In this final chapter, I explore Anjan Chakravartty’s semirealism, or selective scepticism, as a peculiar version of structural realism. To contextualise his proposal, I explain why Chakravartty rejects both entity realism and two other versions of structural realism—namely, epistemic structural realism and ontic structural realism. While I praise his attempt to mediate the debate between realism and anti-realism by providing a middle ground that offers neither the optimistic nor the pessimistic meta-induction full support, I find his proposal lacking. He endeavours to find a principled criterion to distinguish parts of a theory that will survive theory change because they will be repeated in future theories from other parts that will not. Only the first, and not the latter, will be good candidates to extract realist commitments. I criticise this approach because it does not consider several important complexities involved in theory change that make it impossible to directly derive realist commitments from the parts that better, new theories repeat from previous ones. I argue that my proposal to explain theory change and the notion of increasing degrees of epistemic verisimilitude avoids the problems that Chakravartty has to face.