And from what I've seen, the spell only physically disintegrates you if you actually lose all your hit points to it -- otherwise, it's just abstract "ouch" damage, like if you get hit by an inflict wounds or harm spell.
Well, fair enough. I can just about concede this one. Good argument
I don't know about you, but I'm changing all death spells (and some others) do deal Con damage instead of instadeath. 4d6 Con damage (say) should make most things sit up and take notice.
Well, to be fair hong, what you do in your house rules doesn't affect me nor the core rules. If you HR that death spells do Con damage, then more power to you. Just as an aside, I do see the problems that for a large number of creatures, 4d6 Con damage is virtually death anyway. The argument that disintegrate did Con damage sounds a good one, and if all SoDs did Con damage, that would be conceptually coherent. However, it just struck me as weird that they nerfed disintegrate without the other save-or-dies. Surely if one is broken, then they all are?
There's nothing special about the construct type that makes it immune to magic. Check the retriever for instance.
Not per se. However, the multitude of core constructs at high level are golems. The retriever is the only double-digit CR construct not immune to magic- and it only scrapes on at CR10. Given that disintegrate is available initially to an 11th level party, they will not see retrievers for long.
Poppycock. None of the iconic undead monsters -- vampires, liches, death knights -- are incorporeal
What happened to ghosts, advanced spectres and the like? In any case, liches usually have good touch ACs with their spells and vampires tend to be reasonably respectable in this department.
Is this more postmodern statistics I see before me?
Not this time

. Even I can do a straight damage analysis, even if I once screwed up a % chance to kill. 50/50 chance to make saves in what WotC generally recommends, and if you observe the archetypal DMG characters and put them against each other, then 50/50 is probably roughly what it will turn out at. Of course, the skew would be campaign-dependent- in a campaign chock full of paladins, then this is going to fall.