I think he's referring to how some competitions designate specific scramblers for each event, instead of calling up volunteers from the crowd at random. It depends on the organizer or if it's even possible to have enough known people to make said schedule.
It's only 7!*3^6 = 3,674,160 positions.
As far as I know, it can't solve states optimally that quickly. The robots I've seen just take an early sub-optimal solution instead of running it until the two phases converge on an optimal one (which can take minutes depending on specs).
Why...
1) 66
2) 59
3) 67
4) 65
5) 69
Notes: http://pastebin.com/mN4ixgYR
I solved the last 2 edges before doing the cross, then subtracted the length of the L2E alg from the total move count. E moves were counted as 2 moves, and flipUF/flipFR is 7. The +x at the end is how many moves it took me for...
It didn't look fully assembled in the video? Unless he took a chunk of it out and inserted those cards within 5 minutes, then yeah.
Edit: It even says in the description of his video:
Did you mean to quote biscuit? :P
He never finished assembling it; an unassembled cube is not a fully functional cube. That picture only shows one view of it - it's empty on the other side (as seen in the video of it exploding).
This little thing, although slow, costed less than a Speedstacks display:
That said, having used it at a competition, it definitely couldn't replace human scramblers. It heavily traded off speed for reliability and simplicity.
I heard that the faster, multi-armed ones (that would be...
I work in a brain research lab that does a lot of learning curve-related experiments, and from the regression models I've used, the steady-state value is always really wonky; the lab cares more about the learning/exponential decay rate.
Here's a fit to the 3x3 WR averages over time using one of...
You could try:
1) Solving a cross with two opposite edges swapped
2) F2L
3) CLL
4) Opposite-cross ELL
It's like Roux with less-efficient block building, but only 30 possible L6E cases.