All these reverse engineering models are of course artificial. That doesn't mean that one of them might not help a writer tease out a plotline, but they're not necessary or essential. I think the 3-act model, with Act Two being twice the length of Acts One and Three individually is useful as a yardstick (a yardstick is a way of using a familiar measurement (literally a yardstick) to represent a quantity that is meaningful to the audience (e.g. an acre is about the size of a football field). In my reviewing of scripts by inexperienced writers (over 300 of them--I'm an inexperienced writer myself), if the writer doesn't get to a major plot point (traditionally the end of Act 1) by the midpoint of the story, then the plot will fail to engage a reader, no matter how exciting the remainder of the script is. There will be way too much expo. I think writers should hold the three act yardstick in their heads while writing (and certainly while rewriting), but complying with the "formula" in no way guarantees an engaging screenplay, and it is possible to modify the 1-2-2-3 grid and still have an effective plot.

BTW the worst script I ever wrote was one that followed Snyder's approach faithfully (including the note cards). That's not Snyder's fault but mine in focusing on a template instead of the needs of the characters.