Adaptive learning
Many are asking "didn't they test the software before sending it?". Since NASA is not Boeing, I'm going to assume they did, but it also seems reasonable to assume that conditions on the ground 15 light minutes away may not be replicable with 100% exactitude. If the slight hiccup was caused by a local variable that could not have been replicated Earthside, then rewriting the software to factor in the new data would seem to me to be evidence of sensible engineering, not the opposite