It may very well be working in more than one mode - but I strongly suspect its a far better shield, than it is a forward power reflector. It also may strongly effect the VSWR of the existing dipoles used here.
But to be fair, it's a wild guess. It's a semi-flat piece of foil placed at an angle, at a random distance from 3 individual antennas. That's why I linked to the nice and simple, easy-to-calculate yagi-uda for if you want to hack with antenna gain.
No disagreement that the yagi thing you pointed out would work better.
That said, this is a MIMO router, so that may change the modulus on the multipath problem.
A regular home Wi-Fi network is already a complex multi-path system, as the primary propagation mode is reflections from walls and ceilings, leading to plenty of self-interference with many hot- and cold-spots from constructive and destructive inteference. We are at most introducing a few stronger reflections in an already extremely reflection-heavy environment. The big question is whether it does any good, either through attenuation or gain.
As for MIMO, I must admit to not knowing much about it—I know it utilizes spacial separation to increase bandwidth, but I do not know much of its actual implementation, and thereby weaknesses.
It won't be a satellite dish as the comment thread on the page mentions, though, which is both a passive reflector and a focusing device, whose curvature requires at least some level of precision to be of any help.
For something much more helpful, see http://ham-radio.com/k6sti/wifiyagi.htm