What can we learn from noncoding regions of similarity between genomes?

What can we learn from noncoding regions of similarity between genomes?