The interpretation and presentation of data is often just as important as the data itself. If we're going to get anywhere with this, the statistics need to be presented clearly and without bias.
I don't have much time this weekend, so I will be brief. I don't think the basic idea of IP concerns are all that much in dispute. At least not as far as I'm aware. There might be a big question as to the probabilities in question: i.e., how frequently you are likely to suffer a catastrophic event and how frequently you are subject to them. And, if those probabilities were low enough, then this might get filed under "why care?" b/c it's so infrequent over the course of an ordinary campaign.
At the expense of stating the obvious given the thread title, the reason I started this thread was the implications of this concept. And, b/c it has cropped up in numerous threads, lurking behind the arguments so that around page 12 or so the reader finds that is the real idea at play. It's kind of like the whole magic v. mundanes thing in that regard. In particular, it seems to be used as a bludgeon against a whole swath of builds, classes, and archetypes, which given the materials available in 3.5 D&D seems surprising to me. So, creating a separate space for such a discussion seemed useful.
I'm strapped for time, but in brief here's my thoughts on the implications:
IP-proofing is a good idea -- adventuring is a hard and dangerous line of work. Assuming the relevant probabilities aren't too low, and my gut tells me they scale upwards as level does, you're going to suffer enough catastrophic events that you're likely to fail ... eventually.
IP-proofing ain't that hard to get* -- much of it is actually de rigeur in D&D. Getting good saves, a good AC, and so on helps lower those probabilities, making the instance of catastrophic failure less frequent. However, it's still worth investing in a few magic items, spells, or class abilities that give you some rerolls, avoid attacks, or other "get out of jail free" cards. And, it's nice for everyone at the table if you do -- the DM can go at you harder knowing you can take it, and everyone doesn't have to worry so much about each other.
IP-proofing tends to be talked about in too limited a fashion, to wit, it can be done ex ante, ex post, on a temporary basis, or by another character* -- this is just a personal impression, but usually when IP-proofing comes up in another thread the implication is that once you've failed that critical save, etc. it's "game over." So, the only way to make it in D&D is, by implication, to never fail that critical save, etc. While it may generally be the case that ex ante forms of IP-proofing, i.e., those that avoid the hazard entirely like a reroll for the dreaded save, are more effective, there are still ex post ones. A delay death may be just as good in some circumstances, especially if the fight is near over, and the same can be said for the humble remove paralysis. You're expending resources one way or the other: whether you get a wand of greater mirror image or delay death it's still committing resources to the issue. Likewise, the 8000 gp on an Amulet of Second Chances needs to be weighed against the 1000 gp cost of Revivify. If the probabilities are fairly low, then 1k every time you fail catastrophically ends up being more efficient.
Perhaps more important, though, is that in the usual course of discussions troupe play seems to have been lost. IP-proofing can be the responsibility of some subset of the party. I've long contended that the healer's real job in D&D is usually condition removal. Sometimes it's damage mitigation -- although in-combat healing is usually not a good idea, it's sometimes handy or necessary -- but really it's condition removal to keep the party in fighting strength against the terrible things enemies can do. This may end up being more efficient in the long run, or more fun, especially since some people enjoy the "medic!" role.
House rules to make the IP issue either more transparent or more easily dealt with for a variety of character classes -- I think most of us are well beyond playing vanilla 3.5 D&D. I know I am, and each gaming group develops their own corpus of house rules over time. There are house rules that would allow a wider variety of characters to better cope with IP-proofing, since nowadays like everything else it's typically the so-called Tier 1 and Tier 2 spellcasters that manage it best, if only b/c there are more spells in D&D than there are anything else. This would also make the IP thing more obvious and transparent, as well as sparing people the effort of diving through books finding the right magic item for it.
Such house rules could include changing the material components on spells (my gaming group has not used material components for years now, and doing so make revivify much handier and so on. Other options include Trailblazer's action points suggestion or Mutants & Masterminds' hero points, both of which are very similar.
*I expect these to be controversial.
