[MrRubix]

It wasn't too bad if you used "reasonable Occam's-Razor statistical logic". By this I mean "most likely Dev is not going to be super evil and create a bunch of red herrings -- if I find a cool pattern, it's likely relevant to the final answer in some form". The only "red herring" in this case was the fact that the first three sum to numbers ending in 0, but this was mainly because I misunderstood the problem at first. Once I noticed that the inner numbers were the ANSWERS, that means the resulting numbers were numbers ending in 0 minus the answer, which is peculiar and led me to this reasoning:

e.g. "There's no way that the fact that the first three are divisible by 9 is a coincidence. Therefore there must be some rule that actually applies to all four of these boxes and just-so-happens to produce a /9 for the first three, but something else for the fourth one. Since the /9 must come from these numbers, I should figure out how to work the numbers in the first three boxes to somehow yield a /9, and then see if this rule works on the fourth box"

After thinking of it in this manner I got the answer pretty quickly