Most journals and book publishers follow a consistent house style, which authors have to implement. This leaves authors very little discretion over design choices (probably for the better).
When it comes to colored figures, which are usually allowed, each additional color costs money. With xerox printing, this is now less than ten years ago, when for each primary color, the page had to run through an additional stage of the offset printer, but it still adds considerable costs. Of course, this only applies if the text is printed at all.
Some textbooks use colors and other graphic design elements. If well done, this can help to convey information rather than distract from it. But in order for it to be well done, it needs to be taken care of by a professional designer.
With respect to individual manuscripts, you can of course do whatever you want. But unless you have a talent for design, it's easier and more efficient to use a standard template and to follow "less is more" as a rule of thumb. It is also safer, because aesthetic judgement is quite subjective, and because more people overestimate their design talent than underestimate it.
Moreover, academics are a conservative bunch when it comes to scientific conventions; and they tend to suspect that "eccentric" form only serves to cover up weak substance.
Finally, one remark about color blindness. Colorblind people can see colored objects quite well, they just can't tell the difference between certain colors (mostly red vs green). Therefore, colored text doesn't hurt them (unless you use red type against green background), nor does it help them to print a colored document in black and white.