(Cross-posted from my facebook)
The term "scientist", in modern times at least, refers to a practitioner of science- someone who generates data in accordance with the scientific method (defined in a standard, ecumenical way). Scientists may also make inferences based on this data. This involves, among other things, noting correlations among sets of data, making generalizations based on a limited dataset, choosing a hypothesis among many based on simplicity and other criteria, and so on. All of this amount to "practising" science, the same way you "practise" woodworking. Science is, at the end of the day, a craft; and scientists are craftsmen. Craftsmanship is measured in terms of one's knowledge of what constitutes good craftsmanship and best embodying it.
All of this goes on within, and assuming, the system of science. In "doing" science, scientists have to assume laws of mathematics, logic, induction, inference to the best explanation where "goodness" of explanation is decided on the basis of explanatory power, scope, simplicity, ontological economy, predictability etc, perhaps methodological naturalism, and so on. The crystal palace of scientific knowledge can only be constructed when one uses these rules and tools obediently.
But can an average scientist be expected to deliver verdict on the system *as a whole*? For example, if someone were to ask- how do you know that obedience to the laws of induction would generate knowledge that corresponds to reality as it is? A scientist who works on a specific part of the crystal palace, but hasn't seen (much less examined) the entire thing from a bird's eye point of view, cannot be expected to have expertise on the topic. In fact, that's a question about whether science works at all. A less esoteric (from the perspective of the empiricist) question would be- what laws and assumptions do science have to make in order to work? I think many scientists wouldn't be able to give a rough-and-ready answer to this either.
What makes things more complicated is, due to the explosive advancement of scientific knowledge in especially the past few centuries, individual scientific disciplines have become more and more self-contained, each with its own set of rules and tools and methods. Like a Russian Matryoshka doll, individual scientific disciplines, in terms of their scope, can be said to nestle snugly within broader disciplines. Each smaller doll requires the efficacy of the one it snuggles into in order to work. Microbiology, for example, works on the assumptions of biochemistry, and has higher-order assumptions and methods of its own. Biochemistry in turn assumes laws of chemistry; chemistry, those of physics. The inhabitants of these smaller chambers cannot be expected to have knowledge on the bigger rooms in the palace (just as those who inhabit the bigger rooms don't know what goes on in the smaller ones). So in many cases, the scientist cannot even explain, or be expected to have expert opinion about, the more fundamental *scientific* principles on which his own discipline works, simply because they inhabit a higher scientific plane he may not be entirely privy to.
Answering these broader questions is the duty not of the scientist, but of the philosopher. A philosopher doesn't work with science, he thinks about science. He examines the data produced by the craftsmen, analyzes them, and makes broad method-related conclusions about them. This gives him the privilege of not being enclosed within the palace, much less being cooped up in individual rooms. From this vantage point, he can see the bigger picture, and make conclusions about the entire system without having to assume any of its rules. So in response to the earlier question about whether scientific knowledge corresponds to reality, the philosopher would wade through the muddy waters of scientific realism vs. antirealism, examine arguments both for and against, and be in a position to give a verdict. In response to whether induction is a good guide to knowledge, he would pore over the literature on the problem of induction, develop models on which induction can be said to work, and give his expert opinion.
Even for questions which are about particular disciplines of science, as opposed to being about science itself, there are philosophers specializing on specific rooms. While a philosopher of microbiology, for example, wouldn't be completely outside the palace, he would be standing at the door of the microbiology room. This would present him with a unique vantage point to observe the room's structure, as well as details, from inside and out.
In the crystal palace of scientific knowledge, scientists build, philosophers work maintenance. If you want to know how to build, how to generate scientific knowledge, talk to scientists. If, however, you want to know whether that knowledge corresponds to reality, or the dimensions, scope and limitations of that knowledge, talk to philosophers.