Alternatives (roughly: things a speaker might have said) are useful in semantics; they underlie popular, explanatory theories of questions, focus, implicature, and free choice.
Alternative semantics, which conceives of compositional interpretation as fundamentally alternative-propagating, is also useful, but for distinct reasons. Saliently, alternative semantics is one way to compositionally derive alternatives! But there are other routes to alternatives: Karttunen's (1977) theory of questions and Krifka's (2006) structured-meaning account of focus both use scope-based mechanisms to arrive at sets of alternative meanings, and both theories can be repurposed into fully general accounts of alternative propagation, without any need for an alternative semantics. Recently, these techniques have been re-discovered by Ciardelli et al. (2016).
Yet alternative semantics is claimed to offer various advantages over the scopal approach, principally an immediate account of island-insensitivity (Rooth 1985; Kratzer & Shimoyama 2002). But alternative semantics is also known to suffer from various problems, chief among them a problematic relationship with binding (e.g., Shan 2004). Finally, both scopal and alternative-semantic theories suffer from a seemingly baked-in commitment to unselectivity outside of scope islands (Wold 1996; Rooth 1996; Krifka 2006).
I'll show that we can, indeed, have it all: island-insensitivity, a robust account of binding, and full selectivity outside islands, with a conservative syntax-semantics interface and a scopal approach to alternative propagation that, nevertheless, provides alternative semantics on demand.