DrakeFox: What is it we really want?
The best puzzles, IMO, are those that are integrated in the game world and that make sense, but are not solved simply by clicking on everything. You have played the first GK: consider the cemetery message puzzle. That was brilliant. Perfectly logical and making complete sense, requiring some thought on the player's part and putting two and two together. (Unlike, for example, one of the first puzzles in the game with the clock in GK's grandmother's attic, which makes sense in retrospect, but not much when you encounter it.) For me personally, the single greatest puzzle I remember is the Le Serpent rouge sequence in GK3 (yes, the same GK3). That was pure joy to solve. Jane Jensen can do this stuff amazingly well when she tries.
You are right that puzzles that lack creativity are boring, but puzzles that are too creative are frustrating. I think the trick is in being consistent with the world, and creating a world that allows for wackier things to happen. Some actions in the
Goblins series are downright insane, but once you understand the world's logic, they are not all that difficult to figure out (it was
Woodruff who took that too far), while still providing a challenge, because the solutions are not immediately apparent.
Sam and Max Hit the Road or
The Day of the Tentacle are other good examples of that. It's much more difficult to design a solid puzzle for adventure games that have a serious, real life setting -- which is why the cat moustache thing was ridiculed so much, I suppose, or the rubber duck insanity in
The Longest Journey. And it's also one of the reasons why the puzzle design in the two
Syberias is so atrocious.
I don't really like the
Myst /
Dr. Brain way of doing puzzles. It's fine to throw in a traditional puzzle like that every now and then, but designing an entire game around that feels sort of cheap and rather gimmicky. Unless you are a very creative puzzlemaker and your name happens to be Cliff Johnson.