it's already been thought of. If you're still stuck on that and genuinely believe it could be good, come to this thread and we can discuss it properly and go through all the details (like doing a HARCS analysis, genning algs etc.) I know a lot of people are new to speedsolving and want to develop methods and that is great! The only caveat is to think critically instead of splurging out all your ideas onto here. That doesn't mean to stop having lots of ideas. In fact, have many ideas as possible at home but again, think critically about them and understand whether they'd be good or not. You will get better at this the more experience you have.

If you want to gen algs, there is software that exists (cube explorer, kubesolver, alg explorer, HARCS and KSolve) and the software thread is a great place to go if you need any help. If you have any questions about anything else, there will be a thread that exists (just don't bump 10 year old threads) and ask there and people will be more than happy to help. Also, read the first post in here. Anything such as Roux but you solve DFDB was an old idea back when this thread began and something such as CFOP but you don't solve the bottom properly is similar (remember, there's a reason why that isn't done at top speeds). You might think of something that is good but isn't new - don't get dishartened; it happens to all of us. Instead, keep on thinking. Basically, what I'm trying to say is that to come up with a good, original method, you need to put time into it, be it in the form of lots of ideas until one works or handcrafting it over a long time from theory, but the method you thought of in 2 mins was almost certainly thought of by someone else in 2 mins too. If you need help with alg genning, there's a thread for that and if you have a method with proper potential (or a subset), this is the thread for that, not for unfiltered ideas.

Keep methoding.

 

[Nir1213]

You save maybe 1 move on cross for about 8 extra moves at the end, not worth it. Also, use alg.cubing.net for example solves, works a lot better.

I mean i guess its about the same as cfop but just a few moves easier
it can also make you luckier on your cross, plus color neutral which means almost all your crosses will be almost perfect
people can just learn cfop and my method like they are color neutral and it would be pretty easy for them to learn the other method as they are both pretty similar
overall my method is not worth it but if you want to maximize efficiency then go for it
also what should i call my method?
what about Fefop? (Flip edge f2l, oll pll)
also about the 8 extra moves at the end i think you can shorten that and since its pretty simple just some MU moves people executing it would take no time at all,
and saving one move is not that all its going to save like 3 moves and some time as well.
anyway in all its not really worth it
it saves time but really just dont

 

[LNBFilms]

You save maybe 1 move on cross for about 8 extra moves at the end, not worth it. Also, use alg.cubing.net for example solves, works a lot better.


What do you mean by "start with Roux"? If you mean something like this, that's worse than Roux and CFOP (and ZZ I guess) for multiple reasons:

1. Movecount is high 50s - low 60s compared to CFOP's high 50s and Roux's high 40s.
2. Lookahead is bad as you're forcing a back slot and you have DB as a blind spot.
3. Mix of movegroups. For example, Roux goes from RULDBF to RrUM pretty quickly, then MU after CMLL, wheras this goes from RULDBF ro RrUMFL to RrUMFL to CFOP F2L. Not really that great.

TL;DR: you're better off using normal Roux or CFOP. There's a reason Roux has barely changed since it was first proposed - it's already so good. Any optimisations involving non linear blocks and mixing with CFOP have all been discounted for that reason. But it's not all bad as you can learn from this.

[This applies to this thread in general atm]
If a method is similar to one that exists, think "why don't they do it like this already?" Chances are, it's already been thought of. If you're still stuck on that and genuinely believe it could be good, come to this thread and we can discuss it properly and go through all the details (like doing a HARCS analysis, genning algs etc.) I know a lot of people are new to speedsolving and want to develop methods and that is great! The only caveat is to think critically instead of splurging out all your ideas onto here. That doesn't mean to stop having lots of ideas. In fact, have many ideas as possible at home but again, think critically about them and understand whether they'd be good or not. You will get better at this the more experience you have.

If you want to gen algs, there is software that exists (cube explorer, kubesolver, alg explorer, HARCS and KSolve) and the software thread is a great place to go if you need any help. If you have any questions about anything else, there will be a thread that exists (just don't bump 10 year old threads) and ask there and people will be more than happy to help. Also, read the first post in here. Anything such as Roux but you solve DFDB was an old idea back when this thread began and something such as CFOP but you don't solve the bottom properly is similar (remember, there's a reason why that isn't done at top speeds). You might think of something that is good but isn't new - don't get dishartened; it happens to all of us. Instead, keep on thinking. Basically, what I'm trying to say is that to come up with a good, original method, you need to put time into it, be it in the form of lots of ideas until one works or handcrafting it over a long time from theory, but the method you thought of in 2 mins was almost certainly thought of by someone else in 2 mins too. If you need help with alg genning, there's a thread for that and if you have a method with proper potential (or a subset), this is the thread for that, not for unfiltered ideas.

Keep methoding.

Problem with using software is my iMac. Thank you!

 

[shadowslice e]

how does it compare to 3lll?

In my estimation, it's probably about the same speed with fewer algs. On the other hand, it does have the disadvantage of not really having anywhere to progress to so if you want to be really fast at mega (pushing world class and stuff), you'd probably be better off learning and using 3lll.

 

[TheSlykrCubr]

In my estimation, it's probably about the same speed with fewer algs. On the other hand, it does have the disadvantage of not really having anywhere to progress to so if you want to be really fast at mega (pushing world class and stuff), you'd probably be better off learning and using 3lll.

theoretically, how many algs would a 1lll have? I'd guess way too many

 

[shadowslice e]

theoretically, how many algs would a 1lll have? I'd guess way too many

About 1.8 million

 

[TheSlykrCubr]

About 1.8 million

is there any sort of winter variation/vls for last pair on megaminx?

 

[Nir1213]

is there any sort of winter variation/vls for last pair on megaminx?

i mean megaminx is similar to 3x3 so i guess it could have.

 

[PapaSmurf]

It probably does, although VLS would be a lot of algs. I wouldn't be surprised if WV exists and it would be possible to use some sort of "pair up pair while doing EO, insert with WV", although this would most likely be slower than insert+OLL at the very top.

 

[PetraPine]

It probably does, although VLS would be a lot of algs. I wouldn't be surprised if WV exists and it would be possible to use some sort of "pair up pair while doing EO, insert with WV", although this would most likely be slower than insert+OLL at the very top.

recognition for WV on mega would be pretty bad

 

[PapaSmurf]

It definitely wouldn't be amazing, but it shouldn't be any different from OCLL recog on mega. Anyway, it's probably not worth pursuing.

 

[LukasCubes]

I am prepared to get criticized. Well this is a 2x2 method so here ya go.

3/4 of a layer
TTLS (23 algs)
TTLL (corners only) (6 algs)

I have not learned all these algs yet but once I do, ima make a doc about this method. I do not have a name of the method yet so... yeah.

 

[Ashton]

Hi. The H method uses few algorithms. Two examples:


D' F2 U' B2 F2 D2 U' L2 D2 R2 B' L B2 F2 R B' U' F L R2

z2
F2 U L R' F L R' D
U L' R' U' R2 U2 L U2 R
U2 L U L U L2 U L U' L // LS
R U2 L' U' L U2 R' // HLS
U R' F' R U2 R U2 R' F U' R U' R' // COLL
U2 M2 U' M' U2 M U' M2 U' // EPLL

D' R2 U R2 U2 B2 F2 U F2 L2 U' F' D2 L' B' F2 D' B R2 B R2

z2
L' B' U L' D' F
U2 R2 U2 R U2 L R2 U2 L2 U' L'
U R U' R U2 R2 U' R2 U' R' U // LS
U R U2 R' // HLS
R' F' R U2 R U2 R' F U' R U' R' // COLL
U2 M2 U M' U2 M U M2 U // EPLL

 

[PapaSmurf]

I am prepared to get criticized. Well this is a 2x2 method so here ya go.

3/4 of a layer
TTLS (23 algs)
TTLL (corners only) (6 algs)

I have not learned all these algs yet but once I do, ima make a doc about this method. I do not have a name of the method yet so... yeah.

The HD method is this, is pretty good but still doesn't match up to EG because EG is OP.

Hi. The H method uses few algorithms. Two examples:


D' F2 U' B2 F2 D2 U' L2 D2 R2 B' L B2 F2 R B' U' F L R2

z2
F2 U L R' F L R' D
U L' R' U' R2 U2 L U2 R
U2 L U L U L2 U L U' L // LS
R U2 L' U' L U2 R' // HLS
U R' F' R U2 R U2 R' F U' R U' R' // COLL
U2 M2 U' M' U2 M U' M2 U' // EPLL

D' R2 U R2 U2 B2 F2 U F2 L2 U' F' D2 L' B' F2 D' B R2 B R2

z2
L' B' U L' D' F
U2 R2 U2 R U2 L R2 U2 L2 U' L'
U R U' R U2 R2 U' R2 U' R' U // LS
U R U2 R' // HLS
R' F' R U2 R U2 R' F U' R U' R' // COLL
U2 M2 U M' U2 M U M2 U // EPLL

So basically ZZ-blah but purely H? And OCLL/PLL is 100% faster and only has 28 algs.

 

[PetraPine]

I am prepared to get criticized. Well this is a 2x2 method so here ya go.

3/4 of a layer
TTLS (23 algs)
TTLL (corners only) (6 algs)

I have not learned all these algs yet but once I do, ima make a doc about this method. I do not have a name of the method yet so... yeah.

i mean this is just a bad corners first, because of the algs solving 1 layer-1 edge and than doing CLL,top edges than E slice is better,
or just doing waterman which is better than that.
and btw me and probably others have already thought of this

 

[LukasCubes]

The HD method is this, is pretty good but still doesn't match up to EG because EG is OP.

This is like the CLL version of HD. HD wants you to create a V shape that doesnt have to be permuted correctly and mine does; limiting the algorithm count from 59 to just 29.I dont know all 29 yet but when I do, ima make a thread about the method and a google docs on the method.

i mean this is just a bad corners first, because of the algs solving 1 layer-1 edge and than doing CLL,top edges than E slice is better,
or just doing waterman which is better than that.
and btw me and probably others have already thought of this

My point is just to create a good 2x2 method that has NOTHING to do with CF on 3x3.

 

[PetraPine]

My point is just to create a good 2x2 method that has NOTHING to do with CF on 3x3.

oh than that already exist too and me and others have thought of it multiple times after its creation
its ok but just not as good as 1layer cll or eg
if you are going to create a 2x2 method look at the list of methods first...

 

[LukasCubes]

oh than that already exist too and me and others have thought of it multiple times after its creation
its ok but just not as good as 1layer cll or eg
if you are going to create a 2x2 method look at the list of methods first...

i did. HD requires you to make 3/4 of a side and my method requires you to make 3/4 of a layer. This reduces to the total alg count of the method from 59 to 29. I know this isnt as good ass CLL or EG and possibly Ortega just as well as you do. I am just poining out the differences of this and HD.

 

[BenChristman1]

People just need to understand that sometimes (most times, actually), the methods that you make actually suck, or is just a worse version of another method. (I’m just telling how it is; don’t get mad at me.)