The Music Genome Project tracks songs into broad genomes with specific genes. The project aims to match lesser known music to listeners' existing tastes. Its obvious utility lies in signaling to artists how they can construct music that a clearly defined market demographic will buy. Consider that some white people like hip-hop and some African-Americans like country/Western. Race and gender demographics matter less in an era of Big Data when algorithms can match individual consumers with music they like.
Let's consider further that modern algorithms have something to learn from ancient philosophers. The golden ratio has guided architects and artists for centuries. Musica universalis was an imaginary form of music occurring naturally in the universe. I believe algorithms plotting these two concepts into something like the Music Genome Project could produce marketable music. Classical music has always followed precise modes, like 4/4 time for an orchestral work or the standard twelve notes in Western music notation. Applying math to ancient theories gives us more musical options.
Other musical theories like Sonido 13 are ripe for experimentation with algorithms. Music innovation should no longer be confined to obscure composers or theorists working in isolation. Computer science can theoretically generate new music automatically through an algorithm's continual processing. Generating enough new songs and compositions means some of them will find an audience. Pythagorean musical algorithms can apply the golden ratio to the music of the spheres, and to all other kinds of creations. Someone will buy it among the market for space and ambient music. Ancient concepts can be new again.