If you don't trust an identified host to execute your transaction how does blockchain magically make that better?
Tanktalus provided one example upthread, what you might call "trusted competition" (or "competitive trust"). More generally, "trust" in the real world has many complexities, many shades of gray, and they are are not static. You might trust Ben Carson (or at least a younger one) to perform complex neurosurgery on your infant, but you might not trust him to run HUD. (Ben Carson didn't trust Ben Carson to run HUD. Oops.) As another analogy, during the Cold War the United States and the Soviet Union trusted each other in certain ways, not in others. In particular, they trusted each other to blow each other up if sufficiently provoked ("Mutual Assured Destruction").
You could easily make the same basic argument about Bitcoin, and I might even join you in making that argument -- that the world is full of currencies (including many better currencies in terms of what currencies are supposed to do), and there are many other ways to create and to operate a currency than to use Blockchain algorithms (and to consume the equivalent of Holland's entire electricity demand in the process, last I checked). I remember the early commercial Internet when lavishly funded startups like Pets.com could advertise on the Superbowl but made absolutely no business sense. Blockchain is going to have its share of dubious applications and hucksters. Bridge well crossed, actually. Blockchain's first use case (Bitcoin) might be extra dubious. That said, I'm not brave enough to predict that Blockchain algorithms have no reasonable use cases that are "best fits" for the technology, especially if parties who understand business and government at least fairly well are working together. No matter. We'll find out soon enough, probably within the next year or so.