# 4x4x4 FMC (computer and human)

#### rokicki

##### Member
My pleasure! Happy to report some submissions have come in already.
Hope to get many more, and then learn about what people did, and how
they did it.

I am *amazed* at what people can do with FMC on the 4x4x4 looking at

#### irontwig

##### Member
Okay, you really need to have a box for comments about the solution, as FMC solutions rarely are self explanatory. As for the applet, I think it would be nice to be able to label the sides in some ways (like have an "R" somewhere on the R face), since after some rotations it's hard to know which side is which.

#### rokicki

##### Member
Okay, you really need to have a box for comments about the solution, as FMC solutions rarely are self explanatory. As for the applet, I think it would be nice to be able to label the sides in some ways (like have an "R" somewhere on the R face), since after some rotations it's hard to know which side is which.
Excellent suggestions! Don't expect them to happen immediately, however.

Is there a syntax for comments that people use? Perhaps // means that
and everything through to the end of line is a comment (so the comments are
actually interspersed throughout the actual algorithm)?

#### cuBerBruce

##### Member
Is there a syntax for comments that people use? Perhaps // means that
and everything through to the end of line is a comment (so the comments are
actually interspersed throughout the actual algorithm)?
I think there are a number of problems with trying to intersperse comments in the solution itself. For example, if you use inverse scramble, NISS techniques, and insertions, the flow of the actual steps used don't line up with the actual end result. I think it's better to just have a separate text box for writing a free-form explanation.

Another point is that the explanation will generally include additional moves that get cancelled out or combined in the final solution. The skeleton of the solution is generally shown first, with insertion points indicated with special characters, which often are in the middle of some skeleton step. The actual algs used for insertions are generally shown at the end of the explanation.

#### rokicki

##### Member
I still don't have comment support; sorry. But I do want to mention; please, one solution
per submission please. I'm judging these automatically, and newlines in a submission
are allowed. If you have five solutions to the five different scrambles, submit them one
at a time. Thanks!

Happy to see the activity that is taking place in the contest!

#### qq280833822

##### Member
Hi rokicki, would you mind making some results public. Such as the length of the shortest solution?

Btw, how can I know whether my solution is submitted successfully? (I haven't received any confirmation email after my submission.)

#### rokicki

##### Member
Howdy!

I will make some intermediate results public soon, including information about who is
leading, but I probably won't make intermediate lengths-so-far public at this point.

Best way to make sure your submission is successfully received right now is to
submit it through the applet; if the cube is solved in the applet, your submission
should be fine. But I plan to close the loop soon and give you explicit confirmation
on which scramble (if any) is solved and the calculated length per metric.

Stay tuned!

-tom

#### rokicki

##### Member

Computer:
Shuang Chen
Charlie Tsai
Kåre Krig

Human:
Charlie Tsai

I have sent the leader of the Computer division and the Human division a
gift card for \$30 to add to their puzzle collection or whatever else they may
want to use it for.

At this point I will not share any more information on the actual scores.

All submissions I have received have correctly solved a scramble, except for
one, and I have already notified the relevant participant and he has already
corrected his submissions.

#### cuBerBruce

##### Member
Just thought I would post a few of my thoughts regarding the competition.

It appears to me that the applet has a serious memory leak. It seems to cause the browser to eventually use well over a gigabyte of RAM, resulting in problems with the browser freezing for periods of time, and even crashing.

In official competitions, obviously you can't use algorithm sheets or look up algorithms online. This doesn't seem to be explicitly disallowed for human solutions. Of course, if someone pre-generated a huge list of algorithms by computer (such as for all 3x3x3 last layer cases, for example), then used that list in developing a solution, one might argue that is in effect making use of a computer search. For my own human solutions submitted so far, I have not made use of any printed or online algorithm lists - only algorithms I either knew or could figure out without using a computer.

I am guessing Tom is allowing a human solution to be developed after already having developed a computer solution for the same scramble, as long as the person does not use anything from the computer solution in developing the human solution. I think ideally the human solution for a given scramble should be developed prior to working on a computer solution.

#### rokicki

##### Member
Howdy, Bruce!

Thanks for the note on the applet. I'm sorry for any problems you are having with the
applet; I shamelessly lifted and modified the alg.cubing.net applet, and I could possibly
have broken something. The applet does save its state, so if you suspect there is an
issue coming you can always close it and reopen it; I hope that will resolve the memory
issues.

The human portion of the competition is somewhat on the honor system; I have no way
to enforce any sort of constraints. This is why the official cash prizes are only for the
computer solutions (although I may award random prizes at random times for anything
I feel like). I plan to add a textarea to the applet so people can explain their solution,
but even then there's no guarantee they haven't cheated in any number of ways.

On the other hand, people do have names and reputations, and this is pretty much a
friendly crowd, so I'm not too worried about people cheating. If they obviously cheat
I may disqualify them or terminate the human part of the competition (or terminate
both).

I am allowing people to submit any solutions they want in any order; computer or
human, or both. Yes, it's possible someone "picked something up" from a computer
solution and used it in the human solution. Given the few people competing so far,
and the fact that they are all fairly well known in the community, I'm not too worried.

I'm curious to see how well humans can do right now in FMC (it's certainly well
beyond what *I* can do). Already the results are somewhat enlightening to me.

-tom

#### 10461394944000

##### Banned
Thanks for the note on the applet. I'm sorry for any problems you are having with the applet; I shamelessly lifted and modified the alg.cubing.net applet, and I could possibly have broken something. The applet does save its state, so if you suspect there is an issue coming you can always close it and reopen it; I hope that will resolve the memory issues.
i'm pretty sure the memory leak is in the standard alg.cubing.net app too

#### ch_ts

##### Member
I don't want to reveal my solution lengths yet, but I will reveal that my human solutions were done by reduction (which may not necessarily be the best approach). I did reductions without any algorithms, but using ideas developed in the weeks after Tom announced he would hold a contest and before it opened. Then just a regular 3x3x3 FMC. I will also reveal that none of my 3x3x3 human solutions were sub-30 moves, so that is an area where I could definitely improve. I thought there used to be a 4x4x4 fmc in this forum, however it looks like they don't do it anymore and I couldn't find any old 4x4x4 fmc solutions to study.

I personally think that the way people (including me) do reduction is inefficient and could use some work. For example, imagine if people used a set of ~50 algorithms (or comparable to what people know for CFOP) during reduction. In my opinion, doing reduction without algorithms is somewhat like doing 3x3x3 without algorithms (or using a minimal set) - basically a beginner method. I tried to come up with a good method in the past but nothing came out of it, but maybe somebody else can during the contest. (then again, reduction may not be the best approach)

#### rokicki

##### Member

Bruce Norskog
Charlie Tsai

And that's it. I'd love to see more human solutions.

Congratulations, Bruce, for taking over the lead on the
Human competition!

Computer standings remain unchanged.

#### cuBerBruce

##### Member
Congratulations, Bruce, for taking over the lead on the
Human competition!
Obviously, I was pleased to be on top of the leaderboard (human category). I see no mention of the metrics on the leaderboard listings. I am guessing that means that there is no difference in the rankings between OBTM and BTM. I would not be surprised at all that the rankings for the two metrics would be very similar. I mostly have been aiming to get a good OBTM move count, and hope that I'll get some decrease from that in the BTM count. My best OBTM solution so far has no such decrease for BTM, though.

I note that my solutions, like Charlie Tsai's, are based upon reduction. I think some form of reduction will be the most widely used strategy for the human solutions (and maybe computer solutions as well). Of course, you need to work at doing the reduction efficiently as you can. Getting some edges paired during the center solving phase can make a difference in reducing the move count of the edge pairing phase. And, of course, you avoid the parities. I now have the 3x3x3 phase solved in close to 30 moves for all five scrambles, although only one is actually less than 30 in FTM (OBTM). So someone that can readily produce sub-30 solutions for 3x3x3 could certainly overtake me on the leaderboard.

#### rokicki

##### Member
Current rankings:

Computer, OBTM:

Shuang Chen
Stefan
Charlie Tsai
Kåre Krig

Computer, BTM:

Shuang Chen
Charlie Tsai
Stefan
Kåre Krig

Human (same for both metrics):

Bruce Norskog
Charlie Tsai

Roughly five more days to go!

-tom

#### Stefan

##### Member
If I submit solutions A and B for the same scramble and A is better in BTM and B is better in OBTM, do you use A for my BTM score and B for my OBTM score?

#### rokicki

##### Member
If I submit solutions A and B for the same scramble and A is better in BTM and B is better in OBTM, do you use A for my BTM score and B for my OBTM score?
Yes.

#### rokicki

##### Member
It's over! Results in a few hours. Thank you all for participating.

#### rokicki

##### Member
Tom's Fewest Moves Challenge: The Results

After two months, the results are in! First, the human results.

There were only two competitors in the human division, and here
are their results:

Code:
                   OBTM                Avg.  BTM                 Avg.
------------------------  ------------------------
1. Bruce Norskog   75  75  76x 71  67x 73.6  72  72x 69  67  67x 69.3
2. Charlie Tsai    79  87x 72x 79  81  79.6  79  86x 71x 76  79  78.0
Bruce wins first place in both metrics, and Charlie wins second
place. Congratulations to both competitors; these are amazingly
short solutions!

There were five competitors in the computer division. These are
their results. First, the Outer Block Turn Metric:

Code:
                   OBTM                    Avg.
----------------------------
1. Stefan Pochmann 41   41   42x  40x  41  41.0
2. Shuang Chen     42x  41x  42   41   41  41.3
3. Charlie Tsai    50   48x  49   52   53x 50.3
4. Bruce Norskog   47   53   56x  43x  52  50.6
5. Kare Krig       92x  95  109x 102   99  98.6
Next, the Block Turn Metric:

Code:
                   BTM                     Avg.
----------------------------
1. Stefan Pochmann 38x  37x  38   37   37  37.3
2. Shuang Chen     38x  39x  39   38   38  38.3
3. Bruce Norskog   46   44   49x  41x  46  45.3
4. Charlie Tsai    47x  48   47   49   50x 48.0
5. Kåre Krig       69   67x  73x  70   69  69.3
In both metrics, Stefan Pochmann wins first place, and Shuang
Chen wins second. It was a very tight competition; congratulations
to both winners!

All winners will be hearing from me by email within the next
24 hours. I'd love to see how people approached this problem