if the DM isn't communicating with his players well, is there any set of rules that will save them?
Can I also point out that its odd to assume that its ok for the DM not to play certain monsters to the fullest (ie, that the BBEG wouldn't capture a bodak or basilisk to drop on intruders, or hire a medusa as guard but not give hints of such things so that adventurers would find out) but that its inconceivable to move such meta-rules out into the open with the no deaths rule.
As Malraux points out, there are at least two ways of communicating: ingame, and metagame. Metagame is typically easier. It's no surprise that 4e - a game expressly designed to reduce the preparation burden on GMs - has opted for metagame communication of risks.
From the point of view of the protagonists, there is a potential for lasting harm. From the point of view of the reader, suspension of disbelief includes a willingness to believe that there is a potential for lasting harm. Knowing that there can be no lasting harm when you are playing the part of the protagonist, though, steps outside of that point of view.
<snip>
If the player chooses when the character dies, there is no risk of death. There may be death, but no risk associated with it.
The second paragraph runs together risk in the real world with risk in the gameworld. It goes without saying that in "death flag" play the players do not run these things together. That is, they are happy to "step outside of the point of view" of their PC. "Death flag" play is metagame-heavy play.
I'd say for people who prefer death-lite game, this "break" of POV offers no serious impediment to enjoying the game. The game is full of activities that pull you out of your character (start with "rolling dice"). They are part of the game.
Exactly.
The audience doesn't, however, want to believe that Bond knows that he is in no real danger.
And likewise the players don't want to believe that the PCs know they are in no real danger. Luckily, the players can easily bring about this result, because they get to determine (through the standard means that roleplayers use) what the PCs do and don't believe. And they can choose not to have their PCs break the fourth wall.
It is inherently far, far easier to suspend your disbelief about potential risk to your character when your character actually faces potential risk.
It is inherently far, far easier to suspend your disbelief about potential risk to your character when your character actually faces potential risk. That isn't a nonsense statement. That is tautological fact.
And death flag mechanics don't remove the tautology - the PC faces risks in the gameworld, but the mechanics of the game mean that those risks are never realised unless a certain metagame constraint is satisfied (ie the player raises the death flag).
But I assume that you really means something like "It is far, far easier to suspend your disbelief about potential risk to your character when the risk that your character faces in the gameworld is actually modelled by some randomness in the real world." And this is an empirical claim, not a tautology. What evidence supports it? As far as I know a lot of people are playing with death flag and similar mechanics and do not find it hard to play their PCs without breaking the fourth wall. It is no different from any other time that a player knows something that his/her PC does not.
Here's another question: Is there anyone here who thinks that, if the PCs (in a non-supers game) strip naked and bathe in lava, they should survive the experience unless the players decide otherwise?
I don't know. But in the typical "death flag" game it wouldn't come up, just the same as in the typical 1st ed AD&D game the player of the high-level fighter won't deliberately flaunt the wackiness of the hit point rules by having his/her PC take head-first dives of 100' cliffs for fun.
"Death flag" mechanics, like any other set of RPG mechanics, presuppose that they are being used for a certain purpose and that the players won't set out to break them by turning them to another purpose for which they don't work.
There is, IMHO, no difference between preferring survival-guaranteed and preferring no-paralysis-guaranteed.
This is an entirely empirical matter. It depends upon such considerations as what motivates players to enjoy "death flag" mechanics and what options exist in the game for a player whose PC is paralysed. I don't find it very hard to imagine a death flag game in which paralysis of PCs is possible. To work well, however, it would probably have to be the case that the paralysis was only temporary
in real life, which would mean either that it lasts only briefly in the gameworld also, or else that there is some way of accelerating the gameworld to the point at which the paralysis is lifted.
If the real question is "Does thinking about the death flag make us wonder whether other deprotagonising mechanics should also be dropped from the game?" then the answer is Yes. But paralysis need not be a deprotagonising mechanic, depending how it is handled.