You are probably right if the data you're representing is scaled, but this sort of solution is fantastic for categorical data. Also, see other comments re: accessibility for colorblindness. Not necessarily a silver bullet, but possibly a solid solution in some circumstances.