Interesting stuff here. If I may ask, what prompted the move from turn based to "slow form" RTS? I like both, I'm just curious since you apparently started out with a TBS design. 
My god, the worms are coming everywhere out of this can!
Well, this is a really complex question on the surface, actually, because I had an enormous multitude of reasons for making the switch. Trying to explain this succinctly, though, the shortest answer is that I prefer RTS games in multiplayer, and I wanted this to be a multiplayer game.
I'd had a bad experience trying to play turn-based Civ IV and one of my teammates just always taking 4x longer than the rest of us, making us wait around (yes, I know there was a turn timer). But at the same time, Civ IV was the most strategy multiplayer experience I'd ever had. For years I had thought that our co-op comp stomping sessions in various RTS games were limited in strategy because the AI was bad; that was partly true, but when we started playing pvp SupCom I didn't find it any more strategic. Lots of tactics, sure, and I love tactical games, but I have a desire more for RTS than RTT, and I lump a lot more games into RTT than most folks (any of these shorter-form games, including Starcraft and similar).
Okay, I'm already going off on huge tangents -- time to dial that back. So anyway, I went into this knowing that I preferred RTS games, but having the following beliefs that later proved to be false:
1. That I wouldn't be able to program something efficient enough to run over the network in a realtime gaming context.
2. That any AI I coded would bog down a realtime game too much, even from another thread, in the form of network commands or otherwise.
3. That a realtime game inherently couldn't be as strategic as a TBS game because of the nature of it being realtime.
4. That there was no way I'd be able to have galaxy-wide planets simulated in realtime, and that having 10,000 units on the map at once (particularly in C#) would be a huge stretch even for the latest computers.
A lot of this was based on my past experience. I knew that C# was supposedly 11% slower than C++, which isn't killer, but is significant (since doing AI War, I've begun to question that, though, but that's another tangent). I also just had not ever seen a realtime game with the strategy that I wanted. I also had most recently been playing SupCom FA with the sorian AI mod, and the game was running at 1/3 of realtime in that context because of all the extra AI processing, and the cross-threading stuff wasn't working as it should have been in that game (reportedly, and seemingly). I'd also seen an enormous amount more latency with the largest supcom maps, even when they were unpopulated (but later I realized that was because of all the extra meshes and pathfinding, two things I did without in AI War because of its nature).
God, I told you this was a can of worms question.
Okay, so I'd gone into this with false assumptions, and decided to make AI War a "TBS game in realtime." Basically, to where you'd get something like 20 movement points every turn, and turns would last 20 seconds no matter what you do. You could store about 100 move points total, so that there was a bit of give in it before you'd start losing your move points, but it was still meant very much to keep things moving.
The 0.065 alpha up there is from somewhere during my experiments with that style of mechanic, and in practice it was just... not fun. The planetary areas were large enough that it would take 10ish moves just to get ships from one end of the planet to the other, and there were supposed to be hundreds of planets and thousands of units!? It was boring and slow, and there was far too much empty space for a TBS game. And far too much micro inherent in a TBS game to have that many units.
So, at any rate, I decided to take a leap of faith and see what I could do with an RTS game. I figured that the above factors would make me have to cut back a lot on my planned design, and that fewer units would be the order of the day, etc, but I also figured I could do some tricky things with compound units and "fleets" that were really one big unit group given orders all as one unit, etc. I figured I could bypass a lot of the larger processing, AI, and networking issues that way if they came up.
Of course, as development progressed, I found I never had to do that, and in fact I kept upping the unit count until in late beta we were routinely having 30,000 units in the late game, and sometimes 50,000 (though it would get pretty laggy there). Now, of course, even 70,000 is routine, and 150,000 is not unheard of.
The shortest possible answer is that I started with TBS because I thought that's all I'd be able to do while still meeting all my design goals, but I took a leap with the RTS and it worked out. I do love TBS games, though, and at some point I'd like to do one that is designed more around TBS concepts rather than trying to mesh in so much realtime stuff.
(Whew!)
