Well, thermometers are not as bad. Assuming you invest $100 in a certified/calibrated thermometer you can use it as a reference source. With light meters, there are many variables and many different ways in which they can give different readings. Everything from k-factor and other calibration issues (spectrum), to linearity, to measurement field, to optics (flare). Meters are a horribly complicated mess of variables. Fortunately our films etc. provide plenty of latitude, and "errors" may tend to offset eachother.