The web-based half RTS/half TBS genre has incorporated the idea of X command points per Y time period as part of the norm for a long while (Travian, et. al.) Turns in such games are generally not immediately exposed as mechanics (if they, in fact, exist at all). I recall an old space based RTS (the name escapes me, sorry) which exposed the amount of time left until the next tick, at which point all players would get more command points, while Travian appears to have a tick per second, with long waiting periods (relative to tick length) between updating command resources.
One thing which might be useful as a thought experiment to further this article is to look at TBS games which aren't dying--why is Civ V still popular? Or, to get even more basic, why is baduk still widely popular? Some thoughts regarding the latter: each move is very granular, which keeps the game moving foreward. Each move has an immediate impact on the game state (this relates to your point that "nothing happens in the first move.") There is a deep element of strategy, which is augmented by a need for strict time management. It's easy to review games and find mistakes, as well as to review games of higher level players to find new ideas and strategies.
PS: Frozen Synapse is part way there, but IMO units should be a little more capable when not told what to do.
I'll concede that turn-based has never been a good fit for multiplayer, but I was hoping for some analysis on single-player failings. There are reasons why turn-based single player games have at times been very popular, but many newer entrants seem to have forgotten what those are. The turn-based system gets rid of skillshots at the cost of respondability, which can both improve and degrade approachability.
I also wouldn't consider the two offered solutions to both be turn-based. I feel the Travian system of ticks is realtime, whereas the Age of Wonders system is mostly turn-based (though exploitable). I'm on the fence about a Planetarion style system, with sparser ticks - it retains some good parts of turn-based, but adds some bad parts from realtime.
I'm pretty sure there has been a multiplayer game like Frozen Synapse, which is the third simultaneous option. I feel like this is the only truly turn-based variant, though a very different system from traditional turn-based games.
Of course, there is the final, popular method of simply setting time limits on turns and simplifying the game until turns are fast. I dislike this approach as well - maybe turn-based games are better left in single-player mode.
I think a good example could be Frozen Synapse. Not the first , but he has fixed some of the listed flaws .
I still think a Hero Academy type system is probably best. It doesn't give you the complexity of building up multiple cities Civ 5 style, but it does give you a multiplier final fantasy tactics, which is awesome!
One that comes to mind is Diplomacy. TBH I think Diplomacy is a great game, but it actually takes way longer than a normal board game (probably because everyone spends a long time negotiating during every turn to optimize each move) and usually ends with some group of winners rather than playing the game out.
I think designing a strong game of this type is extremely hard.
The distinction between "turn-based" and "real-time" is largely perceptual rather than actual. "Real-time" simply does two things: it slices the turns down to an extremely fine grain (such that they're more "frames" than "turns"), and it forcibly makes "do nothing" or "continue previous action" the default action.
Thus, the real distinction for a designer is, "How much time do you want your player to have to make a given decision?" The more time you're willing to give, the more likely you should make it turn-based, and vice versa. This is how chess starts to resemble an RTS when you add a clock. If you pared the "time to move" down to, say, 10 seconds... it becomes hard (but not impossible, since you're still right) to argue that chess is really turn-based.
That's the distinction, IMO, not turn length.
What matters is the player experience, not the implementation.
On a digital side, I think the Pokemon games does simultaneous turns very well. There is an entire competitive community that is built around such a simple system, and I think this format has much room for improvement. I'm currently working on an online multiplayer game of this format, which tries to explore design ideas that Pokemon doesn't (as it seems 95% of moves are of the variant, deal X damage of a certain elemental type).
I also had lots of fun playing http://www.kongregate.com/games/Kongregate/kongai that is also turn based and example of great game design.
What I've come to prefer is RTS with a live pause: that is, you can pause at any time to give orders, line up production queues, etc. It works best for single player games, which is what I prefer anyway. Paradox Interactive's games (Crusader Kings series, Hearts of Iron series, Europa Universalis series) are prime examples of this method.
(Disclaimer: I worked on a spin-off of Europa Universalis II; see my profile for a link. I started on the project after having played the game for six years.)
For me, the downfall of that game was the excessive need to write down information outside the game, e.g. plausible enemy fleet positions based on where you last saw them.
Yes XCOM is trading in large part on nostalgia, but I know plenty of people who never played it the first time round picking it up and enjoying it.
Not to mention the slew of TBS games on iOS (and I'm guessing Android) where the stop start nature of the play ties in well to how and where we use those devices.
And for the record the AI in that game could be brutal.