> The primary purpose of feature flags is to provide a way to change system behavior dynamically, without needing a deploy.
That's not feature flags; it's just ordinary configuration. (Actually, seems many contributions to this discussion get those mixed up. Maybe even TFA itself.)
Eh, what? No, it is -- or at least used to be -- the other way around. Features are what the programmer puts in the actual application code. "Feature flags" are a (new-ish) way of making the transfer from older to newer versions of the application code more manageable for the programmer: If they're actually "hard-coded" (as, say, boolean feature-on-or-off constants) in the application code, they require a deployment of the actual executable; if they're "hard-coded" in the totally new-fangled sense of the word that seems to be emerging here (i.e. not actually hard-coded at all), it may require deployment of a "features configuration file" or such.
Just plain "configuration", though, is how the user sets up their software to work. That's saved in a local configuration file (or the Windows Registry or wherever) under the user's control, and doesn't require any deployment by the developer at all.
That's not feature flags; it's just ordinary configuration. (Actually, seems many contributions to this discussion get those mixed up. Maybe even TFA itself.)