MI4 was the first MI game I played, and I thought it was great. So, my opinion on it is probably rose-tinted. But, I did have a blast playing it, and I think that most of the stuff that happens before the final act (when you get Monkey Kombat and that weird... robot... thing... which I won't spoil...) is actually pretty funny.
For me, the Lucasarts design philosophy is so much better formulated. By ensuring there were no dead ends and players couldn't die, they allowed their players to explore, immerse themselves in the world and not be too anxious about having missed something.
Sierra, on the other hand, loved having dead-ends, sudden deaths and stuff that could be considered "unfair". I don't think this philosophy is worse than Lucasarts', though! The effect you got was very different, but still valid: instead of experiencing the game as an immersive story, you experienced the game as a complex system of puzzles that had to be solved over multiple playthroughs (because you'd have to restart to avoid dead-ends). Put simply, in a Sierra game, the "save game" feature becomes part of the gameplay.
My problem with that is that Sierra games weren't very clever about this. Rather than use this philosophy to make games in a setting that benefits from this philosophy, maybe making a science fiction game about parallel universes, where the save game feature is part of the lore because you have a time machine/universe hopping machine and are able to manipulate time/reality to solve these puzzles, they used it to make games about stereotypical characters in stereotypical settings. (Fantasy mashups ala KQ or Mixed up Mother Goose, generic scifi jokes in SQ, the Arthurian archetype stuff in Conquests of the Longbow.) I enjoyed these games (especially Longbow) but I think I enjoyed them in spite of Sierra's design philosophy.
And it's not coincidence that my favourite Sierra games, the GK games, had no dead-ends (I think) and fewer arbitrary deaths. In other words, they aligned closer to Lucasarts' philosophy.