Is English ‘must' strong or weak? Since the literature contains evidence pointing both ways, resolving this issue requires weighing the costs that each approach must incur in order to account for the opposing approach’s key data. For the strong approach, the literature provides at least a ceiling for those costs, but for the weak approach the costs remain unclear. In this poster, I argue that they are less than they appear. I show that the classic arguments in favor of a strong ‘must’ can be replicated with a variety of weak embedding predicates such as ‘guess’ and ‘suppose’. This means that, regardless of what we say about ‘must', we would still need an account of how weak operators can behave as if they are strong. I propose one such account on which strength is derived via Roothian (1992) implicational bridging.