I spoke the other day about 5e, and criticized their saving throw structure. I am not sure I did a good job in that post, so I will attempt to make my position clearer here, as well as to provide some alternatives. A lot of this is derived from Trailblazer's analysis of 3.5's saving throw structure, so I may drop into their parlance on occasion.
When I speak of saving throws diverging, what I mean is this: we divide saving throw categories into two classes, proficient ('good saves' in 3.x) and non-proficient ('bad saves' in 3.x). Each has a starting value and a rate of increase; in 3.x, your good saves started at +2 and gained another +1 per two levels of experience, while your bad saves started at +0 and gained +1 per three levels of experience. Despite increasing reliably, over the course of many levels your bad saves got weaker compared to your good saves and the save DCs you were up against, since save DCs also grew at about one point per two levels. The gap between your good saves and your bad saves goes from +2 at 1st to +6 at 20th, while spell save DCs also rose from 11 or so (we're disregarding ability scores here; more on that in a moment) to 19, an increase of eight points. Your good save odds have gone from 60% (9+) to 70% (7+ with your good saves), but bad saves've dropped from 50% (11+) to 40% (13+). That's not so bad, thought, right?
Well, when save-or-die is in the mix, it sort of is. And there are more aggravating factors too. Ability scores don't stay constant, and tend to diverge as well. Our hypothetical spellcasting adversary has, since first level, probably acquired +6 points of enhancement bonus, +2-4 points of increases from levelling, and possibly a few points of inherent bonuses from Wish or Tomes of Whatever. Let's be conservative here and assume we're still fighting without magic items; let's call it +4 points of self-buffed enhancement bonus and another two points from levelling, for +3 points of DC. Your good saves are typically correlated with your class' strong ability scores; as a result, you might've kept up with him there if you were a cleric or a rogue. But your bad saves are typically tied to your off-stats, which you likely did not invest resources in bringing up over the levels. So now you're looking at something like 16+, or 25% success rate, on your bad saves.
If your class happens to have two bad saves, you're even worse off - even if you do want to invest in ability score increases to try to close the gap, you have to split your resources between two scores while the enemy only needs to increase one stat to attack both of your weak points. Meanwhile, the Adversary's spell selection has only expanded, and he now has more options to target your off-saves than he did at 1st level.
The final straw is the rate of monster HD growth in 3.x. How many HD does a CR20 dragon have? Ranges from 28 for red up into the low 30s for most colors. And since supernatural and extraordinary ability save DCs grow at half HD, they're at +5 points of save DC above our friend the Adversarial Wizard. You're not making that breath weapon save, pal, especially since it's based off of Con, which on big dragons tends to be in the high 20s. Hell, even if you've naked-capped dex (18 natural +2 racial +4 inherent +4 cat's grace = 28, for a +9 modifier) and have reflex as a good save (+12), you're still probably not making the save (DC36 for a great white wyrm, so 15+ for you)... without magic gear.
This is where Trailblazer's work left off, with the conclusion that against high level foes, your odds with good saves were bad and your odds with bad saves were impossible in the absence of cloaks of resistance and save-ability-score boosters out the ass. Their proposed solution was action points. I happen to think that's a lousy, duct-tape-and-band-aids solution, but more on that next post. Ultimately I think their conclusions are sound - having different growth rates between save types is a dangerous position to be in. Even if your good saves grow as quickly as save DCs do, there's still a very real risk that your bad saves will fall (relatively) to the point where they're no defense at all, as your opponents will tend to specialize in support of their offense.
Anecdote: I once saw two-thirds of a 20th level party in 3.0 die outright to a banshee during the first round of combat. Saves at high levels are both serious business and very easy to fail.
To return to 5e, several of these conditions are met, if not as strongly as in 3.x. Proficient saves keep pace with save DC growth, at one point per four levels, while non-proficient saves do not grow at all. We start at 10+ with non-proficient saves at 1st level, and gradually fall to 14+. Fortunately, our opponent can only increase his casting stat to 20; I expect that will give us an extra point or two of DC over 20 levels. Meanwhile, we got some ability score increases over the levels, and subject to the same cap on our main stats, might've spread them out to cover our off-saves. Unfortunately, we have more bad saves than we used to; I don't know if anything calls for a Charisma save yet, but if it does, the Adversary will find it. Defending four off-saves with only six ability score increases, some of which we're going to want to put towards our main stats, is not a great place to be, especially given that something's going to be an off-stat in addition to an off-save. That 10 from the elite array is coming back to haunt you, while your opponent is gunning for it with a 20 Int. Let's say you boosted it to a 12, and he started with a 16 and boosted to a 20... In any case, now you're looking at an 18+ to save with that stat, up from 13+ at 1st with ability scores taken into account, and you're going to have a bad day against a wizard who 1) is prepared to target any stat (not happening at low levels, and since I don't have access to the full spell list I cannot assert that this will be possible for any given spellcaster in Core - when the supplements start dropping, though, all bets are off), and 2) has the ability to perceive that you are weak in that stat.
Even if they can't do that, your basic odds of making a non-proficient save are worse at 20th than they were at 1st, barring buffs or magic items or other trickery, and the stakes have (again) only gotten higher. Save-or-die may be off the table, but Dominate and Petrification and Disintegration are still around, and those can ruin your day.
Let us consider the monsters. The good news is that monster base ability DC grows with CR, which hopefully grows at about the same rate as your level, so you're not at as big a disadvantage as you might've been in 3.x. Unfortunately, monsters don't have the ability score caps you do, which means that they can still crank their save DCs up high enough that you are very unlikely to pass with a non-proficient save. Even though I don't have the whole monsters section, I feel fairly confident in asserting that monster ability scores will tend to rise with CR and level, and that (for example) dragonbreath will utterly outpace any attempts you might make at boosting your Dex save if it's not one you're proficient in. DCs for the sort of effects you really, really want to avoid will go up, your bonus won't change, and ultimately your ability to save against these effects will decline with level, producing high-level characters who are more likely to fail saves and suffer the full effects of spells and abilities from 'level-appropriate' opponents than they were at 1st level.
Will it be game-breaking? Will the action-point-like Inspiration mechanic save the day? Will high-level 5e characters just have so many HP and healing resources that they don't care about failing saves? Will rings of protection rain from the sky, and everyone decide that they only really need two other attunement slots? Remains to be seen. Answer unclear, try again later. Go play the game yourself in the upper-teen levels, and let me know how it goes. But know that the risk is there with 5e saves as they are structured now - as a flattened mirror of 3.x's save structure.
(Even better is that since spell save DC scales with proficiency bonus rather than spell level, a 20th level fighter is more likely to fail his save against charm person cast by a 20th level wizard than he was when he and the wizard were both 1st level. Do not dismiss the utility of low-level off-stat save-or-sucks)
Forecast for next post: substantially less doom and gloom, an alternate paradigm, and a modest proposal.