GM vs PC - You're doing it wrong.

[Lmns Crn]

Of course failures should be interesting and one should be rooting for the PCs to succeed, but part of being a GM is creating a series of challenges.

Steerpike has saved me the trouble of writing five paragraphs in this space.

[Weave]

I'm with Kindling.  Part of the fun as a player for me is when I'm given a situation that seems challenging and really makes me think my way round a problem since the obvious or straightforward solution would be perilous.  Of course failures should be interesting and one should be rooting for the PCs to succeed, but part of being a GM is creating a series of challenges.

In some games of the more sandboxy variety, I'm also in favour of the GM not pulling punches in order to contribute to verisimilitude.  For example, right now I'm GMing a Planescape game.  If my 3rd level PCs decided, for some reason, that they wanted to steal a contract from the Tower of Arcanaloths (Arcanaloth = CR 16), I would have NPCs warn them that's a terrible idea and would give them several opportunities to back out or pursue alternatives, but if they persisted, I wouldn't go easy on them.

Oh, I would certainly do the same in your case. I would give them fair warning regarding the challenges they face, and if they continue to do so, then yes, they would die. Perhaps I should say, players will succeed in their own manner - they'll get what they want, but I'll do what I can to ensure they aren't recklessly marching to their deaths.

Does the failure count as interesting if its existence makes success interesting?
I'm really not a fan of "save or die" type situations, but I do like a degree of lethality and danger in my game. My players (I can't speak for anyone else) would not enjoy the game if their success was entirely guaranteed. Highly likely is fine, but the risk of failure, character death and other Bad Things happening has to be there to some degree to make the successes feel worthwhile.

That's a good question! I think so long as the failure isn't "save or die," then it's fine. The main point of what I was trying to get across was that failure shouldn't be the end all of the adventure. Maybe towards the end, in the final battle, when it's all or nothing, but for me, death shouldn't be the difference between success and failure. Multiple failures? Sure. One? Not in my book.

Bad Things happening are awesome. I'm in full support of Bad Things happening, but those Bad Things should be interesting and not veritable walls for the plot.

EDIT: Also, I don't want people to think I'm saying "don't challenge the players," because that's not what I mean to say. By all means, challenge them, throw everything you've got at them! I started this thread because a lot of people seemed to lean towards challenging them through character exploitations, as if they deserved punishment for building a powerful character. Challenges should be opportunities for growth, change, and heroism - I find many of my players work their best when they're at their (character's) worst. I want to make sure that the game isn't becoming something akin to PC vs. GM, but rather a collaborative building process.

[LordVreeg]

I'm with Kindling.  Part of the fun as a player for me is when I'm given a situation that seems challenging and really makes me think my way round a problem since the obvious or straightforward solution would be perilous.  Of course failures should be interesting and one should be rooting for the PCs to succeed, but part of being a GM is creating a series of challenges.

In some games of the more sandboxy variety, I'm also in favour of the GM not pulling punches in order to contribute to verisimilitude.  For example, right now I'm GMing a Planescape game.  If my 3rd level PCs decided, for some reason, that they wanted to steal a contract from the Tower of Arcanaloths (Arcanaloth = CR 16), I would have NPCs warn them that's a terrible idea and would give them several opportunities to back out or pursue alternatives, but if they persisted, I wouldn't go easy on them.

In a few recent threads in different sites, I have referenced the threat level and power level of a campaign as something that needs to be clearly communicated to the players.  Not everyone is motivcated by the same things, and even the same player and GM might want a lighter, less intensive game for a while.  And sometimes, the story created by the game is more silver-age comic book or more liughthearted or otherwise superheroic, if that is what everyone wants.

And it is the job of the GM to deliver the type of game and story that is advertised, but also play to what the level of the players. 

Your Sandbox comment is the nost important to me.  I have always said that EL/CR are dangerous ideas, as they can put GMs and designers in a situation where the EL is more important than the internal logic of the setting.  I made the same argument back in the elder days that random placement (big in Gygaxian days) also needed to be used with care as it also sometimes flew in the face of internal setting logic (versimilitude).
My Steel Isle players can attest to this in many areas.  I do try, as a GM, to always give extra avoidability for over powered encounters (but still, only for players playing very intelligently) , but internal setting logic is critical to immersion and to creating a satisying inside-out (played from the inside) narrative.


I recently said, "Smarter Players often come from deader characters".  I stand by it, humorous as it may be.  It may not work for every GM, but I tend to keep players for a long time, and often for multiple characters in the same setting.  I try to tell prospective players these days that Celtricia, at least, is a high-lethality, challenging game.  I try to evoke the feelings of pride and real accomplishment that existed in the earlier days of gaming, where most characters did not survive that long. 
And I feel like it does matter, even from a cooperative level.  Recently, both the SIG and IGBAR groups in Celtricia have started to move up as a group into a higher average experience bracket.  And I can feel the feeling of acomplishment in my players, and (speaking of the cooperative side) I share it. 

[sparkletwist]

I recently said, "Smarter Players often come from deader characters".  I stand by it, humorous as it may be.  It may not work for every GM, but I tend to keep players for a long time, and often for multiple characters in the same setting.

I think you need to be careful what you mean when you say "smarter." However much verisimilitude or detail or whatever a game has, at the end of the day, every GM is still just a subjective human being running a subjective game whose main goal is to be fun for all involved. A player doing something "not smart" might simply be doing something that the GM doesn't agree with, or doing something from a position of inferior knowledge about the GM's world or whatever. For that matter, the player might know more about the real-world basis for what he's doing than the GM and get it "wrong" in the GM's world due to the GM's lack of knowledge. I don't think there's any justifiable or reasonable reason to cut someone out of a shared narrative (which saying "you're dead" is always going to be doing, on some level) simply over what is fundamentally a disagreement between two people who both have a stake in that shared narrative.

[LordVreeg]

I recently said, "Smarter Players often come from deader characters".  I stand by it, humorous as it may be.  It may not work for every GM, but I tend to keep players for a long time, and often for multiple characters in the same setting.

I think you need to be careful what you mean when you say "smarter." However much verisimilitude or detail or whatever a game has, at the end of the day, every GM is still just a subjective human being running a subjective game whose main goal is to be fun for all involved. A player doing something "not smart" might simply be doing something that the GM doesn't agree with, or doing something from a position of inferior knowledge about the GM's world or whatever. For that matter, the player might know more about the real-world basis for what he's doing than the GM and get it "wrong" in the GM's world due to the GM's lack of knowledge. I don't think there's any justifiable or reasonable reason to cut someone out of a shared narrative (which saying "you're dead" is always going to be doing, on some level) simply over what is fundamentally a disagreement between two people who both have a stake in that shared narrative.

Yeah, I'm pretty careful, or at least at this point, damn sure about what I mean when I say 'smarter'. 
the quote comes from Roleplaying games where there are stakes and sometimes permanent effects.  And like many such things, these consequences can be teaching moments about the setting and the level of lethality.  And that is part of the game and risk/reward.  Shining the light of the OP on it, the GM should never be vindictive and should be trying to build the shared narrative with the Players, but in some game, it is the Job of the GM to play the world reacting to the actions of the players.  I rarely have had any disagreement from a player about their death in a very long history with a lot of gameplay.  And again, some of these players have stuck pretty exclusively with me for decades.

[Steerpike]

This thread is getting very interesting.  There seem to be several salient issues.

The main questions seem to be:

1) Should GMs use their control over the game to try and reward certain player character behaviors and punish others?  Should this control be predicated only on the internal logic of the setting?  Is it OK to reward or punish certain gaming styles?

2) Can/should character death in particular be used as a kind of teaching tool for players, to cultivate a certain style of play - i.e. more creative, more cautious, more adaptable, etc.?

3) How important is the possibility of death or other severely deleterious effects to the enjoyment of the game?  Is a sense of risk integral to the fun?

4) What are the major risks of running an especially lethal game?  Will a deadly game frustrate most players?  How should a GM strive to balance this risk/frustration dichotomy?

5) Are "save-or-die" effects and the like something of a hold-over or vestige from an earlier era of gaming, or can they have a valuable role in games?  Were they a "mistake" or are they sometimes still a useful tool?

6) Should the GM's role primarily be to simulate a world, to offer challenges for players to overcome, or to cooperate with players to build a story?  Are these roles ever at odds, or can they always be reconciled?

7) Should player success be assumed, or should the GM be "success-neutral"?  Is such a stance even possible?

8) How should failures be handled?  What are some strategies for making PC failure "part of the fun"?

I'd be interested in hearing how various thread contributors answer some of these!

[sparkletwist]

I rarely have had any disagreement from a player about their death in a very long history with a lot of gameplay.  And again, some of these players have stuck pretty exclusively with me for decades.

Well, if your players like it and agree with it, that's the real point. As I've said before, I'm not going to accuse anyone of "badwrongfun." I think your quote rubbed me the wrong way because all too often I've heard the statement about players "not being smart" to mean "playing in a way different than the GM wants/expects," which, of course, needs to be resolved by discussion at the table, not killing characters. I am personally not in favor of making players learn anything about my game or setting "the hard way," but that's just me. (Characters, naturally, are a different story!)

[LordVreeg]

I rarely have had any disagreement from a player about their death in a very long history with a lot of gameplay.  And again, some of these players have stuck pretty exclusively with me for decades.

Well, if your players like it and agree with it, that's the real point. As I've said before, I'm not going to accuse anyone of "badwrongfun." I think your quote rubbed me the wrong way because all too often I've heard the statement about players "not being smart" to mean "playing in a way different than the GM wants/expects," which, of course, needs to be resolved by discussion at the table, not killing characters. I am personally not in favor of making players learn anything about my game or setting "the hard way," but that's just me. (Characters, naturally, are a different story!)

I think that is the real point on this.  Some players would not enjoy this, or find the risk vs reward worthwhile.  I can actually say that the two times a PC really complained bitterky about their death being unfair were both times that I was covering for the fact other PCs killed those characters, but the player would not have known that....

[sparkletwist]

This thread is getting very interesting.  There seem to be several salient issues.

These are good questions! I'll try to answer them for myself. :grin:

1) As I've mentioned, I don't think "smart" should be used as a code word for "plays the way the GM wants." I feel there should be mutual agreement at the table among all those with a stake in the shared narrative as to how that narrative will proceed.

2) Only if the players have agreed. This goes with #1 and my previous comments.

3) To me, it's not that important. I think risk is important, but that can be the risk of any sort of setback. Failing the mission means the game gets to go on. Character death means the game doesn't go on, for that player. That's something I feel the player should have a hand in deciding.

4) I think it depends on the group. Asura is actually quite a lethal system, in the sense of "it's quite easy to lose a lot of HP." However, Asura has a specific, codified rule that only "important" characters can kill other "important" characters. I feel this is important to allow characters to "own" their character's lives and deaths.

5) I personally dislike them strongly. I don't feel a single roll in isolation should matter nearly that much. (This is far different from the single roll that determines whether you're taken out-- after you've already taken a lot of damage, and probably made a voluntary choice or two that increased your risk!)

6) I like to think that helping the players to build a story and simulating the world go hand in hand, because the world is an essential backdrop to the story and the story gives the world some direction. And, of course, in that world and story, there will be challenges, or it's not very interesting. So, I like to do them all, and feel the three tasks often enrich one another greatly.

7) I really do like it when players prevail in the end. I'm certainly willing and able to inflict setbacks on them, but that makes victory all the better.

A side anecdote: I will admit that the "Sixsura" game was at one point spiraling towards a bad end and I was mentally preparing myself to just grit my teeth and hand out the "Game over, you lose, Cthugha wins." It was a one-shot, though. I don't know how I'd feel in a campaign.

8) I must confess a certain fondness for slapstick hijinks, both as a player and as a GM; failure doesn't feel nearly as bad when everyone's laughing at the way it happened. Giving players some choice in how they fail often helps with this, as they can come up with something amusing and is a setback but doesn't hurt their concept of their character or the narrative. FATE has a mechanic often called a "self-compel" where you can say to the GM, "I want to fail at this because I think it'll enrich the story."

[Lmns Crn]

Fascinating questions, Steerpike. I think this thread can benefit a lot from that sort of distillation.

1a) Should GMs use their control over the game to try and reward certain player character behaviors and punish others?  1b) Should this control be predicated only on the internal logic of the setting?  1c) Is it OK to reward or punish certain gaming styles?

Forgive my slight renumbering, I have clarity in mind.

I consider 1a and 1c to be rewordings of the same question.

I think 1b is the key, here. I will absolutely, absolutely reward and punish certain styles when it is consistent with the game being played and the world the game takes place in. If I run some sort of Game of Thrones RPG and you want to play a pure-hearted, honorable, virtuous soul-- look out, buddy; your life is going to be rough. Because that's what the setting demands.

So this is not a blanket condemnation of a style choice ("When I GM, I hate it when people do XYZ, so I always make sure doing XYZ is fatal to a character"), but it's an acknowledgement that some types of play work less well within a specific context. And I think that's important to making the context seem real and relevant. (If I run a Game of Thrones RPG and people are being all valorous and honest and noble and coming out unscathed, pretty soon it doesn't feel like Game of Thrones at all anymore.)

Which is not to say I make characters die. Because that's removing someone from play, that's removing someone's ability to contribute to our group activity, that's pretty extreme. And I think "wrong move: u ded" is often abrupt and suspenseless.

But this gets back around to the role of the GM, which is tough for me to articulate. Lately I lean towards "not antagonism per se, just make the characters' lives difficult and occasionally miserable, and make it all but impossible for them to reconcile conflicting wants." But that's not antagonism, that's dissection, and we'll get to that in a bit I guess.

2) Can/should character death in particular be used as a kind of teaching tool for players, to cultivate a certain style of play - i.e. more creative, more cautious, more adaptable, etc.?

In general, I think this is not very much fun. There are exceptions, and they are the exceptions that alter the stakes of death, what death means. Eclipse Phase and Paranoia would be examples, I guess.

Tabletop RPG playing is a social group endeavor. If death means "your character is finished, start over from scratch," then a player whose character dies has a reduced scope through which to contribute to the ongoing effect. (Survivors' players can contribute through a "complete" character's story, inasmuch as they're ever actually "complete"; casualties' players contribute through fragmented and abruptly truncated segments, and may never get to touch on things they wanted a character to address.) This is not to say that PCs should never be allowed to die until their stories are concluded (I am loath to take just about anything completely off the table, especially something as potentially powerful and interesting as death); rather that it's a lousy anticlimax for a character with a promising arc and interesting options for future fates to be struck down as a fatal object lesson. ("Died as a way to demonstrate to his player the importance of checking every 5' x 5' square for traps.")

3) How important is the possibility of death or other severely deleterious effects to the enjoyment of the game?  Is a sense of risk integral to the fun?

I think there should always be the possibility of unforeseen, irrevocable changes to a character. As it happens, most things that fall into that category seem to be tragedies.  I don't think those things should necessarily remove that character from play. (Then you don't get to deal with the aftermath.)

Death isn't even the only thing that falls into this category. If you're running a period piece in feudal Japan where one character is a badass samurai who can do anything with his sword, and being a master swordsman is the player's source of enjoyment from that character, for god's sake don't let that PC end up with his hands cut off. Or if you ordain such a fate, don't pretend it's more lenient than PC death, because in many cases, it's not.

5) Are "save-or-die" effects and the like something of a hold-over or vestige from an earlier era of gaming, or can they have a valuable role in games?  Were they a "mistake" or are they sometimes still a useful tool?

So here's the thing. Whenever death is an option, it almost always comes down to one detail or one die roll. If you're a mighty warrior who has suffered many wounds and continues fighting anyway, you get to a point where every attack roll against you is a save-or-die, because if you get hit one more time by anything, you are wormfood.

The difference, and the reason one is acceptable and the other is (I think) unacceptable, is whether or not a player has an option to buy into the stakes of a scene and agree to them by action. If I'm playing a mighty warrior and I wade into battle, I've got to know that there's a chance I might die. By going into the fight anyway, I'm essentially acknowledging that risk. If I continue fighting even though I'm pierced with many wounds, I've got to know it's probable that I will die. Being wounded, and continuing to fight anyway, is a way of raising the stakes.

Choice made, risk accepted, death valiant.

The same character could, in some systems, get killed by a save-or-die effect before even realizing where the deadly attack was coming from, much less buying into the high stakes with a choice. Which is what I think makes them unsatisfying. In some situations it can make sense (you've waded into the heart of the necromancer's lair and now face the twisted sorcerer one-on-one; he raises clawed hands which crackle with anti-life magics...) although I think there are almost always better mechanical ways to represent this stuff.

6) Should the GM's role primarily be to simulate a world, to offer challenges for players to overcome, or to cooperate with players to build a story?  Are these roles ever at odds, or can they always be reconciled?

All three. Probably more than that. If they're at odds, you need to change one. (If I'm working really hard to simulate a world which is boring, and there's nothing that's drawing players into the story, then I probably need to change things about the world I'm representing to make it more... full of stuff to do.)

7) Should player success be assumed, or should the GM be "success-neutral"?  Is such a stance even possible?

In the context of this thread, I feel like this question might be asking about a false dichotomy, as if the only options are a.) succeed or b.) die. Which I think is missing out on a heck of a lot of things!

I think the real meat of the GM's role, much of the time, is to really put the fire to PCs, in order to see what they do. Be a fan of the players' character, but force them to make tough choices, because that's how you get to know them. Set up situations where they cannot have everything they want, no way, impossible-- then see what they prioritize.

Here's a character who a.) never lies, and b.) puts family first, no exceptions. Obviously put him in a situation where the only apparent way to protect a family member is by lying, then see which principle he sacrifices first. (Or if he's clever enough to find a third path that allows him to keep all his principles intact.

This does players the great and interesting service of allowing them to walk the walk. It's one thing to say "my character is honest." Unless that's really put to the test: ho-hum, so what? not interesting. But if this is the character from the above example who chose to let a brother die rather than lie to protect him, suddenly that honesty really means something, it's become the basis for a story players will remember, and it's become a real point of discussion (is he now "too honest"? Maybe!)

So I think that's a way of hurting a character that really sparks new kinds of player participation and reveals interesting things about the character. Character death is a way of hurting a character that stops them from participating and reveals... entrails.

If death is worked into this framework and becomes part of the stakes of an action through a weighing of priorities, though, that's powerful fuel. If a character dies because they legitimately wanted something so badly they were willing to risk death to get it, that's awesome whether they live or die.  

8) How should failures be handled?  What are some strategies for making PC failure "part of the fun"?

[Steerpike]

Fascinating questions, Steerpike. I think this thread can benefit a lot from that sort of distillation.

Thank you!

Some awesome answers... LC, I love what I'm going to call your "Jamie Lannister" style of character-development, where you put the PC in a kind of "something's-gotta-give" situation.

But that's not antagonism, that's dissection, and we'll get to that in a bit I guess.

Or vivisection  :P.

I think it depends on the group. Asura is actually quite a lethal system, in the sense of "it's quite easy to lose a lot of HP." However, Asura has a specific, codified rule that only "important" characters can kill other "important" characters. I feel this is important to allow characters to "own" their character's lives and deaths.

This is interesting.  It seems to imply that you're essentially "anti-death" except when it specifically enhances the drama of a given scene, i.e. when two important characters are fighting.  How do you handle things like traps and environmental hazards?  Is it impossible to be killed by them?

I like to think that helping the players to build a story and simulating the world go hand in hand, because the world is an essential backdrop to the story and the story gives the world some direction. And, of course, in that world and story, there will be challenges, or it's not very interesting. So, I like to do them all, and feel the three tasks often enrich one another greatly.

All three. Probably more than that. If they're at odds, you need to change one.

This is what I'm getting at with the idea of the simulation/challenge aspects of GMing sometimes potentially being at odds with the storytelling aspects of GMing (I'm not saying they always are at odds):

For example, let's say we're playing a gritty, realistic WWII game where a single bullet can spell death.  The GM here wants to evoke the feeling of grim, realistic war films like Saving Private Ryan and mini-series like Band of Brothers.  Accordingly if we build a rules-system with simulation and challenge in mind, then weapons like grenades, gatling guns, flamethrowers, and bombs are going to blow characters to shreds if they put themselves in harm's way, and they're going to have to be very, very careful about their tactics (use of cover, stealth, formation) to survive.  Done well, this makes the game exciting, immersive, and highly challenging.  Much of the fun comes from devising creative ways to avoid being turned to mincemeat, and the very-real possibility of death is essential to this atmosphere and playing style.

Now let's say that one character's major story arc involves saving an old Parisian sweetheart from the Nazi commander who plans to make her his bride.  Unfortunately, said character is wounded in the leg just outside the French capitol, and as a result of his reduced speed, he's unable to make it to a bunker in time to avoid being caught in a hail of German fire.  The dice say that the character has been obliterated at the hands of a nameless Nazi mook.

Now, here I'd be tempted to say "screw the dice," screw the simulation, screw the throwing-tough-challenges-at-players element of GMing and rule that the character is just knocked out and bleeding badly, giving his comrades a chance to patch him up and thus (hopefully) head towards the much-more-dramatic encounter with the German commander, which at this point isn't far off.  Or perhaps a bullet penetrated his brain and he's now slowly dying, but will last just long enough to rescue his lost love before perishing.  That's fine, but it entails a sacrifice of the other roles the GM plays - if we give in to the storytelling aspect, we give up an element of gritty realism.  Moreover, the players will probably be aware that the normally simulationist, gritty rules have been compromised in some fashion.  Now they might feel that they needn't be quite as careful since they can rely on the GM fudging the dice for the sake of story.  See the conundrum?

Now we could say that that type of story is too elaborate for this kind of game, but then haven't we just sacrificed the story element in favour of the simulation/challenge element?  Is there a way "out" of this situation without some kind of tradeoff, a privileging of one GM hat over another?

[Llum]

Man seems like a lot of people here are anti PC-death. Maybe because it's how I like my games (heavily simulationist) or how I cut my teeth on gaming (Guildschool) but I really don't see losing a PC as a bad thing. Sure it sucks, but what's stopping you from rolling another one? It isn't life, you aren't "out of the game" forever. It was a consequence to your choice of actions.

[Seraph]

I like to try to give an extra chance to avoid a PC death, but I will allow a PC to die if it comes to that.

[LordVreeg]

1) Should GMs use their control over the game to try and reward certain player character behaviors and punish others?  Should this control be predicated only on the internal logic of the setting?  Is it OK to reward or punish certain gaming styles?

I have almost never had a need to punish a gaming style.  The game has always done that for me, literally.   About 6 times in my history, the other PCs have knocked off the powergamer in a group for me.  And I do try to run games that are either too simple to really break, or too multidimensional.
And since the internal logic of the setting does not have to be compromised to reward and punish, it can be very synergistic in a good GM's hand.Part of this is merely just cherring along with the PCs who do smart and fun stuff.  And this should not be overlooked.  When the GM is excited about the way the PCs do something, that excitement passes to the players.

2) Can/should character death in particular be used as a kind of teaching tool for players, to cultivate a certain style of play - i.e. more creative, more cautious, more adaptable, etc.?

I think my answer is obvious, but I'll try to explain it a bit. 
First off, I run Old School games, and was brought into gaming in a time where part of the reason you played was the fact that a lot oc characters died, and that was OK.  Roll up a new one and try to learn from the experience.
Because there was a lot of pride in survival. Saying you had a fifth level dwarven fighter in a campaign where only 3 people had ever made it past 5th level meant something.  And I keep this in my current games. 
Now, not every setback or teaching moment involves death, sometimes there are a lot of positive reinforcers, and sometimes players cheat death and that is a big one as well.  But having to stop and roll up a character and rejoin later is a powerful reinforcer to behavior.

3) How important is the possibility of death or other severely deleterious effects to the enjoyment of the game?  Is a sense of risk integral to the fun?

As above, I was brought into this whole thing playing games where you lost characters regularly.  SO for me and what I was brought up in, I have seen that the risk is often tied to the reward.  Success means little when success is all but ensured.
Now, different people enjoy different things, and even the same person likes a break sometimes.  But I once had a very safe RPG game described to be about as much fun as playing monopoly and having the money and property re distrubuted every time around the board so no one could lose. 
But there are a lot of people who hate to lose, and they would rather deal with capping the feeling of success because the chance of losing is no fun at all.

4) What are the major risks of running an especially lethal game?  Will a deadly game frustrate most players?  How should a GM strive to balance this risk/frustration dichotomy?

The biggest issue here is being upfront about this and make sure that players know what they are getting into.  Will it frustrate most players?  DEpends what you mean.  There are levels.  Rarely is a player thrilled about losing a character, but frustrated enough to bail, or really to make them mad enough to avoid the game?  Steerpike and assembled...I have had 2 instances in the last 30 years of my adult phase of gaming (I am not counting the first 6) wher a Player has complained to me about how they died.  ANd both were situations where I was lying to thje player becasue their character would not know that the other PCs killed them, so they were unclear how they died.  and both of them camed back to the table anyways.  I have NEVER seen a player lose a character and not come back to my table.  I've had them  quit 'in medias', but never lose a character and then stop playing.

5) Are "save-or-die" effects and the like something of a hold-over or vestige from an earlier era of gaming, or can they have a valuable role in games?  Were they a "mistake" or are they sometimes still a useful tool?

This is a question and part of an analysis that went on in many boards for years, as 3e and 4e came out.  The answer to the first is both, the answer to the second is that they were never a mistake, but they were misused.
Save or die was part of the earlier games, where the game was balanced around exploration and then balanced around the campaign...not around combat.  There was no EL or such.  Part of this was just the fact the game was supposed to be more deadly.  I know enough people who lost characters to fireballs even after saving (Only half damage...great.). 
But part of the reason that petrification saves, or poison saves, or save or die stuff was kept in the earleir games was to encourage alternate, often non-combat or heavy thinking ways to defeat something.  Once EL and CR came around, and the game was becoming balanced around combat, things needed to change because the whole balancing mechanism was based around slugging it out with things in the appropriate class.  Balancing things around combat with the possibility of Save or Die makes little sense.
 
But running away was always part of the earlier game, or parlaying (there were rules for it), or trying to get the treasure without fighting (you got experience points for gold and for magic found back in the day), or finding a way around that save or die.  Because that was one of the reasons it was there.  It made smart players try to get around those real chances of serious issues despite their power level.
Now, if you played kick in the door style games back in the older days, you lost a lot of characters (this heakens back to the earlier question about death), so you save or die taufght you that no matter how tough your character was, they clould get killed, and as such, save or die taught caution and smart play.   

6) Should the GM's role primarily be to simulate a world, to offer challenges for players to overcome, or to cooperate with players to build a story?  Are these roles ever at odds, or can they always be reconciled?

they are not at odds, but they often have to be prioritized.  And while I have my own ideas how, it is based on the expectations that the players and GM set up in the beginning.
For my games, the world simulation is the one absolute.  But I read other people, and I see that proobably the shared narrative is the top priority.  Heck, in my games still, and especially back in the old dyas, PC cooperation was not a given at all.

7) Should player success be assumed, or should the GM be "success-neutral"?  Is such a stance even possible?

Steerpike, I;ve seen games where it is neutral, anti player, and proplayer.  Seen it all.  I personally don't think it is any fun if the sucess or even the survial is assumed, but that is me.
I take the personal pose of, 'Geez, I cheer for my players and I am here to answer anything and to help you play as smart and as well as you can...since I am playing the role of the rest of the world fairly and if you croak, you croak'

8) How should failures be handled?  What are some strategies for making PC failure "part of the fun"?

By having the dice ready to roll up their new character.  Or to tell them to have  a spare ready.
No, sort of kidding, but the way the party treats this has a lot to do with it.
And the GM should always be sorry and unhappy with PC bad consequences.  This is a game for everyone.
OK, enough for now.

[TheMeanestGuest]

As above, I was brought into this whole thing playing games where you lost characters regularly.  SO for me and what I was brought up in, I have seen that the risk is often tied to the reward.  Success means little when success is all but ensured.
Now, different people enjoy different things, and even the same person likes a break sometimes.  But I once had a very safe RPG game described to be about as much fun as playing monopoly and having the money and property re distrubuted every time around the board so no one could lose. 
But there are a lot of people who hate to lose, and they would rather deal with capping the feeling of success because the chance of losing is no fun at all.

There's a couple things you've said here that bother me, but I think it mostly boils down to equating character death with losing. Maybe more specifically the apparent assumption that a player who is upset by the death of their character is a sore loser, or that anyone who creates or plays in a game that isn't all about the hard reality of numbers is a communistic carebear. I don't particularly see the point of playing an RPG as having a successful character, but rather about the construction of an interesting story. My character could consistently suffer from undesirable outcomes, but that doesn't automatically make the story that is being built a bad story. I'm not saying characters shouldn't die. Character death is very important, and can be a powerful narrative element, but I don't think that it should just be a functionally random occurrence. The choices that a character has made and the tensions that they have created through their actions should feature, in my opinion, in any situation that has significant lethal potential.

I generally become very invested in my characters and enjoy developing them outside the bounds of the game itself, and thusly, if a character of mine were to die because a random no-name mook shot them in the eyeball I would not be upset because I 'lost', but because the work I put into that character wasn't at all valued or considered by the GM. I'm not saying that GMs should pander to their players, far from it, being given what you want all the time with little effort would be just as boring as having your character choke to death on their breakfast because they failed their chew roll. It's difficult to convey my exact meaning here, but I think I at least sort of intimated it.