Mark Brader mentioned that such cubes had been mentioned in rec.puzzles recently; I found a mention with a Google search:
http://groups-beta.google.com/group/rec.puzzles/msg/
c5b75360a6cb1a54 (BTW, Google no longer provides friendly support for viewing Usenet messages; anyone know of an alternative for Internet cafe users?)
That thread suggests more puzzles:
Puzzle 1: Find eight points on the Earth's *land* surface that form a perfect cube. (I use the approximation that the Earth is a perfect sphere.)
Comment: My solution with Cocos Island comes *much* closer than any other near-solution I've found, but my database of obscure mid-ocean islands is far from complete. I don't think even my solution works perfectly: the antipodes of Cocos Islands may all be some miles to sea off the coast of Nicaragua.
Puzzle 2: The old thread asks: Find eight points on the surface of a sphere maximizing the minimum distance among point pairs.
Comment: As pointed out in the old thread, a cube is *not* optimal.
Puzzle 3: Same as Puzzle 2, but replace 'eight' with 4, 5, 6, 7, 9, etc.
Comment: This must be an anciently studied puzzle. Given my Alzheimer's symptoms, I may well have read about it here 15 years ago and forgotten.
Puzzle 4: If my calculation is correct, the best puzzle-3 solution for 5 yields a minimal distance no larger than the 6-solution so call 5 a 'useless' number. Find all useless numbers.
Comment: There should be only a finite number of them, right? James Dow Allen