This post is a response to Mike Mearls's two items on save-or-die mechanics, here and here. I wrote a bit about the situation with save-or-die in high level play back in this post, but I'll expand on it a bit here. (I'll go ahead and spoil it for you: I don't think save-or-die is a good mechanic.) The comments to Mearls's first post raise the problem of whether save-or-die effects heighten or destroy tension during a fight. The former group feels that the situation is tense if players are anxiously waiting to see whether the dice decree that they live or die. The latter feels that because the opportunity for opponents to make save-or-die attacks is out of the player's control and there are no choices for the player to make after the perfidy of the dice has been revealed, there is no tension, only fatalism. Those players feel relieved if they survive the battle, but during the battle they feel there isn't much they can do to prolong their lives.
The dice determine the outcome of every battle, of course; it's technically possible for a 1HD goblin to defeat a 15th level fighter in a stand-up fight, if the fighter rolls nothing but 1s and the goblin rolls nothing but 20s. We're used to dice-luck determining outcomes, but we're also used to players making multiple decisions along the way, basing those decisions on varying estimations of risk.
Mearls talks a good bit about the right ways to use monsters with save-or-die effects. I have to wonder if he knows that this is one of those areas where DMGs have never offered comparable advice, and published modules have directly undermined that advice. Every high-level spellcaster in 3.x has save-or-die effects. Most poison-using creatures that I can recall in 2e that aren't drow have save-or-die effects - wyverns spring to mind. Fortunately the game mostly got rid of save-or-die poison effects in 3.x, but they left things with basically-passive save-or-die effects, such as bodaks and medusae.
The save or die effect represents an interesting point in D&D mechanics. On one hand, fighting a critter with a save or die attack is tense and exciting. Or at least, it can be. A good DM makes a fight like this into something that can grow into a gaming legend over the years. Players will remember how their characters valiantly fended off attacks and either hoped for lucky rolls or came up with a cunning plan to defeat or avoid the critter.
On the other hand, the save or die mechanic can be incredibly boring. With a few dice rolls, the evening could screech to a halt as the vagaries of luck wipe out the party. A save or die situation can also cause a cascade effect. Once the fighter drops, the rest of the party's inferior AC and saving throws can lead to a TPK.
I really like the save or die mechanic because, in my experience, most DMs know how to handle it well. They use it as a spice: something that can keep an adventure interesting or that can serve as a pitfall for foolhardy play. The mere appearance of a medusa or a giant spider changes the game, leaving even the most confident player nervous. Great triumphs require great adversity, and the threat of instant death is one of the game's toughest challenges.
The proliferation of save-or-die effects, though, means that lots of encounters will threaten instant death. I recall a 4e devblog relating a conversation in which the designers concluded that a combat system in which every character was a one-hit kill would be unfair because it made initiative too powerful. What I don't understand is how Mearls doesn't see this as a case of the same. Is it really that hard to imagine a character getting killed by a save-or-die effect before taking an action? "Before taking an action" here means that "foolhardy play" has not occurred. The only decisions a character can make without taking an action are ones that the character made without necessarily having any information on the encounter... or even that an encounter was in progress.
So, fine. Mearls does eventually come around to agreeing that save-or-die has to change in some fashion. His suggestion is that each effect carry a hit-point threshold; if the defender is above that threshold, the character automatically passes the save or the effect does something significantly reduced to the character. Mearls is pretty proud of his solution, but it is equally clear that opinions are divided here - he does, after all, go on to write a second post defending his proposal. The topic that actually concerns me in this post is the use cases he lays out as the strong points of his proposal.
Mearls believes that players should never have access to save-or-die effects, "short of the very highest levels." I see no need for this qualifier; I'm quite happy to connect the historic breakdown of D&D editions at high levels to the emergence of save-or-die effects in PC power sets. It's bizarre to me that he would then correctly identify the problem verbatim (too easy to game your spell DCs), despite having already stated a contradictory conclusion and having given directly contradictory examples in his earlier writing on the topic.
So, about hit point thresholds for save-or-die effects. Okay, this does solve the problem of "killed in the first round of combat"... maybe. Well, it helps fighters. I feel comfortable assuming that 5e will have the same general variation of hit point values we've seen in every edition: front-line warrior-types will have the most hit points, secondary front-line types (clerics and maybe rogues) will have a moderate number, and wizards will be squishy. I don't know how squishy, but if previous editions are any guide, it's not that unreasonable to suppose that a wizard at full health might be near or below the same number of hit points as a fighter at half health. So only the wizards (who probably also have poor Fortitude-type saves or defenses) have to worry about medusa petrification, bodak death gaze, or whatever else on the first round of combat?
See, hit points are this absolute rather than relative measure. Mearls looks at this as positive thing, in a world-building sense:
We can design monsters to model their power in the world. A medusa turns the town guards to stone, but the hero accompanying them has a fighting chance.
Right goal, wrong method. Don't focus player attention on hit point totals as the logic here. The fact that previous editions focused attention on saving throws/non-armor defenses was bad enough. "This guard? He would have been fine if he hadn't taken that one-hit-point scratch earlier. Such a shame about him."
Mearls's heart is in the right place here, for all that I'm castigating him in this post. He wants to set things up so that a monster has to screw a character over just a little more slowly - so that there's time for the player to experience dread and make choices that embrace or avert that risk. He wants these effects to eat away at hit points just like other attacks do. This is solid for two reasons: first, so that a successful save doesn't mean the attacker has made no progress, and second, so that a wizard and a fighter are both whittling away at the same defensive resource. Even though I'm on record as liking multiple damage tracks in games, it's generally best for a single type of conflict (e.g., violent conflict, as opposed to social conflict, mental strain, etc.) to put most of its focus on a single damage track, so that characters work in tandem rather than in parallel to take down a foe.
Hit point thresholds just aren't a good way to mark the progress of a slow-death effect, that's all. Not if you want some lethal effects to be themed around anything other than physical resilience - like a vampire's dominating gaze, for example, where wizards should be some of the best at resisting. Mearls, you work for the Wizards of the Coast. No one loves putting counters on shit quite like your buddies over in M:tG R&D do, though HeroClix and Warhammer Fantasy Roleplay 3e do at least get honorable mentions. You also personally worked on 4e, an edition of D&D that solved for changing the effect that something has on its target based on the target's condition. I mean, of course, Bloodied. So let's try this:
Finger of Death
If damage from any of the below cases reduces the target to 0 or fewer hit points, the target is dead rather than staggered or dying.
If the target is not bloodied: On a successful save, take 4d6 damage and gain a Death Effect counter. (This allows those poor red-shirted guardsmen to die horribly in the opening round.)
On a failed save, take 8d6 damage (here assuming 3.x hit point scales, roughly) and gain two Death Effect counters.
If the target is bloodied: Take a penalty to the saving throw equal to the number of Death Effect counters you currently have, and:
On a successful save, take 4d6 damage and gain two Death Effect counters.
On a failed save, you are dead (resurrection effects include the removal of all effect counters).
To this, I'd add support/utility effects that purge various kinds of effect counters, putting additional strain on the healers within the party. The structure of this effect does increase resolution complexity by one step, since the DM needs to follow the logic tree to figure out which subroutine to run, but "if bloodied" is a term that we had to get used to in 4e anyway, and it wasn't really a problem except in the case of AoEs that targeted a mixed group of bloodied and unbloodied characters.
Maybe it's my lingering memories of Fallen Earth design speaking, but I envision a variety of different counters that can build up on a character: death, poison, petrification, vitrification, ossification... you know what, maybe those last three can all be the same. Actually, that's not a bad idea: call them Transmutation counters and put them in the same group with baleful polymorph effects. Some healing spells purge all of a given type of counter, some purge a few counters from every pool - I dunno, this is all off-the-cuff.