Clarify the situation

Whenever my players propose a plan that will in no way work, I first assume that I didn't describe things well enough.

Rogue: I sneak into the throne room, steal the document we need, and leave.

DM: (This guy must have a death wish. What am I missing?) The hallway leading up to the throne room is brightly lit by torches at regular intervals. There are columns in the hallway, but the closest one is 15 feet from the throne room door, which is currently open. Two royal guards flank the door. One of them is watching your group attentively. How exactly do you sneak up?

Rogue: Oh, I see...well, I'm a halfling, so I hide behind the fighter. I'll then flip up the hood on my cloak of invisibility and creep up to the door.

In this situation, the player had a different image of the situation in their head. Maybe I forgot to mention the guards or the torches in my initial description. Either way, additional information allowed the player to adjust their plan.

What happens if the problem with a plan is something that the player doesn't know? What if every member of the royal guard has visão verdadeira, allowing them to see invisible creatures?

DM: (Crap. He's going to get caught and it's going to feel cheap.) Alright. Your cloak shimmers and you turn invisible. Make me a History check. You grew up in this town, so roll with advantage.

Rogue: (rolls) I got a 13.

DM: (That's not quite enough, but I'll give them something.) A few years back, someone tried to assassinate the king. As a rogue, you heard that the assassin used magic to change their appearance. You also heard that the assassin didn't even get one foot into the throne room before they were cut down. These guards don't bother taking prisoners. What would you like to do?

When giving out information, there are a few options.

• Just tell them. If it's something many people would know, don't hide it behind a roll.
• Make them roll. If you're not sure, have one or more of the party members roll, even the ones who aren't coming up with crazy plans. Maybe the party wizard can advise the rogue about the dangers of visão verdadeira.
• Change the information. Nothing is real until you say it at the game table. You planned for the guards to have visão verdadeira, but maybe it's better that they don't. This one feels weird when you're doing it, but I have found it to be an invaluable tool.

Involve the party

When a character is about to do something crazy, it's important to remember that you control the flow of time. Events do not have to transpire right when a player says they do.

DM: (addressing the other party members) You see your halfling friend hide from the two royal guards, and then go invisible. What would you like to do?

Bard: I'm going to walk up to one of the guards and distract them with conversation.

Fighter: I'm going to watch and see how this plays out.

Paladin: Whatever you three are doing, I don't want any part in it. I walk away from the throne room.

Pausing the flow of time gives the other players a chance to involve themselves, even if it just means leaving before everything blows up. If the other players are on board, you've turned "that one player is running off again" into "cool impromptu heist".

If you've clarified the situation enough and multiple players are discussing their options, it may be time for you to leave the room. (Graças a Skrrp in the comments for this one.) I'll often take a break when a plan has potential, but still needs some ironing out. Like it or not, Dungeon Master is a position of authority. With you out of the picture, the only people to say "no" are the other players, and sometimes it's easier to accept "no" from a peer. Your return can also mark a logical moment to bring the discussion to a close. "So, what's the plan?"

Step through the plan's execution

Getting on the same page when it comes to a plan's details is difficult, so I've found that it's easier to take it one step at-a-time. After a long-winded plan explanation, I often find myself telling the player, "ok, so what action are you taking first?" If the player is just throwing ideas out, they'll often back down with a "well, I was just thinking". If so, then problem solved. A crazy plan is only crazy if you execute it.

So you've clarified things, involved a party member or two, and the doomed plan is still being executed. Now what? Is it time to tear up some character sheets?

In my experience, most seemingly-deadly situations can be made less deadly with a few tweaks. It's easy to say "the guards spot you trying to sneak into the throne room and you die in a hail of crossbow bolts", but that's not very fun. My go-to solution is to have the plan fail at the earliest (or safest) opportunity.

DM: (addressing the rogue) You see your bard friend chatting up one of the guards, as the paladin strides away. What would you like to do?

Rogue: I sneak past the two guards, being careful not to bump into them.

DM: As soon as you step out from behind the fighter, one of the guards points his heavy crossbow directly at you. "What do you think you're doing, small one?" He appears to be able to see you despite your cloak. What do you do?

If the player was caught stuffing important documents into their pack, it would be much harder to justify letting them off easy. Instead, the plan failed at a point where the party has an out. They may come up with a creative lie, the bard may charm a guard or two, or the fighter may bribe the guards and apologize.

Deve haver alguns consequence for failing, but that consequence shouldn't always (or usually) be death. Whenever possible, the consequence should move the story forward, rather than grind it to a halt. (This concept of failing forward is used in Dungeon World, a more narrative system, but applies to 5e as well.)

Outras considerações

None of these steps involve telling a player "no". If I have clarified the situation, involved other party members, and the player has reached a fatal step in the plan's execution, I ainda don't say "no". Instead, I say, "Alright. Describe for me how your character dies." (I ask this right before the point of no return. If a player backs down here, I allow them to do so.)

That doesn't mean that you should Nunca say no, but that you should reserve it for special cases. If a player says, "my character kills the annoying child", and your group has an (explicit or implicit) agreement not to harm children, then "no" is the appropriate response.

My answer thus far has assumed a player acting in good faith. Stepping through the plan's execution can help weed out the trolls ("I run at the guard that's pointing a crossbow at me and stab him in the face"), but it's hard to be sure. If a player is intencionalmente breaking the game, it's time for an out-of-character conversation.

Sometimes you just gotta put your foot down.

I have actually been in a very similar situation myself, from the player's perspective. I wanted to try and free some slaves from captivity, and I had a whole plan formulated. Using stealth, and a few of my druidic spells, I could slip past the guards, and free the prisoners. However, the ultimate outcome was that it was me, vs 5 very highly trained, and well armed guards. It would not work, no matter how you spun it.

If you can't explain to the player that the plan won't work, you need to put your foot down, and estado that it won't work. Talk to the player, and explain that the situation is unfeasible. The PC is not a one-man army, the party is not a 4-man army. You'll end up in jail, best case scenario. Stop wasting time, because it's hindering the gameplay.

Dito isto...

The player's investment in this task could potentially open up a sidequest. Right agora, the target is well protected behind walls and guards, but maybe, somewhere down the line, they may become exposed. Perhaps somewhere in a future game, you can drop a hint to the player, maybe another NPC is clued in to the target's activities, and has learned of an opening. This could be the opportunity they're looking for. Player investment can't always be dealt with on the spot, but it should not be denied, if you can help it.

It sounds to me like you and the player have an out-of-universe disagreement, but you've been trying to resolve it by using in-universe language.

The player is saying, "My character wants to assassinate the government official" (in-universe language). What they really mean is, "I want to accomplish this objective in a dangerous and exciting way" (out-of-universe language). But you're hearing this as, "I want to do something foolish and reckless."

You're saying, "The official is surrounded by armed guards" (in-universe language). What you really mean is, "I'm running a game where dangerous antics like that aren't going to work" (out-of-universe language). But your player is hearing this as, "In order to do that, you need to surmount this one particular obstacle."

I don't have enough experience to offer you specific advice. But you should try to understand what your player really wants, and you should make sure your players understand what you really want.

The answer to this question is 100% dependent on the individual player. There is no one size fits all answer. You are trying to convince a player to act a certain way, and that's simply hard.

That being said, what you've described sounds an aweful lot like the bumpers they put on bowling lanes to let kids bowl without getting gutter balls. Many people respond positively when those bumpers are taken away, and they scratch a few frames.

In other words, let the player sign their character's own death warrant. Well, perhaps not that far. You mentioned you don't want them to die. But you can have their actions take the group substantially further from the goal. Perhaps the quest was to convince Noble A to declare war on Noble B. It wasn't a hard task before, especially since your group has the best bard since the invention of salsa! Want to attack the government official? Fine! Quest is the same, only now you have to do it from a dank cell in the bottom of the tower. You have to work through several irreputable individuals who have connections outside of the tower to influence Noble A from within your cell. And, once you do, now you owe someone a favor... and it's not the kind of person that you want to have to owe a favor.

A handful of these tipicamente leads groups to reign in their trouble-member. Ações têm consequências. In fact, you may find this increases the excitement level of the campaign, simply because players know that their actions do indeed have consequences.

My first response when a player proposes something wacky or otherwise potential game-disrupting is "Why do you want to do that?" or "What do you hope to accomplish by that?".

If the player is wanting to assassinate an NPC for a good reason then I let them try. Roll the dice; see where it takes the story. If they can't justify their decision, or have a weak motivation, I generally get them to confirm they realmente want to do this, or even talk to the rest of the party with the player in question if that is something they support. Roleplaying games are a group activity after all.

Delegate this problem to the other players

I used to run into this problem now and again as a DM, but I don't any longer. Why?

I operate with the understanding, as we discuss things in Session Zero, that the core conceit of this game is that "a party of adventurers, each with a different batch of skills and abilities to offer, gets together and has adventures." What your player is trying to do, apparently, is initiate a series of "split the party" episodes and do a solo adventure.

From the notes in your question, you seem to have a similar idea about how the game ought to proceed: PC's work toward various goals and objectives together. As you put it, there is a plot that you want to maintain.

I am concerned with both derailing the plot and sapping the session time.

You like structure. (Apparently, some of your players also like structure).

Each time the player offers up a zany idea to you, ask the group: Are you going along with this?

Let the other players tell this player "yes" or "no" to these zany ideas.

To the player, a follow up to that question goes something like this:

Why are you telling me this? You need to be selling this course of action to the rest of the group. If they agree, then we'll do that.
If they don't, I am not going to run two separate campaigns: one for you and one for everyone else.

When they try to divert you with a zany plan again, remind them to sell their idea, or their plan, to the rest of the party and thus to the rest of the players.

Let the party succeed or fail together

If the other players agree to try one of these zany plans, then let them try! All kinds of fun and hijinx can ensue. They will succeed or fail on their own merits, as a group. All of the players will have a chance to make an input to the plan. This increases the chances that the plan won't be hopeless.

• Failure need not mean PC death. It may mean being locked into the deepest dungeons of the Cult of the Unnameable Thing's temple, while the ritual for PC sacrifice is being prepared. The next adventure becomes the tried and true Escape/Jailbreak adventure.

If the other players don't agree, then it just doesn't happen.

You get back to the game ...

"OK, where were we? Right, you leave the audience with the Duke and are in the square outside the palace. Where do you go next?" ... and so proceeds the adventure based on what the group agrees to do together.

As a DM, some of your role is as a coach. In this case, I think your efforts at coaching the whole group of players in more team building will pay off.

_{Note: this answer challenges in the frame of the question in terms of the assumption built into the question that this is a problem for the DM to solve. It's a problem for the whole group to resolve together.}

If a player wants to do something, play it out. The closest there is, as a DM, to "you can't do that", is "You failed". If it's truly impossible (scaling a smooth ice wall barehanded), don't even roll -- because rolling gives the impression that they might succeed. Just say, "It's too slippery and you can't get a grip."

Ultimately, it depends on how they're acting. Are they just not thinking things through or are they basically disrupting the game? ("Disruptive" for example, can be players who insist on making everything focused on them, or ignoring the "reality" of the game universe.) In the end, play it out in detail and see what they come up with. Maybe they think of something you didn't imagine. Or... and you have to be prepared to follow through with this... maybe their character just dies trying.