Complexity does not ensure the musical value of a piece of composition. A piece can be totally creative and random but does not hold much significance in a musical sense. The very nature of music makes live coding music more difficult than creating visuals. That is to say, it requires more deliberations and more experimentations with sound from us,. Not only do we need to make 10 times more the effort with which we experiment with images, but we also need to know in what directions are we making the efforts. In other words, we need to know how to improve the musical value of a piece of composition. 

Namely, a good musical piece always has development, evolution, and form. Music, since it carries more emotions, takes on a more important role in conveying them. For example, it is easier to make people feel happy by feeding them an upbeat melody than by presenting them with visual patterns. Admittedly, we can also use certain types of images and visuals to signal emotions, such as slow waves for tranquility, the dark color theme for horror, or messy lines for anxiety. But in my opinion, visuals are not as powerful as sounds. Therefore, the design of sounds comes in the front. 

Looping the music helps increase the musicality, but only to a certain extent. As time goes by, the entropy decreases and it is not that attractive to the audience anymore. This is where we want to notch it up a little by introducing noise, the random factor. At first, it seemed a little bit counterintuitive to me because good music needs to retain a certain form. But noise brings information. It all depends on how well you are using the noise. A successful introduction of noise that takes noise vulnerability factors, the nature of noise, etc. helps us to achieve a balance between entropy(the information) and musicality.