question about new Rubik's cube proof (all solutions require <= 20 moves)

[Arnold_Winkelried]

Update:

I e-mailed the guys at the webpage. They were very friendly and told me the following:
The “superflip” they show has a 22-move process because they are using a java app to display the way to get to a certain position, and the java app calculated a 22-move solution. The page clearly states that it should be possible in 20 moves.

It is standard in the cubing community to count a 180° turn is one move instead of two. If you count 180° turns as two moves (which I thought you should do), then the “God’s number” (minimum number of moves to solve a worst-case scenario) would be somewhere between 26 and 29 moves.
From the e-mail, discussing the case where 180° turns count as two moves:

Interestingly, only one position is known to require 26 moves–superflip composed with “Pons Asinorum” (perhaps the most well-known symmetric pattern other than the solved state).

[Chronos]

I presume that what they’re calling “Pons Asinorum” is the checkerboard? L2R2F2B2U2D2?

[Arnold_Winkelried]

I would assume so. That’s how it’s defined here
http://www.math.ucf.edu/~reid/Rubik/patterns.html

[Lumpy]

I’m surprised that there are multiple positions at the maximum number of moves. I had intuitively believed that there would be a single most difficult possible position, an “anti-solved” state.