In case anyone is wondering, the way to interpet the rules is as follows.
* The state of the system is a black-white assignment of colors to the grid.
* The rules tell you how to compute the next state of the system.
* To update the state at a certain cell x, you look at the colors of x-1, x and x+1. (left, self, and right).
Then use use the table of rules to determine the new color of the middle cell.
For instance, the first rule tells you that if you see three black cells in a row, then in the next time step the middle cell is white.
* This update is done simultaneously over all cells, so you compute all the new cell colors and then update them all at once.
If the pattern matches the top three tiles, then the tile below is defined by the bottom tile in the rules. This is done top down, row by row for each 3-tile section of a row. You're totally right though.
There are no edges, it's an infinite grid. You start with 1 black cell and all the other cells are white.
Because the triple-white configuration does not produce a black cell, you can compute up the any finite time step with finite computational power.
(Actually, later in the article he talks about finite grids with periodic boundary conditions. That means that, if you're on the edge, you 'wrap around' to the other side of the grid).
* The state of the system is a black-white assignment of colors to the grid.
* The rules tell you how to compute the next state of the system.
* To update the state at a certain cell x, you look at the colors of x-1, x and x+1. (left, self, and right). Then use use the table of rules to determine the new color of the middle cell. For instance, the first rule tells you that if you see three black cells in a row, then in the next time step the middle cell is white.
* This update is done simultaneously over all cells, so you compute all the new cell colors and then update them all at once.