Essentially the notion is that, because a player's ability to address their opponent's Titan takes at least three turns, one to place Target markers and one to replace those Target markers with Blast markers and one to take effect, that the opponent will always be able to avoid incoming fire. Now, aside from the feeling like I did not competently explain the concept, I feel like the critique misses a couple of notions about the game itself, namely the concept of there being limits to a Titan's mobility, limits to the length of the game, and incentives in the form of objectives that mean that whether or not to tank a hit is a very real decision-point in the game. The notion, of course, is that catching the other person's Titan in your cross-hairs should be like a check or check-mate in Chess, something that is in principle predictable, but in actuality a matter of clever set-up.
Of course, the random thing about Chess is that it lasts any number of turns. In a game lasting a definite number of turns, then such un-predictability would be nigh-impossible, as players can simply look at the winning conditions on the last turn and calculate backwards. That means that vast sections of game-play would be cordoned off with boredom: a solved game is not, somehow or other, a good entertainment. I would like the game to be finite in temporal size, as I think it's important that players operate with a time-crunch on how long they have in the game to achieve their Objectives. I feel like it's this time-crunch that will force them to do more interesting things than attempt to circle each other's Titans in a boring loop of fire-and-dodge.
Of course, I could simply go with the Foundational Axiom approach, as in Go, and state that acting to repeat a game-state is not permissible. The problem, of course, is that beginner players may not be able to recognize when the game state is repeating, because the game state will be more complex than that of Go, which is very, very complex (and yet requires the Ko-rule).
Now, that is my temptation with Titanomachia, to throw in a randomization mechanic based on a pay-off table. I've seen them before with some varying degree of success, most notably in the Darksword Adventures where the players play a kind of ten-valued Rock-Paper-Scissors which probably existed in order for players to throw it away an concentrate on the story-telling part of the game, which was a pretty important innovation back in its day (or at least to me in the notion of RPG sessions as a way to tell a story rather than to simulate ethnic cleansing). There's always dice, but I've strongly resisted the temptation to just remake Adeptus Titanicus with more modern-flavoured Warhammer-esque rules.
So what I think is the way to go is to take a page out of the Warhammer 40,000 5th edition book (really a page from A Very Short Introduction to Game Theory by Ken Binmore) and make the game last an uncertain number of turns. In Warhammer 40,000 5th edition the game lasts between 5 and 7 turns, with the odds of the game ending on any of those three turns being equal (1/3). So the game has to end on turn 7, but it could end earlier. I think this is sufficient to prevent both the last-turn game-jack problem that faced Warhammer 40,000 4th edition, and the notion that knowing when the game will last will allow players to narrow down and identify the optimal set of actions in the Titanomachia game described in the two prior posts.
The question then is how to implement this notion, avoiding dice if possible. My thought is that I could probably do something neat with the fact that the board itself, as it stands, is a 10x10 grid. Essentially, properties of the grid can be used to measure game-time, such as the number of turns, or rounds, and act as a game timer. Honestly I haven't thought this all the way through, but I think it's pretty interesting in and of itself.
I think I'll make this a part 1 and pick it up later in a part 2 article so that I have some time to develop this notion.