A Go Game

What sort of game do I want? I see the beginning of the Go game as a conversation, where I decide with my partner what sort of game I want to play.
Do we want to play a standoffish game of opaque strategy, do we want to have a singleminded scrap across the whole board. Do we want to play similar styles, or pit different styles against each other? Do we want to cooperate, or argue? Do we want to experiment with new ideas, refine developing skills, or exercise prior knowledge? What combination of these things do we want to do?

So, the opening is like a statement of policy. After that, the outward aim of the game is to see whose policy works the best. Go's excellent handicapping property (it's too basic and intrinsic to be called a rule) is such that the strength of the players should only be a factor in the choice of policy, not a decider of the outcome of the game. (Unless the strength gap is too great, in which case teaching games are more appropriate.)
Go helps you realise your own strength, not waste it, and not have a false impression of it. If you adopt policies that you don't have the strength to follow up, you lose. But if you adopt policies that you easily have the strength to follow up, if the handicap was set correctly, you will lose because you wasted your stones.

If your strategies turn out to be ineffective, you have to find the agility to move onto effective strategies. Go is full of balances like this.

Maybe I will play an inflexible strategy in the hope that it is one my opponent doesn't know and will lose to. Maybe I will try and play as flexibly as possible, partly to hide my lack of a plan, and partly to be ready for unexpected events. Good go is a combination of all these things. And the size of the goban means you can adopt a varying mix of these in different places on the board.

So by the midgame the strategies are decided, there is no more room to change them (barring swindles). It is up to both players to follow through with the promises they made at the beginning. If they don't, they lose.

If the handicap was right, and both players made good choices, the result may be a draw. The endgame is an opportunity to avoid a draw by making up points in the last little fights over small amounts of remaining territory.

So perhaps one view of Go is this. The opening is a gamble. Do you have the skill to pull off the strategy you want to use in order to win? Do you want to win on the strategy, or rely on using superior strength to take a big swing in the midgame, or do you want to make it a close game and take the win with little points in the endgame?
Because stones are played in turns, both players base their decisions on the intentions the other communicates through their stones.
Once the bets have been made, they have to be tested. As the game progresses there is less and less opportunity to change your bets, and you are forced to follow them through.
By the endgame, both players are scraping up the last points, either to emphasize or blunt a victory, or to snatch one.
Gamble, prove, scrape.

An ideal game of Go is a draw: both players chose appropriate strategies for their skills, the handicap was correct, and neither could break the others choices. Because the skill of individuals varies widely and according to context, and because the available strategies and tactics in Go are so great, this means that in reality the game may not always be a draw. A well balanced game that isn't a draw and where no mistakes were made represents both players exploring new territory. A well balanced, flawless game that is a draw represents both players practicing on charted territory.

A good Go player works a balance between practice and exploration. In a full-size game there is often opportunity to do both.

Go has very much depth!

Bearing all that in mind, playing Go against a computer is quite strange, especially when it's better than you! It's as if the computer is beating me at a Turing test.
Once you've gotten over the weirdness of taking a game against a computer seriously enough to play properly, the fact that it's actually better than you is quite interesting. (Remember that number crunching is not a big item in the armoury of a Go player). Perhaps this is the one way in which a computer can actually teach me (albeit only for the beginning stages), rather than merely being a portal through which I view information.