When is character death appropriate?

[Drache64]

That's the question: When is character death appropriate?

My philosophy is that a character death should be earned by a series of bad decisions, or in a fair combat with a series of bad rolls.

But that's not gospel to me, I am open to others and would love some more perspective. Should character death be inflicted upon players if they misbehave? What about used for plot purposes? Can a DM kill a party member just to shake stuff up?

What do you guys think?

[NichG]

In systems with raise dead/etc, I think character death is fair game as long as it could have reasonably been prevented somehow and is not just vindictiveness on the part of the DM. E.g. it's not okay to decide 'I'm going to kill this character', just like it's not okay to decide 'this character is going to be successfully captured'. But it's fine to say 'this situation is dangerous, and the enemy is going to focus-fire on the weakest character in the party to try to inflict some damage that's hard to heal to soften them up for their boss' or 'this trap is designed to kill since just injuring someone is a temporary inconvenience and not really a protection; so if someone sets it off then its straight to the save or die'

For cases where it's a more permanent end, I'd say that there should always be some kind of significant exchange, trade, or consequence to the arc of the campaign associated with the events of the death - this character sacrificed themselves so that another might live; or they got themselves killed overreaching for a power that is now running wild over the world; or they were betrayed by someone the party thought was an ally and now the direction of things moves towards obtaining vengeance. But with permadeath, getting killed by a random encounter with boars just encourages a 'disposable character' mentality.

[Silly Name]

First of all, this question heavily hinges on what system you're using and what tone you want to convey. For example, if you're running an AD&D dungeoncrawler, players will know their characters can die to a bad roll or a no-save trap or any number of brutal obstacles they'll meet, and will act accordingly. Likewise, if you're playing Call of Cthulhu we all know we came here for supernatural eldritch horrors that can unravel our bodies and/or minds with a mere touch or even their sole presence.

But maybe we're playing a session of DungeonWorld or D&D 5e or Exalted, and the players are a bit more attached to their characters and their role in the overall plot of the campaign. Deaths that feel random and lack dramatic contribution don't really fit with the base assumptions of those games. You can still die to the dice or by making exceedingly dumb decisions, but we expect those events to be a rarity and to not be sprung on the players without warning.

I think your base assumption, however, is a very good one: no matter system or tone, players should never perceive character death as the consequence of the GM "punishing" them, or purposefully creating unwinnable scenarios. Using it as a tool to keep a party in line screams dysfunctional social skills and evokes tales of horrible GMs.*

Character death as a plot point can be one of those fair moments, too - but, again, how you go around handling it depends on the baseline assumptions I spoke of above: going back to Call of Cthulhu for a moment, players know their characters are likely to die in the course of an adventure, but also know those deaths can integrate in the plot (compare to the hypothetical AD&D dungeoncrawler, where expectations of dramatic, plot-affecting death scenes are low). So in a CoC campaign, the GM doesn't need to worry particularly about character deaths.

In other types of games, like 5e, an usual tidbit of wisdom is that you should make sure the player is ok with their character dying because the plot demands so. Either by discussing this with them, or because you assume you know them well enough they'll be ok with it, but I'd still recommend talking about it even in this case.

I also know you can run DungeonWorld and 5e as highly-lethal dungeoncrawlers, but it's not the usual type of game you'd run in those systems - both system and agreed-upon style of play matter, with some systems being better suited for certain styles and tones.

In short, make sure everyone's on board with how you are planning to handle character deaths, don't be a jerk about them and remember that we play games to have fun.


*Also note there's a difference between actual fairness and perceived fairness: while we tend to accept "screwed by the dice" as nobody's fault but statistics', if you spend a whole session getting screwed by RNG you probably won't think it's fair.

[arrowed]

Silly Name's statement is good.
System and individual game inform a lot of where Character Death should be in prominence.

Something that I've learned over my experience with RPGs is that death is an easy consequence to write, but often unhelpful to actually follow through on. The threat is compelling because the actuality of character death is BAD. Your plot threads are cut short, you lose not just the character but all the ways they connected to the world, and your equal-level-appropriate-equipment replacement is still going to play catch-up to the rest of the group.

There are 2 games that spring to mind for how they deal with this: Torchbearer and Spellbound Kingdoms.

Torchbearer is a dungeon-crawler, with limited inventory, risky and fickle magic, and a very gritty feel. Dealing with goblins is generally done in a Conflict, but a Kill Conflict is only one type next to Drive Off, Trick, Flee, Capture, etc. You absolutely CAN try to massacre every goblin in sight, but if your team loses all of you will die, and even if you win, if you don't win perfectly, some of you might still die. 9 times out of 10 in play, the people I've played with have chosen to Drive Off rather than kill, and use traps, strategy, and environment to keep our enemies out for good. Importantly, outside of a Kill Conflict, your character CANNOT die in one roll. This is specifically stated in the GM section. You might get cursed, become injured, become sick, etc., but unless you literally leap off a cliff or otherwise bypass the 'make a roll' step, you have at least 2 rolls to make to not die.

Spellbound Kingdoms is a game of Swashbuckling. It is the setting I'd use to run Dragon Age oddly enough, as well as Pirates of the Caribbean or the Princess Bride. All Characters in Spellbound have a Heart Score, which determines how many Inspirations they can take. If one of your Inspirations is 4+ in intensity, you will not die an unnatural death. Regardless of whether someone tries to stab, poison, or drop a warship on you, Fate will intervene and you WILL survive. You might lose an arm or an eye, or something similar, but you will not die. Importantly, PCs are always capable of hitting the magic 4+ inspiration. As long as they don't intentionally build to NOT have such an inspiration, they will have literal plot armour. The only way to get around this is to destroy an inspiration (e.g. killing a beloved, destroying their pride) or being a high-level assassin. As a GM you have to go out of your way to have your enemies be able to kill the players.

My opinion's mostly that it depends on game, players and setting, but that the game is more interesting where the consequences of failure both in and out of combat are regularly more varied than Character Death. A DM should never just kill a party member to 'shake things up', unless their player is looking to retire them and specifically co-ordinates with the DM. There are 101 better ways to inject excitement into a campaign. Character Death is merely one of the most universal aspects. Killing a Character might create buy-in for 5/6 or 3/4 of your players at the price of utterly butchering it for another.

I'm going to stop here, because I could write like 10 more paragraphs and I need to do other stuff.

[MoiMagnus]

First, character death and player elimination are different things.
A character can die, but if the universe has resurrections (high level D&D) or cloning (Paranoia), that's not a character death.
If the players play a family or a guild or some sort, individual character are just the face of the entity played by the player, and the player will just come back with an heir to continue. (BTW, some players will use that trick when their character die and they don't want to lose all the efforts put in the background: just bring a family member to seek revenge).
Conversely, a character heavily injured in a realistic system can result in the player being eliminated without the character dying. And even if the system assume character death is meaningless and you should just reroll a new character in 10min, the player might feel otherwise because of his emotional bond with the character.

Character death is not a problem per itself, as it's just a game mechanics. As every other game mechanics, it should be used fairly. And depending on the system, player can have significant agency on it (no death if no mistake or willing sacrifice), or have few control on it (random rolls can instantly kill you), or anything in between.
[Note: Character death is also frequently linked to failure. But "When is PC failure appropriate?" would be a full thread on its own. But a very important point is: just because you chose that your PCs cannot die for meta reasons does not mean your PCs cannot fail.]

Player elimination is very much one. It usually forces the player to be reintroduced into the game (through a new character), with all the usual problem coming with introducing a player in a campaign (difficulty of linking the new character to the old ones, feeling of unbalance, ...). It potentially put the player out of the game for the remaining of the session, which is never fun. In the end, player elimination is something to be very careful with, and you'd rather have player agreement for it.

[Quertus]

It feels like I really ought to have a pithy one-liner for this, like "balance to the table" or "knows your group" - but, if my senility is to be believed, I don't.

As an advocate of Player Agency, and an opponent of railroading, unless the system says otherwise,

Do this:

My philosophy is that a character death should be earned by a series of bad decisions, or in a fair combat with a series of bad rolls.

Not this:

Should character death be inflicted upon players if they misbehave? What about used for plot purposes? Can a DM kill a party member just to shake stuff up?

If the system *does* encourage narrative railroading, you should make sure that the players understand and buy into that. Really, making sure that the players understand the system's and your own stance here really sound like they're a part of a balanced breakfast successful session 0 one way or another.

it's not okay to decide 'I'm going to kill this character', just like it's not okay to decide 'this character is going to be successfully captured'. But it's fine to say 'this situation is dangerous, and the enemy is going to focus-fire on the weakest character in the party to try to inflict some damage that's hard to heal to soften them up for their boss' or 'this trap is designed to kill since just injuring someone is a temporary inconvenience and not really a protection; so if someone sets it off then its straight to the save or die'

Strongly agree.

But with permadeath, getting killed by a random encounter with boars just encourages a 'disposable character' mentality.

Just? I disagree. A death of logical consequences of events outside players' control can also encourage the players to play smarter, whether that's attempts to predict these events (Divinations, Gather Information), attempts to control these events (some random encounter systems activate at different rates or with different encounters depending upon player actions), better ability to handle random encounters (specific preparations, or straight-up stronger characters), better tactics, better contingency plans (one-offs, willingness to retreat, hirelings to throw under the bus as you retreat, etc).

Railroad death, OTOH, probably *only* encourages that "disposable character mentality".

-----

An interesting offshoot of this thread is, is "save or die", "one failed roll and you're dead" good game design?

[Pauly]

Firstly it depends heavily in the system. In some games The characters are very squishy (warhammer RPG comes to mind, as well as some of the more realistic modern systems) and players understand and accept that there will be casualties if you’re playing that type of system.

You may be playing in a hardcore setting of a usually more forgiving system, for example a D&D setting with very limited access to resurrection. Again the social contract in entering that arena is that character death is a real risk and the players understand and modify their actions accordingly.

However, generally speaking RPGs are very forgiving to characters. The DM will question Leroy Jenkins if he really wants to do that, healing is plentiful and easy, other party members will try to prevent character death from happening, resurrection is an option. In a forgiving environment character death needs to be unusual and extra-ordinary. Players will feel cheated if death is sudden, arbitrary or capricious.

In a traditional RPG players need to feel that it was their agency that caused the death. A series of bad decisions, an extra-ordinary run of bad rolls, a noble and heroic sacrifice, having the rest of the party turn against them, etc. Players should be able to recognize, unprompted, 3 or 4 opportunities they had to prevent the death.

I have no qualms about killing characters when the player has had the opportunities.

If a player’s death is about to arise from something they had no agency over (for example a save or die roll on crossing a bridge, another party member deliberately antagonizing the ancient red dragon in it’s lair) then I will fudge the result to a non lethal outcome.

Edit to add:
In addition to the in-game character reasons there is one situation where I feel it is OK from the meta game view for the DM to assassinate a player’s character. That is when the player is using the character to disrupt other player’s fun. If you’re being a male appendage to the other players you lose your narrative protection. If you continue to be a male appendage you will get a target placed on your back.

[NichG]

Just? I disagree. A death of logical consequences of events outside players' control can also encourage the players to play smarter, whether that's attempts to predict these events (Divinations, Gather Information), attempts to control these events (some random encounter systems activate at different rates or with different encounters depending upon player actions), better ability to handle random encounters (specific preparations, or straight-up stronger characters), better tactics, better contingency plans (one-offs, willingness to retreat, hirelings to throw under the bus as you retreat, etc).

Railroad death, OTOH, probably *only* encourages that "disposable character mentality".

I don't think its death that actually encourages smarter play. I've been in a campaign where there were super-lethal things and we played very carefully just like you describe - use hirelings as ablative armor, never go into a dark tunnel, don't touch anything, etc. Notably, I was still playing the character I started with when that campaign ended. It wasn't actual random character death that encouraged smart play, it was very very close shaves and near survival and understanding how mistakes could lead to death.

On the other hand, I was in an L5R campaign where we were walking along the road, got ambushed with no warning or real opportunity to adjust, and got three arrows in the heart and character death during the surprise round in session 1. I definitely detached a bit from that campaign. It wasn't railroaded, it wasn't what the GM intended (in fact he even ended up retconning it since it was SO different than what he had understood about the combat system), but it also was clearly not something that any sort of reasonable attempt to engage with the premise or game could have done anything about. Since it was essentially random death, not due to a mistake or lack of reasonable preparation, it communicated 'what you do doesn't matter'.

The real art to using death is to make it 100% clear when there are things such that 'if you mess this up, you will die' but do it in such a way that the player is actually able to understand that, respond appropriately, and as a result of reacting to the world correctly, have their character survive. Seeing caution rewarded with survival is more effective than seeing death happen for a reason that may be hard to pin down or explain beyond 'the dice were unlucky tonight' or whatever.

An interesting offshoot of this thread is, is "save or die", "one failed roll and you're dead" good game design?

IMO, only if there's always a clear chain of causality leading to that moment. A good example of this might be something like the Vrock dance. You have a bit of warning and you're informed 'if you don't deal with this, you're probably going to die'. So if you deal with it or fail to deal with it, you can explain why things went that way. If I take that to Save or Die mechanics, then I'd say that Save or Die abilities should tend to occur on very iconic antagonists which leave some kind of hint or suggestion before you actually encounter them. The classic example is seeing a bunch of statues in a dungeon and then, hey, basilisk! If you get petrified in round 1, you can at least look back and say 'I should have known'.

But if you just had some cultist who happened to have a Flesh to Stone wand and was mixed in with the other cultists, and now suddenly you're dealing with a Save or Die, then I don't think that's a good usage of what that kind of mechanic is for.

[Composer99]

That's the question: When is character death appropriate?

My philosophy is that a character death should be earned by a series of bad decisions, or in a fair combat with a series of bad rolls.

But that's not gospel to me, I am open to others and would love some more perspective. Should character death be inflicted upon players if they misbehave? What about used for plot purposes? Can a DM kill a party member just to shake stuff up?

What do you guys think?

I would say it is always acceptable for a character to die as a result of an aggregate of bad decisions and/or bad die rolls, with some systems making such events easier or harder.

I would say it is never acceptable for a DM/GM to kill a character for plot reasons without buy-in from the player, and never acceptable under any circumstance to kill a character in order to punish the player. (Once you have kicked a player from the game, then what happens to their character - dies, becomes "disappeared" in the Orwellian sense, whatever - is fine.)

Beyond that, I would say that the extent to which a character's death can be a result of seemingly capricious events ("you try to cross the rickety bridge [*player rolls badly*], but it gives way and you fall to your death") depends on both the game system and the "table culture", as it were. I would ensure, however, that you avoid "gotcha" results. Adapting the plot of, say, the Star Trek episode "Justice" would be, I think, going a step too far unless it's an NPC who is under the gun. I do think that if characters are delving into monster and trap-infested holes in the ground (aka dungeoncrawling), they are doing so in full knowledge that they are taking their lives in their hands, at any rate.

The key is good communication. If the game system and table culture entail highly-lethal gameplay, it is important that players understand and accept this as a condition of their participation.

An interesting offshoot of this thread is, is "save or die", "one failed roll and you're dead" good game design?

It depends on the game and the tone it is trying to convey, I would think. If you're playing, say, Dread, then darn tootin' you're done once that tower collapses, but it's not for every game. Such a mechanic would be very off-putting in the example given of Spellbound Kingdoms.

[GreatWyrmGold]

That's the question: When is character death appropriate?

My philosophy is that a character death should be earned by a series of bad decisions, or in a fair combat with a series of bad rolls.

That's basically my philosophy, with more emphasis on the "bad decisions" and less on "bad luck". Losing a character to stupidity feels embarrassing; losing a character to the dice feels asinine.

The campaign in question also matters, of course. Dice deaths in a dungeon crawl don't feel so bad as they would in a more story-driven campaign, for instance, while plot-mandated deaths can fit as the conclusion to a character's arc in the right character-driven campaign.


One cardinal rule: Deaths from DM incompetence are always worse. Two examples from a certain Pathfinder adventure path my local group has been running:

• In the first segment, when the players are level three or so, there's a CR 7 fiend in a locked room. If the players break the lock, the fiend (who casts invisibility and some buffs when it hears people doing things) attacks anyone who enters the room. Our genius DM missed everything except "fiend" and "room," so the party was thrust into a nigh-unwinnable fight, which lead to an in-retrospect-entertaining trio of character deaths. First the cleric threw himself at the monster, telling the others to save themselves; then the barbarian joined, declaring that he wouldn't let the cleric hog the fight to himself; then the gunslinger (whose gun was shattered in the previous fight) also stayed behind, because the player didn't want to buy a new gun.
• Later, after I took over from that DM (for obvious reasons), the players were attacking a fortress occupied by ogres. I had more ogres come out, a few at a time, but didn't realize how hard I was pushing the players until they started dropping. There were only two survivors. (The players generally saw it as a consequence of their own recklessness—and fair enough, some of the dead ones were being extremely reckless—but I probably shouldn't have thrown all those ogres at them at once.)

[Spellweaver]

Any Time.

Assuming your playing a RPG with rules like attack, damage, hit points and creature/character death.

If, for any reason, you don't want character death....and the game has rules for it: You might be playing the wrong game.

Part of the fun of any game is loosing.

Also, in most games like D&D, death is not a big final ending. A character can die and come back. Ghostwalk had some fun with this.

Kill a character for a players misbehavior? Er...no. Not exactly.

Kill a character to shake things up? Er...no. Not exactly.

Kill a character as part of a plot? Er...no. Not exactly.

The plot one works....like if the player wants their character to die next week so they can play a new character or such. But for the DM to have a plot saying not matter what Bob's character will die two days from now is wrong. Though it works the other way around: the character death(s) make the plot.

For misbehavior? Well, if they like tell a joke and the DM gets mad they should not have rocks fall on a character to kill them. But like if, in game, the character attacks a huge dragon...then sure, kill the character.

Shake things up? Well, you want to shake things up....and have character death be a part of the shake up....not the end.

[Tvtyrant]

Ask your players when they think CD is appropriate.

Personally I let players choose whether their characters live or die, instead of it being random chance.

[Aneurin]

I treat character death as always being appropriate... when a character is in a situation that could plausibly result in their death. So, walking down a sunny street isn't going to kill someone - but I've absolutely had a character trampled to death by a horse because they got in its way.

That said, I'm not going to arbitrarily kill someone because they're crossing the street, for instance. I try and give - at a bare minimum - three points of failure before character death, one of which must be a choice that the player controlling the character makes. To take the trampling incident; the character heard the horses coming, and chose to linger in the worst possible place. The character then failed a "get the hell out of the way" test at the last moment. The horse then failed a "don't run into the idiot" test. There was also a damage roll, making a fourth failure point (the horse rolled perfectly for damage; almost anything less and the character would have lived, albeit horribly maimed). Wasn't the outcome I expected; wasn't what the player expected, but we accepted it as reasonable and moved on.

I find this extends fo combat, too; the characters can choose not to engage (fleeing, surrendering and playing dead are valid, if not always good, options), there's an enemy attack roll and their defense roll (making three points of failure total; two if your system of choice lacks an active defense) and, often, damage.




I do not, have not, and never will punish (or reward, for that matter) out of game behavior in game. I can't imagine a situation in which it would help anything.

I also don't deliberately kill characters or set them up to fail and/or die. I'll establish a scenario and turn the players and characters loose; whether they live or die is up to them, and a little luck. The only time I can think that I'd make an exception is if I knew a player wanted a character killed off rather than retired for some reason. Because at the end of the day, what is gained by a character death? It can add to the story your table is building and open a world of character development to other characters, but it's the end of one character's story and means the controlling player can't usually interact with the game until they make a new character and get them introduced - and where's the fun in that?

[Quertus]

I don't think its death that actually encourages smarter play.

OK, fun!

I've been in a campaign where there were super-lethal things and we played very carefully just like you describe - use hirelings as ablative armor, never go into a dark tunnel, don't touch anything, etc. Notably, I was still playing the character I started with when that campaign ended.

… that sounds like "we were already playing smart", so… I'm not sure what to make of that.

It wasn't actual random character death that encouraged smart play,

Well, of course not - "rocks fall, <roll> - *you* die" does not encourage an intelligent response.

it was very very close shaves and near survival and understanding how mistakes could lead to death.

Death also shows how mistakes can lead to death (I'm picturing Lost here).

I'm not disagreeing that close shaves cannot *also* show people the importance of improvement (well, some people - others, sadly, don't learn shy of drastic consequences). I'm saying that character death can *also* be an educational experience.

On the other hand, I was in an L5R campaign where we were walking along the road, got ambushed with no warning or real opportunity to adjust, and got three arrows in the heart and character death during the surprise round in session 1. I definitely detached a bit from that campaign. It wasn't railroaded, it wasn't what the GM intended (in fact he even ended up retconning it since it was SO different than what he had understood about the combat system), but it also was clearly not something that any sort of reasonable attempt to engage with the premise or game could have done anything about. Since it was essentially random death, not due to a mistake or lack of reasonable preparation, it communicated 'what you do doesn't matter'.

Well, yes, "high fatality" ("realistic") systems, where you *can't* really learn much to reduce death (beyond "don't play this game", figuratively or literally) tend to, you know, not result in much learning. However, lower-fatality systems, where you can actually make bad choices and die vs make good choices and live, where you can see "this is what I did wrong" (or, at least, "this level of bad luck is unrecoverable, and i need *and can have* a viable backup plan / exit strategy"), OTOH, can produce postmortem epiphanies.

The real art to using death is to make it 100% clear when there are things such that 'if you mess this up, you will die' but do it in such a way that the player is actually able to understand that, respond appropriately, and as a result of reacting to the world correctly, have their character survive. Seeing caution rewarded with survival is more effective than seeing death happen for a reason that may be hard to pin down or explain beyond 'the dice were unlucky tonight' or whatever.

… I'm still confused. If, say, it's "three strikes and you're out", why would getting two strikes be more educational and easier to understand than getting three strikes?

IMO, only if there's always a clear chain of causality leading to that moment. A good example of this might be something like the Vrock dance. You have a bit of warning and you're informed 'if you don't deal with this, you're probably going to die'. So if you deal with it or fail to deal with it, you can explain why things went that way. If I take that to Save or Die mechanics, then I'd say that Save or Die abilities should tend to occur on very iconic antagonists which leave some kind of hint or suggestion before you actually encounter them. The classic example is seeing a bunch of statues in a dungeon and then, hey, basilisk! If you get petrified in round 1, you can at least look back and say 'I should have known'.

But if you just had some cultist who happened to have a Flesh to Stone wand and was mixed in with the other cultists, and now suddenly you're dealing with a Save or Die, then I don't think that's a good usage of what that kind of mechanic is for.

My players tend towards the "extreme luck" of somehow targeting any such "hidden threats".

To flip this notion, though - is it fair for PC threat level to be hidden of NPCs / monsters are required to broadcast their threat? I ask because I don't play favorites, and allow both sides to play Intel games (if they want - most (on both sides) don't bother). I *allow* 5d Chess, but also allow (and usually see) much more straightforward approaches.

I do think that if characters are delving into monster and trap-infested holes in the ground (aka dungeoncrawling), they are doing so in full knowledge that they are taking their lives in their hands, at any rate.

Lol. You'd think that, wouldn't you?

[NichG]

… I'm still confused. If, say, it's "three strikes and you're out", why would getting two strikes be more educational and easier to understand than getting three strikes?

I think this comes down to autopilot and 'standard operating procedure' ways of playing. If I've made some assumptions about how the game should be played given my character (such as, when playing the Barbarian I should Charge + Pounce + Leap Attack + Power Attack at -4), and I'm just playing that out, then when something goes wrong there are a variety of explanations and reactions:

- This is just a harder opponent/scenario than the previous ones but I'm already doing the optimal thing, so if I die that's just the statistics
- The GM is actively countering me or trying to screw me over
- Someone else messed up, it's their fault!
- I made a mistake but it was a mistake of detail - I should have delayed half a round, or I should have buffed first with a consumable, but my overall strategy is still the optimal one
- I made a serious mistake and I should have taken out my bow and fought from range, or otherwise avoided letting the enemy get a full attack, even if that reduces my damage output by 90%
- Whatever, death happens, that's part of the game. Time to roll up Bob the Second, I'm not going to try to explain it.

If I do my SOP and die, depending on how I'm feeling or what else is going on, I could pick on any of these justifications for better or worse.

Now, lets say we have the same scenario, but before rushing in I see the monster make a full attack against some NPC and get 12 attacks in the sequence, each dealing about 30 damage on average, and with to-hits in the +20 to +30 range (where my AC is 25).

If in response to seeing that, I decide 'I don't want to be in full attack range' then I've sort of already played out the previous scenario of applying SOP in my head and seen the consequences. But now, rather than half of the explanations either blaming dice or someone else for it, because I reached the conclusion that I should change my behavior by myself then 100% of my explanations will have it that it's my own decision of how to react to the situation that matters.

So if you can get someone to alter their behavior in the moment before the consequences hit in order to avoid them, I think that's much more of a teaching moment than if they rush through, suffer consequences, and then have to do a post-mortem to figure out what factors were responsible for the outcome.

If my goal is teaching smart play, places where the players succeed by being slightly smarter than they were before are a lot more valuable to that goal than places where players fail because they're insufficiently smart.

(There's a machine learning rant here to be made, but I guess I'll leave it)

To flip this notion, though - is it fair for PC threat level to be hidden of NPCs / monsters are required to broadcast their threat? I ask because I don't play favorites, and allow both sides to play Intel games (if they want - most (on both sides) don't bother). I *allow* 5d Chess, but also allow (and usually see) much more straightforward approaches.

I'm not really concerned with that kind of fairness as a matter of principle. I'll definitely have antagonists that play intel and mind games, 5d chess, and the like, but I broadcast that information as well - not the specific moves in the game of 5d chess, but the fact that the PCs are up against someone who operates that way. I'll also make sure that there are high level objectives on the board whenever that kind of antagonist is in play, which means that the PCs can get a nasty surprise when their enemy is smart and tricks them, but that surprise is that the enemy achieves something they wanted to prevent rather than the entire party waking up in Limbo after being killed in their sleep. A sort of fail-forward design.

If the opponent is an animal or magical creature, then actually broadcasting lethality is completely realistic. Poison dart frogs are brightly colored, because it means there's a chance a predator will be afraid and not eat them. If the frog gets eaten but kills the predator, the frog is still worse off from that exchange.

[HappyDaze]

Ask your players when they think CD is appropriate.

Personally I let players choose whether their characters live or die, instead of it being random chance.

Alternately, tell your players when CD will be appropriate in your game and ask them if they want to play it.

I don't let players directly choose whether their characters live or die, but they have lots of meaningful in-game choices that can greatly influence the odds.

For me, the game needs to stay a game, so I don't fudge anything.

[Tvtyrant]

Alternately, tell your players when CD will be appropriate in your game and ask them if they want to play it.

I don't let players directly choose whether their characters live or die, but they have lots of meaningful in-game choices that can greatly influence the odds.

For me, the game needs to stay a game, so I don't fudge anything.

Yeah but this is in the context of asking strangers on the Internet, and if you don't have strong feelings you are going to get a lot further asking your players then asking us.

Different strokes on that one. I prefer whatever will get the players to see their characters as people as much as possible, if that involves making them immune to death unless they feel their character arc has finished I'm for it.

[Kaptin Keen]

I have only ever killed characters when I had could no longer avoid it.

The worst example I have is from Shadowrun. A new player had just joined, he was new not only to SR, but to RPG's. In his first combat, he jumped out of cover onto the roof of a minivan - right in the line of sight of a drone armed with missiles. I warned him this was unwise, that cover was how smart heroes stayed alive. He insisted, then rolled atrociously when shooting at the drone.

I did not roll poorly for the drone's counter attack.

I even did the Rambo bit - with the smoking boots being all that was left behind. But he left, and never returned.

As a matter of fact, this was in 2000. 20 years ago. I ... don't think I've killed any player characters since.

[Tanarii]

Whenever you meet the game system requirements. Or the house ruled version of them, if your table doesn't like the default version.

[Jay R]

When the character reaches -10 hit points, or whatever the rules stipulate for death.

If the answer is anything other than what the rules state, then the players and DM need to agree to that in advance, just like any other house rules.

The more important question is how encounters should be designed, to give players the opportunity to save their PCs.

It's been a long time since a PC died while I was DMing, but it's possible, and the players know it's possible. That's why they retreat when they are losing.

The problems come when a group don't realize that death is possible, and therefore do not consider retreat as a sometimes necessary decision.

If a character dies in my game, I will consider it one of the following:

a. poor design on my part, that I couldn't fix on the fly,
b. really poor decisions by the players, or
c. an astoundingly bad series of rolls.

[If a single roll kills a PC, then that falls into category a, not category c.]

By the way, I'm speaking as somebody whose character died two weeks ago. Nobody in the party realized that the 2d6 hit points we were losing each round was a single, multi-round area effect spell, rather than a series of instantaneous attacks. So we never moved out of the area of effect. It was a Holy Storm against a 3rd-level party, and three out of five of us died. The bad decision on our part came from not understanding the situation. I don't think it was the DM's fault; we've been playing at a table for years, and are now using roll20. That format makes full communication a little harder, at least until we get used to it.

No problem; D&D goes on. Last week, our new characters avenged the old ones and are taking their bodies to a church for a proper burial.

[False God]

It is appropriate when it is appropriate.

[Drascin]

If you ask me when I feel a player character death is appropriate, as a GM... well, generally speaking, I feel that for a character death to be appropriate it has to involve one of two things: either the player knew perfectly well that there was a high risk of death and decided to go in anyway, or the player is actively okay and ready for the character to get straight up deaded.

Let me elaborate.

"The player knew perfectly well that there was a high risk of death and decided to go in anyway": What I mean by this is simple. There has to be at least one fully informed decision involved by the player if total character removal is to be on the table. A "random encounter" should never kill a player due to a string of improbably bad rolls (because the fun thing about being a player character is that you will make three orders of magnitude more rolls than any NPC, so law of averages says that you WILL hit the low probability disaster string at SOME point). Or if it turns out that players didn't pick up on your small clue that the vagabond was a disguised dragon and someone gets killed. Or there just was a random ass trap what kills you. Or you got into a fight you didn't expect to be this hard but retreating is damn near impossible (which, btw, is most RPG rulesets with tactical heft - I've noticed in most RPGs with an emphasis on tactical movement I've played, from D&D to Lancer, retreating is almost always way harder than engaging). And so on.

Basically, death by suckerpunch is almost never interesting. If death is going to happen, it should be in situations where players acted in full knowledge that there was a good chance of dying, but decided the matter was important enough to risk it anyway. And no, telling people that they are probably going to die in this fight after they have already engaged doesn't really count as informed decision. When in doubt, it's probably safer to lean away from killing people than not.

Similarly, sometimes players are actively alright with characters dying. Perhaps it would be a good cap to their character arc to pull a Skurge and die heroically holding the bridge. Perhaps they feel it would be hilarious if their death seeker warrior died in the most anticlimactic fashion possible. Who knows! There's as many reasons for character death to be a good idea as there are possible characters. But again, it's a player's decision. I as GM don't get to just go "oh, I think Character A dying would be really neat and make for a good story". It has to be player prompted.

[Democratus]

Character death is appropriate when something happens that kills them.

The rules of whichever game you are playing will explain how death occurs.

Other than that, you must lean on the social contract of the game table.

[GreatWyrmGold]

The real art to using death is to make it 100% clear when there are things such that 'if you mess this up, you will die' but do it in such a way that the player is actually able to understand that, respond appropriately, and as a result of reacting to the world correctly, have their character survive. Seeing caution rewarded with survival is more effective than seeing death happen for a reason that may be hard to pin down or explain beyond 'the dice were unlucky tonight' or whatever.

This argument isn't wrong, but it's supporting the wrong point.

Having your character die for unclear reasons doesn't teach anything—that is 100% true. But the problem isn't the death; having your character survive for unclear reasons isn't any better. The problem is the unclear reasons. Dying for clear reasons is an excellent way to teach people!

Character death is appropriate when something happens that kills them.

The rules of whichever game you are playing will explain how death occurs.

Other than that, you must lean on the social contract of the game table.

No offense is intended, but this is the worst answer in the thread, even worse than the tautologies. You simultaneously say that death should happen if and only if the rules say the characters die, and that the social contract of the game table determines when characters should die. These two answers aren't inherently contradictory, but they are unrelated.
It's not quite as bad as simultaneously asserting deontological and consequentialist morality, but it's close. Maybe it's more akin to simultaneously asserting deontological and subjective morality?

[Tanarii]

No offense is intended, but this is the worst answer in the thread, even worse than the tautologies. You simultaneously say that death should happen if and only if the rules say the characters die, and that the social contract of the game table determines when characters should die. These two answers aren't inherently contradictory, but they are unrelated.
It's not quite as bad as simultaneously asserting deontological and consequentialist morality, but it's close. Maybe it's more akin to simultaneously asserting deontological and subjective morality?

No offense intended, but amazing. Every word of what you just wrote was wrong.

The rules define death, and when those conditions are met. Unless you house rule them.

Other than that, the social contract at the table defines when you'll avoid or apply those rules. Things like "no gotcha insteadeath traps should be in the game" or "wandering encounters should never be difficult enough it's possible to result in death from them" or even "all encounters should be planned so a wandering encounter can't result in death" are all social contracts in how to apply the rules.

The social contract is about when and how to apply the rules. Or if you should modify them. Or cheat in certain circumstances.

[NichG]

This argument isn't wrong, but it's supporting the wrong point.

Having your character die for unclear reasons doesn't teach anything—that is 100% true. But the problem isn't the death; having your character survive for unclear reasons isn't any better. The problem is the unclear reasons. Dying for clear reasons is an excellent way to teach people!

I'd still say any kind of retrospective analysis will be less effective than if the player is able to see those reasons why their character might die just far enough in advance of making their decision that they can change course, and then feels compelled to actually change course in advance of the event.

Or maybe to put a rhetorical spin on it, if you think the reasons why a certain course of action would lead to death are clear, but the player still follows that course of action and the character then dies, it's likely you've overestimated the clarity of those reasons.

[GreatWyrmGold]

No offense intended, but amazing. Every word of what you just wrote was wrong.

The rules define death, and when those conditions are met. Unless you house rule them.

Other than that, the social contract at the table defines when you'll avoid or apply those rules. Things like "no gotcha insteadeath traps should be in the game" or "wandering encounters should never be difficult enough it's possible to result in death from them" or even "all encounters should be planned so a wandering encounter can't result in death" are all social contracts in how to apply the rules.

The social contract is about when and how to apply the rules. Or if you should modify them. Or cheat in certain circumstances.

This is not the same argument as you made the first time. This one clearly establishes the rules as a subset of the social contract, rather than treating them as independent entities. Which is true, but not obvious from "The rules of whichever game you are playing will explain how death occurs. Other than that, you must lean on the social contract of the game table." After all, the "other than that" implies that the social contract and the rules of the game are separate entities, which is a common way of understanding them.

I'd still say any kind of retrospective analysis will be less effective than if the player is able to see those reasons why their character might die just far enough in advance of making their decision that they can change course, and then feels compelled to actually change course in advance of the event.

Or maybe to put a rhetorical spin on it, if you think the reasons why a certain course of action would lead to death are clear, but the player still follows that course of action and the character then dies, it's likely you've overestimated the clarity of those reasons.

You're basically arguing that failure is an inherently ineffective teaching method, because if it was an effective teaching method they would know before you taught them. Could you try rephrasing what you intended to mean?

[Democratus]

This is not the same argument as you made the first time. This one clearly establishes the rules as a subset of the social contract, rather than treating them as independent entities. Which is true, but not obvious from "The rules of whichever game you are playing will explain how death occurs. Other than that, you must lean on the social contract of the game table." After all, the "other than that" implies that the social contract and the rules of the game are separate entities, which is a common way of understanding them.

They are two separate things that must both be taken into account.

When a character reaches a state where the rules say they are dead, death occurs. However, if the social contract at the table precludes the particular kind of death then it doesn't happen.

An example would be when I'm running a game for children. There's a different social contract at a table with young kids than with adults. And the kinds of character deaths (if any) that are allowed is constrained by that contract.

[NichG]

You're basically arguing that failure is an inherently ineffective teaching method, because if it was an effective teaching method they would know before you taught them. Could you try rephrasing what you intended to mean?

I mean, I wouldn't disagree with the part of the statement 'failure is an inherently ineffective teaching method'. If you're stuck in the mode of teaching someone who has absolutely no clue how to be successful at all on their own, asking them to try random things and punishing them every time something doesn't work is never going to teach them anything at a reasonable rate. At best you'll teach them how to have a thick skin and be persistent, at worst you'll make them not care.

The way you get out of that kind of situation is by providing guidance rather than by providing feedback. Feedback works when there's a clear sign of some attempts being better than others and where the credit assignment problem of 'why was this better than that?' has easy solutions - that means dense feedback rather than sparse, and as little randomness and hidden information as possible. Tabletop games aren't a good setting for reinforcement because consequences can be quite delayed (all the way back to character building choices in session 0), dice add randomness, and the DM is a big pool of hidden information waiting to impact the game.

So instead of feedback, you provide guidance. Guidance can be demonstrations of positive behaviors that should be performed (an example of this is if there's a mechanic in the system your players are ignoring, you can have an NPC use it effectively), or by using some sort of auxiliary partial feedback that's already familiar and which kicks in before the extreme sparse consequence does (e.g. before they random walk themselves off a cliff, you have a 'Are you sure you want to do that? You'll fall and die.' prompt). If there's something that they know would have been a failure and, via prompting or guidance or whatever, it is turned into success because of a single change in planned behavior, then that immediately solves the credit assignment problem and makes it clear at least in that one instance what sorts of actions are reasonable and what sorts of actions are not. There's still an abstraction that needs to occur - they need to recognize that same kind of action or strategy in new situations - but it's much, much easier than the general credit assignment problem.

[GreatWyrmGold]

They are two separate things that must both be taken into account.

When a character reaches a state where the rules say they are dead, death occurs. However, if the social contract at the table precludes the particular kind of death then it doesn't happen.

An example would be when I'm running a game for children. There's a different social contract at a table with young kids than with adults. And the kinds of character deaths (if any) that are allowed is constrained by that contract.

That is also not the argument I was originally criticizing. It is different than that one, and both it and your second attempt actually solve the problem I was criticizing. Why are you trying to argue with me? I don't disagree with you anymore, unless you're trying to claim your current iteration is logically identical to what you initially typed.

I mean, I wouldn't disagree with the part of the statement 'failure is an inherently ineffective teaching method'.

Naturally. That's your conclusion. I was not disputing that it was your conclusion. I'm curious why you felt the need to say that you agree with your own conclusion.

If you're stuck in the mode of teaching someone who has absolutely no clue how to be successful at all on their own, asking them to try random things and punishing them every time something doesn't work is never going to teach them anything at a reasonable rate. At best you'll teach them how to have a thick skin and be persistent, at worst you'll make them not care.

This is demonstrably incorrect, in everything from lab rats to Dark Souls players. You'll need to do better than asserting that learning from failure is impossible.

The way you get out of that kind of situation is by providing guidance rather than by providing feedback. Feedback works when there's a clear sign of some attempts being better than others and where the credit assignment problem of 'why was this better than that?' has easy solutions - that means dense feedback rather than sparse, and as little randomness and hidden information as possible.

I agree with everything here except the first sentence. Guidance and feedback are not separate; the only fundamental difference is that guidance can come before action, instead of after.

Tabletop games aren't a good setting for reinforcement because consequences can be quite delayed (all the way back to character building choices in session 0), dice add randomness, and the DM is a big pool of hidden information waiting to impact the game.

Then why does learning by failure work in video games? Consequences can be delayed (all the way back to character-building choices before the first cutscene, in some cases); randomness and de facto randomness are present in more games than not; and no DM is such a deep well of hidden information as a video game with intricate plot and mechanics.

Yet, from puzzlers to platformers, from stealth games to shooters, from Fire Emblem to XCOM, and especially with basically every game citing Dark Souls as an inspiration, learning by failure works even when they fall short of your criteria of clarity.

Obviously, not all video games pull this off. From Sierra adventure games to the clumsier Soulsborne wannabes, there's no shortage of video games where failure is just failure. But these games make it clear that the trait most required to learn from failure isn't immediate consequences, absolute determinism, or even relevant information being obvious. The trait most required is predictability.

"Predictability" doesn't mean "an absence of randomness," though extreme randomness can certainly detract from it. It means that a player can look at a situation, draw on their past experiences, and predict what is likely to happen if they do X. This can be on a vague macro scale ("if I let any archers get in range of my pegasus knight, she'll probably die"), on an instant-to-instant scale ("if I don't dodge-roll right now, that big attack will hit me"), or anything in between.

Obviously, there is some overlap between predictability and determinism, as well as between predictability and transparency. But you don't need perfect determinism or transparency to achieve predictability. The player doesn't need to know exactly how much cover improves their soldiers' odds of not getting blasted into alien chum, nor does there need to be 100% certainty that getting hit will always kill them. Just knowing that a single shot takes off most to all of a soldier's hit points at this level, and that moving into cover greatly reduces the chance that the single shot will hit, is enough for them to learn.

The obvious response is some form of "That's video games, those are different". If you want to make that argument, I'd like to ask ahead of time that you explain how. What makes learning through failure viable for Baldur's Gate but not Descent into Avernus?


The last paragraph of your post is just an explanation of what you think should be done instead of learning through failure. I don't think there's anything wrong with those methods, I just take issue with the assertion that you need to do any of that because failure is somehow incapable of teaching people anything.