An interesting little titbit came up in the recent AI thread in General Chat. Including staying where he is, a 6 MA player has 169 possible moves, excluding any dice rolling. With 11 players on the team, the number of possible moves is upwards of 1800; add in different setups and the dice rolls, and the number of possible moves and consequences expands again. Multiply this by the next 7 turns, and before half time, the number of potential outcomes has exploded.

I'm prompted to consider this by some of the ongoing strategy discussions in the forum, because it seems to me questions of general strategy revolve around picking the best possible route through these branching possibilities; a brute force consideration of every option is impossible, pro-elf players in the last turns of a particularly rough match aside. Instead, we have to choose goals for the turn, for the half, for the game: these help us narrow down the moves from the possible to the manageable. In most situations, this still leaves a staggering range of possibilities, made more complex by skills & races, so we have to add in a consideration of what is plausible from our opponent: what limitations their set of possibilities has, and what guides we think they're using to narrow down their choices.

Yet even this process leaves us, usually, with more possible moves, all of which could apparently serve to advance us towards the goal at the end of the probabilities, than we can accurately evaluate. How do we decide which ones should be abandoned, and which are worth pursuing? It's clear that some players are more adept at pruning the probability tree. Somewhere, experience and talent allows better players to select optimal routes, while worse players are overloaded with choice and end up in cul-de-sacs. The actual moves that occur to players in-game are only the conscious result of a much more complex and instinctive process.

I think the point is probably three-fold.

First, the sheer mass of possibilities involved in every turn mean that few games will ever truly turn on a single dice roll; this is an illusion based on a bottleneck in the probabilities. Back up, and there are multiplying routes to and around this point; go forward, and the possibilities will expand again.

Second, experience is important; although few coaches consciously consult their memories of every single match they have played for early boundaries on the probability forest, games clearly lay down some instinctive sense of it: my own record is proof of that, in a move from very to merely bad that cannot just be explained by knowing the client better. We could take a stab at defining the prerequisites for bloodbowl talent, but these are given; the best we can hope for is that experience teaches us to use them to our best advantage.

The third is one which was touched on in the AI thread: BB is flat out hard. I seem to remember reading somewhere that top level go-players suffer a disproportionate number of cerebral haemorrhages. Win or lose, perhaps we should be more amazed that we play without our brains exploding from our ears.