It annoys me how a lot of "new ideas" for 5e already existed in earlier versions.
Symbol of Death always had a HP threshold (60 each in 2e, 150 total in 3e), Power Word Kill always had a HP threshold (60 in 2e, 100 in 3e), Disintegrate in 3e dealt normal damage (only disintegrates if it deals enough to kill the target anyway), and all kinds of effects had HD thresholds or caster level thresholds.
What would be new is if they weren't implemented poorly, or at least were better integrated into the other system assumptions. Thresholds, especially static thresholds, in past editions often meant a spell started out extremely powerful, and then became worthless. Sleep and the Power Word spells (which I dearly love) are paragons of this thought.
I think rethinking thresholds would go a long way to fixing things up. As pointed out earlier, thresholds based on current hp (as a % of full) have inelegant effects on low-hp but 100% healthy creatures. The first thought to fixing that is to go bad to old-style current hp thresholds, but those have problems of their own (especially if you want more than a binary it works/it doesn't outcome).
Why not do both, with a twist? Rather than check current hp vs. a threshold one could apply "phantom" damage, check what % of hit points a creature would have left, and then apply the effect accordingly. The phantom damage isn't actually damage, it is only used to determine hp for the purposes of applying an effect.
For example, consider the medusa's gaze. Say the gaze did 30 points of this phantom damage, and the effects were something like:
<= 0 hp: petrified (no save)
1 - 1/4 hp: petrified (save), slowed on save
1/4 - 1/2 hp: slowed (no save)
1/2+ hp: slowed (save)
For creatures with 30 hp or less it is unavoidable petrification, whether they were injured or not. (It emulates the sort of threshold that is familiar to us.) However, suppose a creature with 100 normal hp, is currently at 50 hp. When the medusa gazes at it the phantom damage means we check the effect as if it had 20 hp instead. In this case that means save or be petrified. If he makes his save he is slowed but still has 50 hp.
Another example might be Power Word Stun. Suppose it does phantom damage dependent on spell level, but the effects with respect to % of hp remain constant. Something like:
Phantom damage is 5/spell level, and the effect is
<=0 hp: Stunned for 1 minute
1-1/4 hp: Stunned for 1 round (no save)
1/4-1/2 hp: Stunned for 1 round (may attempt save at start of next turn)
1/2-3/4 hp: Stunned for 1 round (save), if failed may make second attempt at start of next turn
A 1st level wizard walking around with this spell would be stunning housecats without a save, while a 20th level wizard is doing it to 10th level wizards, but not to the Tarrasque.
For effects that do damage anyway one doesn't need phantom damage to avoid the problems of determining effect by % of health. For example, Polar Ray might do 2d6 damage per spell level, and an effect as follows:
1-1/4 hp: Immobilized for 1 round
1/4-1/2 hp: Slowed for 1 round
Likewise, it is easily to replicate the binary results of a more traditional hp threshold using phantom damage: One just gives an effect for <=0 hp, and no others.
It seems to me this is close to the best of both worlds for thresholds, and very tunable to the specific mechanical and thematical needs of any given effect.
The action economy is the trickiest thing in my opinion. Solos need ways to damage multiple PCs every round. I don't think this is something that should scale by level.
Take the 4e beholder, which gets a free eye ray any time somebody activates within 5. Great for a solo, but can you imagine fighting against five of those?
The action has always been the most important resource in the game, especially since D&D doesn't have much of a death spiral, so I definitely agree that it's tricky.
I think the important thing here is that extra actions can be actual, or just "effective." I mean, in many respects a fireball is just like taking a lot of actions to attack a bunch of people. In that respect, is it so bad if it's a single action?
If this sort of thing did go by level, maybe one could divvy it up smartly. Fighters get extra attacks each round, as per tradition. Perhaps wizards also gain extra "actions" but their high level spells generally require several to cast. So a 20th level wizard might have 3 spell-casting actions, but spells of levels 1-3 might (usually) require 1 action, levels 4-6 2 actions, 7-9 3 actions. A wizard might choose to cast his most powerful spell, or 2-3 lower power spells on a turn. (I'm not in love with this implementation, but it does point in the direction I'm thinking.)
Extending that kind of thinking across the game might allow some cool things that never really worked well in previous editions. For example, the haste spell might actually give one an extra action, and slow could take one away, without some of the terrible brokenness of earlier editions because creatures are expected to have 2-3 attack actions anyway. (At low levels it would still represent a doubling of attack capability, so at the very least the low level version should probably only affect a single creature.)
I grant you, the beholder would have to be done very carefully, so that it has enough actions to be a solo without being onerous to run in larger numbers. The 4e beholder has a very 4e-like method of making sure it gets extra actions, but if we change the action economy it wouldn't make sense to keep it that way. For example, what if the 5e beholder could shoot 5 rays in a round, and could do so at any point during the round. That bounds the total number of actions it can take so it could roughly match action-for-action to a low-level party. (One automatic ray per enemy turn bugs me because it is unbounded, and because it depends on circumstances wholly external to the beholder itself). When the party gains several levels the beholder might still have a few attacks more than usual for a monster of that level, but that isn't so bad if the eye rays can be resolved quickly.
