First read this - if we're going to have a conversation, context is good. It's an idea they're currently testing internally about how the fighter class might work in 5e, and I'd like to discuss what I suspect are the origins of the idea, some strengths, and possibly some weaknesses (or at least lingering questions). Ironically, it's also a lot like Samhaine's post this morning, but differs in one critical assumption. Let me start by saying that I absolutely encourage more decision-making and Stuff To Do on a fighter's turn than just rolling attack and damage, and more new benefits in character progression than higher numbers for attack, damage, and defenses. Bigger numbers are fun, but they aren't decisions.
So 4e had encounter and daily powers, which were more powerful than at-will effects, but could not be used as often. This was particularly important for fighters, as the game was going way out of its way to give fighters more tactical choices to make and more control over the playing field. At higher levels, the available encounter and daily powers do more damage, attack more targets, or have more desirable side effects, whether good for allies or bad for enemies. There's enough internal variation that first-level encounter powers can still be useful at tenth level and later, but eventually you are expected to drop some of those earliest powers in favor of higher-level options, often with very similar status effects, but more damage. If you gather enough data points, you could almost construct a point-based powers system (for all I know, WotC has something like this, but I kinda doubt it), allowing you to trade damage potential for status effects.
There's one other significant thing here: powers with the Weapon keyword (basically all martial-sourced powers, as well as most melee classes of other sources) based their damage on the weapon's damage die, expressed as 1[W], 2[W], and so on. That is, a d8 weapon deals 2d8, 3d8, and so on. They didn't keep talking about it later in development, but I recall a mention early in 4e of why this was important, as a development from 3.x. Specifically, at higher levels in 3.x, the base damage die diminishes in overall importance to a character's damage output, as the damage bonuses of Strength, feats, enhancement bonuses, and variable adds (flaming, etc.) such have scaled pretty steeply. Critical hits (and, in AE, Crushing Blows) multiply damage dice, but otherwise it's not common in 3.x. I've felt that 4e went in a good direction here, to make sure that weapon choice remained relevant and players didn't find it too easy to go for smaller damage-dice weapons with miscellaneous advantages.
That, then, is one of the things I see as a potential problem with the Combat Superiority plan here. All I have to go on is what they've shown in Legends and Lore (and other columns) and what's in the playtest documents. We can reasonably assume that a fighter's damage expression for a normal attack will be [W] + [CS dice] + Str + enhancement; there may be other factors as well, but all of these factors will definitely be present. It's probably reasonable to guess that the longsword will continue to use its classic d8 damage die, and a greataxe will use a d12 or 2d6. A fighter dealing [W]+d4+3 at first level is totally reasonable, as the d8's average damage is only one point less than the total damage of the rest of the equation. If the damage dice have scaled to 2d6 by fifth level, it's now less than half of what's granted by the rest of the equation, and a d12 is only barely better off - less so if the fighter has picked up a +1 weapon by this point, which seems pretty reasonable to me. For all that I've gone on about it, this question of math isn't a deal-breaker, but it does surprise me, and it factors into a point I want to make later.
There's also the issue of class comparison. I was pretty shocked when the playtest documents had the rogue's Sneak Attack scale at 1d6 per level. My experience of 3.x told me that 1d6 every other level was too fast, such that mid-to-high level monsters were either immune or one-round bloodsmears. The good news in 5e is that flanking isn't enough to grant advantage from what we've seen so far; the most reliable source of advantage for the rogue is spending a round to hide (sometimes behind a party member), therefore attacking only every other round. Now, I don't have any evidence that Sneak Attack damage continues to scale in this way past third level, or that it hasn't changed - this is pure extrapolation. If it holds up, though, we're talking about a fighter dealing d8+2d6+4 damage on a hit (attacking once per round), while the rogue deals d6 (weapon) + 5d6 + 4 (Dex and enhancement) every other round; also, the rogue is potentially hidden and thus immune to anything that has to be targeted, half the time. It's all just... kinda weird.
Anyway, let's move on past damage-increasing uses of Combat Superiority. The article mentions using the Combat Superiority dice as damage mitigation; this is at the same time very interesting and very scary. The main question is how often the fighter can roll and apply these dice: every attack? Once a round? When in the turn order do you decide how you're using your CS dice? We're probably better off with variable Damage Reduction than a flat value, even if the die roll slows down combat, because characters are less likely to become simply immune to, say, a monster that does d6 damage. At the same time, the whole plan of making low-level monsters stay relevant at higher levels is in serious jeopardy. The whole plan is based on not scaling AC and other defenses, but if a fighter's hit points and his mitigation are scaling, that's right back where we started.
Thus far in this post I've been pretty negative about the Combat Superiority system, but that doesn't really encompass my feelings on the idea. There's a lot of good here too - I think figuring out where to assign dice is an interesting decision, and I'm glad to see the game look for ways to make defense feel more active. This was - as I've said before - the single best part of Warhammer Fantasy 3rd Edition. I also like the possibility of a semi-freeform stunt system built off of trading dice to tack on effects, or to make riposte attacks, or any of the other awesome things that 4e powers granted.
I mentioned Samhaine's post, about various types of "energy" currencies that allow characters to use better powers. The difference here is as simple as whether you start with your fuel tank on empty and have to build it up, or receive the same amount of fuel each round, regardless of what you did last time. As Samhaine mentions, his version has the benefit of stopping the alpha strike dead in its tracks, and does offer a good amount of game support for narrative tension. Its downside, which he acknowledges, is that tracking a lot of fiddly things (even with tokens) can be a burden in play; Iron Heroes and WHFRP 3e are pretty clear example cases.
The unstated benefit of Samhaine's version is that if the enemy is prepared for the battle but you're not (here I presume some kind of preparation mechanic that supports ambushes), you could be in real trouble before you can turn things around. To put that another way, you're not always at the top of your game. 5th edition can create situations where a fighter is denied CS (probably hand in hand with having disadvantage), but that's not directly indicated here. Also, I have to wonder if this will be one of the mechanical dividing lines between PCs and NPCs, or if they'll create NPCs with Combat Superiority dice (even if restricting that to named, important characters).
Samhaine's system defines and rewards archetypal behaviors for all classes, while the Combat Superiority system is explicitly intended to be restricted to fighters. Beyond that, the CS system has you choose fighting style in terms of ways that the fighter will be permitted to spend dice. Only final implementation can really tell us how restrictive either of these would feel in play.
This brings up another question about the CS system, actually - will it Play Nice with multi-classing, or will class-dipping in fighter come to be painfully suboptimal in 5e the way class-dipping in a spellcasting class was in 3.x? The reason fighters and rogues were such good options for class-dipping in 3.x was that the abilities the classes granted stacked well with whatever you were getting from most other classes.
In conclusion, there's a lot going on here, and not enough to make a definitive judgment (or even get serious about applying it to public playtests), but they've hinted that the next public playtest documents will hit in late summer, so maybe we'll get to try this out in a month or two and find out if my hopes are more justified than my fears. As with a lot of things in 5e, I'm cautiously optimistic; even if the final printed material isn't perfect for my needs, it sounds like it will be close-enough to doing something new and different that I like. (New and different: technically here I mean recombining ideas found elsewhere. Like you do.)