Xylix's problem is he's useing the wrond definition of 'death.' He keeps referring to
www.dictionary.com for some reason.
D&D has a special
operational definition (also known as jargon) for 'death effect.' A 'death effect' is
not any effect which causes the target to die, in spite of the fact that common sense (i.e., logic) would dictate otherwise.
That is why Xylix's logic fails him. His logic argument is just fine, but his
premises are flawed.
In D&D, a 'death effect' is any effect (meaning spell or magical effect) with the [Death] descriptor ascribed to it.
That's it. By this definition of 'death effect,'
phantasmal killer is not a 'death effect'. So while
phantasmal killer fits
a definition for 'death effect,' it doesn't use the
correct definition for 'death effect.'
You can make the
opinionated argument that spell X or effect Y
should be a 'death effect,' but that would not make a logically deductive argument (which is what you're trying to say you've got).
Oh, and saying "the PH might be misprinted!" is a really bad argument. It is
equally likely that: a) the [Death] descriptor was placed on spells it shouldn't be on, b)
death ward is in error, c) a more explicit definition of 'death effect' was mistakenly left out, etc. We
must assume that the PH/DMG/et al are
correct until we receive
direct evidence in the form of errata that they are not. Suggesting that errors in the PH/DMG/et al might be the cause is a logical fallacy in itself: appeal to fear or emotion.