What concept analysis in philosophy of science should be (and why competing philosophical analyses of gene concepts cannot be tested by polling scientists).

What should philosophers of science accomplish when they analyze scientific concepts and interpret scientific knowledge? What is concept analysis if it is not a description of the way scientists actually think? I investigate these questions by using Hans Reichenbach's account of the descriptive, critical, and advisory tasks of philosophy of science to examine Karola Stotz and Paul Griffiths' idea that poll-based methodologies can test philosophical analyses of scientific concepts. Using Reichenbach's account as a point of departure, I argue that philosophy of science should identify and clarify epistemic virtues and describe scientific knowledge in relation to these virtues. The role of concept analysis is to articulate scientific concepts in ways that help reveal epistemic virtues and limitations of particular sciences. This means an analysis of the gene concept(s) should help clarify the explanatory power and limitations of gene-based explanations, and should help account for the investigative utility and biases of gene-centered sciences. I argue that a philosophical analysis of gene concept(s) that helps achieve these critical aims should not be rejected on the basis of poll-based studies even if such studies could show that professional biologists don't actually use gene terminology in precise ways corresponding to the philosophical analysis.