It's possible that any test of intelligence that the educational system isn't measuring and working on is going to be a better predictor of various kinds of success than those it is. There are a huge number of factors in success, and it's possible that they are correlated more with each other than they are with the managing to get good scores on those few traits that are tested and taught for. (I don't know whether the original papers address this possibility; I'm just going by the linked article.)
Which isn't to say that it's bad to take these things at face value and start testing things like this as they're found to be relevant - it seems better to test and teach more things than fewer.
On a tangent, I seem to remember a study found that Quake improves spatial skills more than Tetris.