Makes sense, as I don't play 4e, but that doesn't change anything: a quick unavoidable death is and remains far preferable to a slow unavoidable death.
Um, I think the whole point to having multiple saving throws is to ensure that death
isn't unavoidable.
And I think that's the other aspect of the SoD issue that has been raised, but not really explored yet (at least, not in this thread): the number of opportunities for action between the time the player realizes that his character might die and the time the character actually dies.
Of course, this is not really an issue if you take the perspective that the whole point of a saving throw (and IIRC, the original idea behind a "saving" throw in the first place) is that the character
should be dead: he's been hit by a death spell, he's looked at a medusa, he's gotten a faceful of deadly poison gas, etc. However, because he's a hero (or he has the potential to be one), he's got one chance to
not die. Go ahead. Roll your saving throw.
If you take this perspective, by the time a player is rolling a saving throw for his character, he would have already made enough mistakes and/or been unlucky enough that his character should be dead. You might have to deal with issues such as whether the players and DMs have different ideas about what is a "mistake", and how much bad luck and/or how many mistakes are "enough", but here, saving throws are always to the player's benefit because the alternative is simply "die, no save". Smart play consists of avoiding the need to make a saving throw in the first place, whether through research, planning, magical divination and/or protection, avoiding combat through negotiation or evasion, etc.
However, sometimes the DM simply wants to increase the threat of death in an encounter without making the fight arbitrarily lethal. IMO, straight SoD doesn't work as well in such scenarios because the player usually can't do much after his character has been targeted by a SoD effect except make a saving throw. The multiple saves before death model works better in such cases because the player will usually have a number of options he could potentially take: try to run away (if being away from the monster stops the effect), try to kill the monster as quickly as possible (if the monster's death stops the effect) or try to remove the effect (if he can).
Incidentally, I think that this chance to take action between realizing the increased likelihood of death and the actual death of the character (should it occur) is one key reason why some gamers who don't mind it (much

) when their characters die in normal combat still don't like SoD.