Hacker News Reader

Comment by callesgg | Hacker News Reader

I assume the time depends very much on the initial state.

Speedcuber here: each competitor starts with the same initial state for each solve in their best of X. Yes, there is some luck involved but it's distributed equally.

Per competition though, not across competitions, which is relevant WRT world records.

Yes - the fact that the record was broken twice in this one competition suggests that its particular initial state should be analyzed for solvability before we assume that we have entered 'a new chapter for speedcubing.'

The two records were actually on different scrambles (i.e. different initial states).

It's one thing if there was a particular scramble that everyone was solving faster than usual, but that wasn't the case here.

Well, yeah. You can't account for everything. Traditional sports don't consider elevation, air quality, wind, etc.

To an extent. Road race world records are only valid if they fall within certain limits of elevation change. You can run slightly downhill, but can't descend 500 meters. Ditto for tailwinds, I believe.

Wind speed is accounted for in athletics. Records won't get validated if there is a strong tail wind.

This seems to be the only somewhat common variable.

They do consider wind. Too much tail wind can invalidate a world record.

https://en.wikipedia.org/wiki/Wind_assistance

That's one factor in one sport.

Does anyone disregard a track world record because the weather was pleasant that day, and the previous recordholder did it on a hot, humid one? Does anyone argue Barry Bonds' home run title should be disregarded because of the thin air at his away games in Denver?

Take away wind assistance, and Bucky Dent loses his place in history. Take away a pebble in the grass and the Red Sox win the 1986 World Series. Add some rain to give pitchers an extra day of rest, and all sorts of baseball history gets upended.

Is there the scramble/analysis of the 4.90s solve somewhere? Looks to me like a nice fast cross, standard F2L, standard OLL, and then a clean PLL skip?

edit: found something on page 16 of a thread: https://www.speedsolving.com/forum/showthread.php?55952-WR-L...

looks like the last move was an OLL(CP) alg (https://www.speedsolving.com/wiki/index.php/OLLCP), not just standard OLL, making the last layer skip 6x (?) more likely at the cost of remembering a few hundred more algs.

Sort of, technically, no state needs more than 20 moves to be solved (but finding that optimal solution is kinda hard for humans). So while luck can play a part (i.e.: you may get a configuration that only requires three moves), it's much more a matter of skill to find those 15, 20, 23 moves.

http://www.cube20.org/ https://www.youtube.com/watch?v=yF2J39Xny4Q

The strategies humans use to solve it are not nearly so optimal in terms of number of moves. More like 50-100 for a typical solve.

There are standard randomized scrambles generated for competitions. Most competitions use some subset of 12 solves and take the average, although this appears to be the single solve record.

I wonder how they set the initial state. I think it's fairly easy to calculate the minimum number of moves required to solve from a given state, so you'd think they would control for this.

Most states can be solved in around 21 moves by the best computer algorithms, but the fastest humans usually solve it in significantly more.

The initial state is a random computer generated scramble. It's easy to calculate if you have a computer but it isn't easy when you have to do it in your head.

It is, in fact, not easy at all to calculate that.

https://en.wikipedia.org/wiki/God%27s_algorithm

(There isn't as much info as there should be on that Wikipedia page but it was an open area of research for decades.)

Isn't it though? I thought the hard part was proving the minimum number of moves that can solve all possible initial states. Korf's algorithm and several subsequent improvements seem to be able to find the minimum number of moves to solve any given initial state.

A yes. You are correct.

Finding minimum # of moves from a single state = not so hard.

Finding the worst possible state (the state where the minimum # of moves is a maximum) = pretty hard, though now solved.

Agreed. More interesting is the variant where people try to solve a cube in as few moves as possible.

Reply on news.ycombinator.com