I think there are a number of points to be made here:
(1) I agree that TBT doesn't have to be "perfectly on target" (whatever that means) for a given learner in order for the learners to learn from it (remember, to quote my favorite IDer: "people learn in spite of instruction..."). Presumably, the closer to "on target" the TBT is, the more/better/faster learning takes place. The same is true for any kind of instruction, regardless of the strategy and the medium. I even think Roger Schank would agree.
(2) That said, I have often wondered, when doing evaluations of TBT, why a given hunk of (badly designed or inappropriate for the learner) instruction in a textbook will be incomprehensible to a learner, but the learner will merely conclude "I'm dumb" or at best, "that's not written for me--I'll go find something else." The same instruction, implemented in TBT (especially a style which uses a lot of meaningful interaction and feedback), will actively annoy the learner and cause defenestration of the computer, desexing of the mouse, etc.!
(3) Why? I think the reason is that the very act of interaction creates the expectation that the thing with which one interacts is supposed to have at least a three-year-old's ability to understand a conversation and its context. The more meaningful the interaction, the more stringent the expectation is. Which is why, I think, that badly-designed interactions attempting to teach something non-trivial really annoy the learner and stick out like a sore thumb in the evaluation. (If you don't believe me, take a look at the client feedback through our hotline, formative and summative evaluations--very humbling.) Majel Barrett, where are you when I need you?
(4) It's particularly bad in instructional situations, when compared to entertainment-style games. Games often use a fantasy world as the context. Among other things, that lets the designer off the hook when it comes to making unrealistic simplifying assumptions about the context: transfer is not an objective (unless you really want Mech Warrior II), so both the designer and the user understand that things aren't supposed to make sense, or correspond to the user's world. That said, there are two limitations, even in fantasy games: (a) if the context gets too unrealistic (referenced to "our" world), the game becomes unplayable, and (b) the whole game industry has never really broken out of the 11-to-18 male demographic. (Maybe the world doesn't make sense to teenage boys anyway, so they don't even realize they're disoriented.)
Alas, for us poor IDers, the rules are different: the context is supposed to encourage transfer to just exactly the right real-world contexts, so neither we ourselves, nor the purchaser, nor the learner will cut us any slack if the context starts getting unrealistic. That also means the interactions have to have high cognitive fidelity, too. Sigh. It would be much easier to design my next math problem solving simulation if I could just suspend the law of gravity and count number of photon torpedoes used instead of having to figure out how to measure some real problem-solving skill.
(5) It is all the above factors which, in my humble opinion, cause the multimedia game designers who lack ID expertise to design trivial TBT or to absolutely go down in flames when they try to do something non-trivial. I would also argue that, among other things, it was failure to understand this dynamic which led all those venture capitalists to think that interactive TV was going to work as a mass medium. Too bad nobody told the customers in the market tests.
(6) It's also one of the factors (not the only one) which has led textbook publishers to kill TBT companies they buy, consistently over the past 25 years.
To conclude, life ain't fair. We really do play by a different set of rules than the textbook publishers, or the lecturers, or the entertainers, or the game designers, or the film makers. Effective TBT really is something more than another branch of show biz (though there's lots of show biz in it).