Li Shenron
Legend
Great thread!
Personally I love traps and puzzles. Minor traps can be placed safely (!?) and just let them be handled with checks, but the most interesting traps are indeed those who are turned into a puzzle, so that they are an encouter of their own.
Actually I don't think the OP's orbs-searching puzzle was a bad idea at all. Probably what went wrong is that (1) the search itself was dependent on search checks instead of players' decisions, and (2) there was no way around the puzzle. These two combined, it means that the DM at some point has to give up and just let the PCs succeed, which raised the question why having the puzzle at all.
But what if instead the 'search' had included something like these:
- the red orb is already in the possession of the PCs (as a puzzle kick-off: they notice the similar shape and size with the lock -> they try to put the red orb in -> first lock element opens -> they get confirmation this is how they will open the whole thing)
- the white orb belongs to the village stingy ol' pawn broker, do you ask him to sell it to you, or do you try less gentle alternatives?
- the blue orb is hidden in a fairly obvious location, maybe there is a clearly visible set of vases in a room that are there for no apparent reason, the players just need to have the idea of checking them out (spice it up with some vases containing a venomous snake or a fake orb)
- the green orb was previously clearly seen embedded into the collar of a guarding/wandering beast: do you go killing the beast or find a more peaceful way to get it?
Each one of these could be a mini-quest or encounter on its own, with multiple possible solutions. In the worst case, if they can't solve the whole puzzle at least the mini-quests should be interesting on their own.
In general, I want to avoid showstoppers i.e. traps or puzzles that must be solved otherwise the whole adventure cannot proceed. If you do use them, then you just have to let the PCs win. It might not always be a bad thing! For instance, even if you guarantee that the PCs will complete all the orbs quests, it doesn't mean they won't be fun, and it doesn't mean you have to let them know that you will effectively let them win
Otherwise, use traps and puzzles only to block non-essential goodies. A puzzle that opens up a small stash of magic potions, or extra treasure, or another boon (e.g. a map of the current location), does not have to be solved at any costs. If they solve it, they get the goodie, otherwise they can just move on.
As a final remark about puzzles, sometimes I have really used some challenging ones that can take hours to solve. Obviously, don't give these puzzles in the middle (or at the beginning) of a gaming session. Try to give them out at the end, and tell the players that they have time until the next session to solve them
It's a kind of homework, but if you have at least a player or two that love puzzles, they will solve it of course, and they will feel very satisfied when you tell them that it was the correct solution. Players who aren't interested, they won't bother and let the others do the job. Only if you really have nobody interested in this sort of thing, then it won't work, so in general you should at least know your own players a little bit 
I don't mind very much...
Solving the story or an investigation case isn't that different from solving a puzzle or trap, and in those cases you don't expect such deep level of RP immersion that players would choose not to understand something about the story on the ground that their PCs shouldn't understand it.
It might be a possible gamestyle of its own, but it has a major drawback: a clever player can choose to play dumb to better portray her Barbarian PC, but a dumb player cannot choose to play clever to better portray her Wizard PC!
So for me it is acceptable (and should be encouraged) that all players play as smart as they can. If a player purposefully does something stupid "because my Int 3 Half-Orc Barbarian would do so", I accept it (and maybe even reward it with Inspiration).
But I also have to accept that it is incredibly horrible for a player to feel she has to shut up when she has a good idea about story/investigation/puzzles, just because her PC's Int is low. So if I put a puzzle/trap in my game, I expect everybody to contribute to its solution, just like they are allowed to discuss out-of-character about the story.
Personally I love traps and puzzles. Minor traps can be placed safely (!?) and just let them be handled with checks, but the most interesting traps are indeed those who are turned into a puzzle, so that they are an encouter of their own.
Actually I don't think the OP's orbs-searching puzzle was a bad idea at all. Probably what went wrong is that (1) the search itself was dependent on search checks instead of players' decisions, and (2) there was no way around the puzzle. These two combined, it means that the DM at some point has to give up and just let the PCs succeed, which raised the question why having the puzzle at all.
But what if instead the 'search' had included something like these:
- the red orb is already in the possession of the PCs (as a puzzle kick-off: they notice the similar shape and size with the lock -> they try to put the red orb in -> first lock element opens -> they get confirmation this is how they will open the whole thing)
- the white orb belongs to the village stingy ol' pawn broker, do you ask him to sell it to you, or do you try less gentle alternatives?
- the blue orb is hidden in a fairly obvious location, maybe there is a clearly visible set of vases in a room that are there for no apparent reason, the players just need to have the idea of checking them out (spice it up with some vases containing a venomous snake or a fake orb)
- the green orb was previously clearly seen embedded into the collar of a guarding/wandering beast: do you go killing the beast or find a more peaceful way to get it?
Each one of these could be a mini-quest or encounter on its own, with multiple possible solutions. In the worst case, if they can't solve the whole puzzle at least the mini-quests should be interesting on their own.
In general, I want to avoid showstoppers i.e. traps or puzzles that must be solved otherwise the whole adventure cannot proceed. If you do use them, then you just have to let the PCs win. It might not always be a bad thing! For instance, even if you guarantee that the PCs will complete all the orbs quests, it doesn't mean they won't be fun, and it doesn't mean you have to let them know that you will effectively let them win

Otherwise, use traps and puzzles only to block non-essential goodies. A puzzle that opens up a small stash of magic potions, or extra treasure, or another boon (e.g. a map of the current location), does not have to be solved at any costs. If they solve it, they get the goodie, otherwise they can just move on.
As a final remark about puzzles, sometimes I have really used some challenging ones that can take hours to solve. Obviously, don't give these puzzles in the middle (or at the beginning) of a gaming session. Try to give them out at the end, and tell the players that they have time until the next session to solve them


That's meta-gaming. In a role-playing game, the PLAYERS don't exist.
I don't mind very much...
Solving the story or an investigation case isn't that different from solving a puzzle or trap, and in those cases you don't expect such deep level of RP immersion that players would choose not to understand something about the story on the ground that their PCs shouldn't understand it.
It might be a possible gamestyle of its own, but it has a major drawback: a clever player can choose to play dumb to better portray her Barbarian PC, but a dumb player cannot choose to play clever to better portray her Wizard PC!
So for me it is acceptable (and should be encouraged) that all players play as smart as they can. If a player purposefully does something stupid "because my Int 3 Half-Orc Barbarian would do so", I accept it (and maybe even reward it with Inspiration).
But I also have to accept that it is incredibly horrible for a player to feel she has to shut up when she has a good idea about story/investigation/puzzles, just because her PC's Int is low. So if I put a puzzle/trap in my game, I expect everybody to contribute to its solution, just like they are allowed to discuss out-of-character about the story.