This dissertation is a collection of three papers. The second and third papers are primarily about inference webs. An inference web is a structured collection of claims and principles in which inferential relationships (including association and common use) dictate the covert meaning of the individual parts of the structure. Here, I introduce the notion of inference webs and argue that, while they are an essential part of our epistemic practices, they can be problematic and, when they are, they obscure our theorizing and wrong those who operate under them. In the third paper I argue that the mental illness literature operates under a problematic inference web whereby legitimate principles are imbued with empiricist assumptions that cause them to go awry. For example, they assume that the social and the scientific are fundamentally distinct. Most philosophers of science will deny this principle if asked directly about it, but I will show that the same principle is operative in many of the inferences that are routinely made in the literature. In the second paper, I argue that inference webs play a role in epistemic injustice. For example, women who have heart attacks are frequently misdiagnosed comparative to their male peers. I argue that this is, in part, due to the fact that someone’s being a woman blocks the inference to ‘this person is having an heart attack’ even when there is both i) a good understanding of what heart attacks are, and ii) good evidence for a heart attack. In these cases, the individual notions are well understood, but inference webs prevent us from correctly inferring one from the other (alternatively, encourage us to incorrectly infer one from the other). The first paper represents a part of my philosophical work that is independent of the above project. I offer a solution to the problem of advanced modalizing which is supposed to show that genuine modal realism is unable to accommodate claims like ‘possibly, there are many possible worlds.’ I argue that the ontology of modal realism rules out those claim as category mistakes.