For the last spot in the codon to bind to an mRNA the first two HAVE to be bound already. But for the bind to be good enough for the ribosome to actually add the amino acid to the polypeptide you don't always need that last one. Since the first two HAVE to be bound, they cannot vary. But the last one is not vital for the ribosome to decide it has the right tRNA attached. So that third site can vary, and the ribosome can still attach the amino acid to the polypeptide.
So if the codon table wasn't redundant what would happen?
instead of CTT, CTG, CTC, and CTA, all coding for leucine they may all code for different amino acids. If this was the case when the ribosome messes up and lets a CTG tRNA attach to a CTT codon the protein is going to be WRONG. So evolution favored organisms where the tRNA for CTT and CTG coded for the same amino acid, so that if the ribosome messes up it does not affect the protein.
Think about the hoodie analogy. The left side of your hoodie is the tRNA the right is the codon. If you zip your hoodie up just a little bit the zipper will slowly slip down until your hoodie flies open. If you zip it up 2/3 of the way though it isn't going to fly open, it is zipped enough to stay stuck together. If the hoodie is zipped enough to stay together then the ribosome will take the tRNA (left side of your hoodie) and attach it to the growing protein. It would be even better if your hoodie was zipped ALL the way, but it is not necessary.
