During my whole cubing career I've been most interested in advances we can make in every event, be it methods, algsets, concepts, techniques and so on. after talking about the unknown future of sq1 advances after oblp(w/ multiple obl algs) I am so excited to see something like this, it really bring so much joy to my heart. Thank you Antonie for sharing this
This is really cool! Though don’t top cubers generally not agree on optimal algs for even EG? All learning sources provide multiple alg options. I don’t see how you can make this objective. Really interested in this hypothesis that everything can be solved sub 1 without rotations though!
I have to be honest, only knowing full EG for 2x2 and my TPS having a possible 10+ tps makes me completely able to be sub 1 on 2x2 with enough practice. With a majority of sub 2.5 2x2 solvers learning every single case. I think averaging sub 1 will be possible for all cubers with decent enough hand abilities. Likewise, I think taking AUF into account may be the best thing for 2x2, as it could reduce the number of cases to a manageable amount. 154,608 of anything is ridiculously hard. Most English speakers know how to use around 10,000 to 20,000 words, Chinese speakers can use around 3,000 Chinese characters in the Chinese language, humans will meet 80,000 people in their lives, what is the limit?
Hey! This is really cool. I have a feeling I could help out as I have a kinda cool way to gen algs such that they are regripless to begin with. I've posted some of 3x3 algs on CF found using that which you might have seen (e.g. Gc-perm: r' F R F' U2 M' U' R U r' U2 r U R', not that it's objectively better but it's pretty good and unique at least). I'll not bore you with the details, but the gist of it is that I can make a move-transition table which says which moves are allowed to be applied at which point. So for example, if you start in homegrip, then one of the moves you can do is R. However, then you reach a new grip, and doing another R would be awkward so that is not part of the moves you can apply (alternatively another R is fine, but doing R when you're an R2 away from homegrip would not work for sure). This does not only apply to grips, I can for example define the table in such a way that max 2 F moves are allowed in an alg, or such that if an F is done, then the next F-move has to be F' - the possibilities are endless. I have 3 questions: 1) Do you have access to a list of all the scrambles up to symmetries, and can I find that somewhere? I could generate them all myself but would be nice to be spared the trouble ;) 2) Could you formulate some more rules? What I mean is that things such as no rotations isn't objectively a criteria, it's concievable that some y solution could be faster than an solution - I highly doubt it of course, and my point isn't to try out solutions with rotations, but rather to figure out other kinds of rules that could make the search easier. 3) Do you have any definite criteria that would instantly mean a solution is oo? If so, I imagine we could progress fairly quickly in the beginning by autoaccepting some certain types of solutions. I am not a pro 2x2er in any way, but I could imagine regripless move optimal QTM solutions would be hard to beat. Or maybe not autoaccepting them, but I could definitely flag solutions by such criteria such that you could be even more confident when testing and accepting them. I'm open to discussing this somewhere better than here if you're interested, I'm sure for example that I could make a really good move-transition table with some input from you. Thanks for all the work so far!
Are we considering less than 4 move solutions in the scrambles because eliminate they would reduce the total a bit there's not much states where you can solve the cube in 3 moves but even eliminating a 100 would be good
not to be a hater but there is simply a 0% chance anyone will ever know all of these. its literally 10x 1LLL and only one person knows that set. i think its a useful resource though, and a good way to expand the cases you're knowledgeable about
In fairness, while it might not be feasible to learn full OO 2x2, it's also not exactly fair to compare it directly to 1LLL or 5style because those require recalling algs during the solve, while OO gives you 15 seconds to recognize/recall cases. The algs for OO are also all good (in theory), while that isn't the case for 1LLL and 5style
The first problem I see is recognition. There are methods that some people in the community have created that require so much recognition (at individual sticker patterns and symmetries) that it would be impossible to learn bigger alg sets with thousands of algs. 2x2 TCLL is kind of where I would myself end. I wonder how the method will advance in the future.
The number of positions without taking symmetrics into account has actually been known for decades. Anyone with some programming skills should be able to find it without too much trouble.
