The article discusses different types of noises in scientific measurements, focusing on heteroscedastic noises and their models. It explores how these noises can vary in chemical measurement systems and the challenges in choosing the right noise model. The researchers show that both linear and "hockey stick" models have solutions, despite common misconceptions. They also demonstrate that the δcritical method doesn't work with heteroscedastic noises and highlight issues with the MARLAP example. Overall, the study emphasizes the importance of accurately modeling noises in scientific measurements.