Here is our first example! Figure 1 shows a timeless beautiful position: the empty board, Black to move. After playing 10000 random games, Gobble assigned to each field a number, ranging from 0.9 to 5.3 in this case. These values are the average game results for all games in which Black was the first to move onto a field. For example, if Black was the first to take the center field, whether it is in move 5 or 50, all such games resulted in an average point count at the end of the game of 5.2 points. In particular, 5.2 is not the average value for games with Black's first move at the center.
The main feature is that moves near the center give better results than moves onto the border and corners. This result is non-trivial, indicating that strategy A is able to find one of go's simple rules of thumb. The absolute worth of the first move amounts to about 5 points, which is lower than it should be, but remember, these 5 points are just the advantage that Gobble knows to derive from the first move. During the computation it is nice to watch how the average value of all games increases steadily as Gobble becomes smarter and tries better moves.
This one example already gives an impression of the statistical errors involved. The board is symmetric, but equivalent moves do not get exactly the same values (consider the corners, for example). An estimate for the error could be Δv = ±0.6. The values near the center are more evenly distributed because good moves are more often tried at the beginning of games which leads to more consistent results. Bad moves are played near the end of the game when they tend to be irrelevant and less correlated with the outcome of the game.