When certain layouts are not possible (which is why certain things must be either true or false), broadly speaking, there are two possibilities – either something will have to be false because it explicitly violates a single rule, or possibly an obvious inference from a couple of rules combined; or, alternatively, an invalid layout will be invalid because of the interaction of multiple rules in a way that’s not readily apparent.
Often, you can tell which is more likely to be the case on a particular game – if there is a bunch of available inferences up front, it’s more likely you can deduce things that can tell you right away that certain answer choices must be false. Also, on acceptability questions, you’re pretty much always looking for the MBF answers by running the rules, and eliminating wrong answers, using a process of elimination approach.
But on a lot of questions, it won’t be that clear why the wrong answers must be false, in which case you’re typically testing answer choices. When that happens, it doesn’t really matter if you say you’re looking for a CBT or a MBF. You’re just going to test it, and you’re going to determine which is the case – if you can make it work, it’s your answer; if it won’t work, then it MBF and you can eliminate it.
I wouldn’t say that the game is “about” what Must Be True; the rules are there to prohibit certain layouts, and may questions just want you to figure out what’s possible. It all depends on the question.
The real crux of the issue, I would say, is this – If there are easy inferences that allow you to eliminate an answer, you want to find them, but very often, students spin their wheels looking for quick inferences that aren’t there, when they’d really be better off just grabbing an answer choice and proving it right or wrong, and on a CBT question, you do that by starting with the assumption that an answer choice is true – if you can generate a complete layout around it that doesn’t break any rules, that’s your answer, and if not, then it MBF and it’s a wrong answer. You don’t want to miss the easy inferences, but you also don’t want to burn a lot of time looking for inferences that aren’t there. So typically, I’ll adopt a compromise approach – I’ll spot check the answer choices and see which, if any, I can eliminate at not much more than a glance. But I’m not going to spend more than about 5-10 seconds per answer choice; if that doesn’t answer the question, I’m brute forcing it. I’m also (before I do anything, actually), looking at my past layouts – often, I’ll have one that tells me that one of the answer choices could be true, in which case the question is over before it starts.
Hope this helps, and don’t hesitate to post about any particular game or question.