Today and tomorrow I am attending a workshop on "How To Design Better Software Using FDD", which is essentially about color modeling. Color modeling uses the 4 archetype, Moment Interval (MI) (something happening in time like transaction, business process), Person/Place/Thing (PPT) (something in space like customer, site, product), Role (how are PPT interacting with MI) and Description (catalog /metadata for PPT like list of video titles). Each one of the archetype has a color to make things easier to read UML w/o adding more clutter: MI - pink, PPT - green, Role - yellow and description - blue. Here is like a reference domain neutral model called Archetypal Domain Shape. Your model will almost end up with pink - MI being like the center.
For me color modeling is a nice concept and puts some fun and simplicity in modeling the problem domain. Whether or not its effective and useful is another question, others and the creators have claimed its effectiveness in real world projects... so maybe it does have some benefit. I do think I will give it a try as I see it as something useful to have in communicating ideas of a project. The main lesson was after all is not the colors and model, but ask the business and do not assume. I wish there was some better model similar to a state diagram model which conveys a point in time, UML class diagrams does not convey very well non static stuff. How can you model properly dynamic languages such as Ruby that can redefine classes in run time?