I'm not sure this sort of logic scales. The examples he uses with a single boolean are simple enough because you're just writing two methods instead of one. What if, however, you have a signature that takes multiple booleans, either as formal parameters or as part of a struct? Do you write exponentially more methods, requiring the caller to be familiar with your naming conventions, just to free the caller from having to remember what the parameters represent? For that matter, what if you're passing in an enumeration? That's also a "flag" that is frequently used to switch behavior, and again, would require expanding one method into n, each with human-readable names.<p>If there's some a way to win big by doing this, I just don't see it.