Because software quality begins at the source.

This point is the most obvious one. All computer software deals with data. If there is any chance that data might be corrupted or manipulated uncontrolled by the software, the software is useless and should not be used in productive environments. One important principle here is the DRY principle which stands for "Don't Repeat Yourself". Redundant data is very hard to keep coherant. So we should avoid it under any circumstance.