If you put little hints and clues in to make things easier, your players will come to rely on the DM giving hints and clues. Take it to extremes, and they'll come to expect a safe sandbox world where there are convenient "danger!" signs whenever they're at risk, and even see that as "normal" D&D. They'll come to see character death as a rare and exceptional event, or even a punishment, and if it's allowed to continue they'll start to complain that more difficult situations are "unfair".
But any situation that isn't "unfair" contains so little risk that they'll start to play it for laughs. And they start to get bored.
Any single element of any game taken to extremes can make for a bad game experience. The idea is that it's not taken to extremes -- they don't rely on the DM, nor do they expect to die from levers. They mistrust levers, and investigate them, and know that there will be a fair oppotunity to either die or live from one. They know the DM will give them a good chance for survival, that he isn't against them or trying to kill them, that death, if it comes, will be as earned as success is.
The idea that we must all spend the session rolling dice is as extreme as the idea we must play a diceless game. The middle path -- between the extremes of lever paranoia and invincible PC's -- is often the most satisfying.
Now, people do enjoy lever paranoia, and people do enjoy more narrative games where characters rarely die. I think they're both ends of the bell curve, but they do have their place. I would not have fun at either place, myself.
In order to retain their interest the DM starts building plots, usually involving BBEG's, and bases these plots around the characters (at the expense of the world setting, which becomes blander and more vanilla all the time). You finally end up with a cookie-cutter series of games that really aren't any different from somebody else's game down the street, and it doesn't seem terribly interesting, but nobody really seems to know what's wrong with the game.
A doesn't seem to follow from B here. Plots based around characters don't come at the expense of the world setting, and cookie-cutter doesn't follow from either plain characters or a plain setting, and bored doesn't always come from cookie-cutter.
For anecdotal evidence, I've been involved in games that largely focus on the characters' pasts. This enriched the characters by drawing them closer to a world on the brink of collapse due to the evil force that they know face, that they only knew vaguely of in the beginning. The games played to each character's unique strengths as they meshed with the setting -- the warrior had a warrior's school, trained under a weapon master, and found an ancient swordmaster locked away in his own sword atop a precarious mountain peak where towns of goblins lived on the back of rocs.
Character-based, but neither cookie-cutter adventures nor vanilla worlds.
What's wrong with it, of course, is that it's seen as the DM's job to arrange matters so the players can't lose -- which means that winning loses its value, so the whole thing starts to feel somehow pointless.
Believe me, I've seen it in other peoples' games.
The DM arranging things so that players can't loose would be as extreme as arraning matters so that the players can't win. The same is true for "extreme ends" -- if there is only a 5% chance of living, or only a 5% chance of loosing, there is often something lost from the game. In the case of it being vastly unlikely to succeed, you lose a sense that your character matters in any sense. In the case of it being vastly unlikely to fail, you loose the sense that your abilities matter in any sense.
The alternative to traps like this is not cakewalk bland quests, and one does not need to like cakewalk bland quests to consider traps like this unfair.
I repeat that the trap is not undetectable. It is merely undetectable by rolling d20's. Surely the view that some situations should challenge the players, rather than the numbers on the character sheet, isn't an extreme.
Atoms aren't undetectable, they're just not detectable with the naked eye. Would it be fair to request the players to "think outside the box" and "develop magic capable of observing atoms" in order to avoid death from them?
I do believe it *is* an extreme, actually. I would never demand my players showcase themselves as master orators to play a spoony bard, nor would I request my players to demonstrate knowledge of swordsmanship to wield a weapon. I would never ask my players to try to outsmart me, either. They probably can, more often than not, but I wouldn't require it of them. If I had a puzzle, it would be up to the CHARACTERS to find it out, not the PLAYERS.
What I would demanad of my players is to put a high score in Strength if their character is a strong and mighty warrior, or to give them the Ride skill if they want to charge into combat astride a noble steed -- I demand they play the game by using the rules to give their characters capabilities. Then I test these capabilities.
I also don't believe that this trap was a really a challenge for the players. A challenge implies something that tests your ability to succeed. This trap did not test any ability to succeed -- they already had succeeded in their goal of getting the McGuffin, and their ability to precieve the trap failed on every count. Even giving that, having the monk pull the lever shows that they are thinking about how to succeed. If this was a graded test on what to do when encountering a lever in a dungeon, they scored an A+. But the monk died regardless.
So not only is it rather unorthodox to test the players to be what their characters are, this lever only tested the abilities of the players to read the DM's mind.
Surely it's the crowd who see no other answer save the numbers on their character sheet for the resolution of in-game problems who represent the extremes.
True, they would be. Unfortunately, those who oppose your view don't always fit that mold. Rather, in speaking for myself, I see the numbers on the character sheet as an essential skeleton for the resolution of in-game problems that should not be ignored. Generating those numbers and choosing when to use those numbers is the "player challenge" of D&D -- interacting with a world in a way that will lead them to success and treasure. Forbiding them the valid use of these numbers (a trap "undetectable by rolling d20s") is something I consider unfair.
They are not the only answer, but they are where most answers should start.
And those who refuse to deviate from the numbers on the sheet ARE just as extreme as your position -- they have no history, they have no personality, they have no item that does not translate into some form of bonus or bost for a mechanic. But those people have fun doing what they do. They just share, with your position, a gameplay style that I (and, I'd argue, the majority of D&D players) see as not only not fun, but also not fair, according to where I (and, it seems, this poll) believe the standard assumptions of D&D fairness lie.
My style isn't the One True Style, I don't hold anything against those who want to outsmart the DM or those who want to tell a story without fearing death. But it is a valid middle ground which neither inspires a fear of levers in my players, nor makes my players feel that winning is meaningless. Rather, this makes them feel that winning is hard-faught and noble, and that death is a constant threat but that they can feel confident attempting the risky and chancy without being assured of their demise.