Yes, I agree some things are more underhanded than others. But it interests me that Mathematica has something like that in one operation; It may interest others.
On the other hand, here, I've demonstrated that no less than two languages have the same single-operation vector compression. So this is not a one-of-a-kind underhanded thing; It's actually in production use and should be considered a legitimate operation.
Besides, in x86 that took 5 bytes to express; APL took one character. So in retrospect the x86 code wasn't so effective anyways.