One situation where Knox has an advantage over humans is when the board *quickly* divides up into a few regions. To try to

take advantage of this, KnoxB has a modified opening strategy where it will try to divide the board up into regions *even* when it “thinks” the resulting parity will be bad for it. When this happens, KnoxB should still have a good shot at taking control through the use of its “nimstring strategy.” A good example of this occurred in the Knox vs. ebm game of the second championship. On move 8, ebm divided up the board into two regions which appears to be in his favor. Knox was able to apply nimstring on move 11 and it turned out that ebm's 10'th move was a mistake. As a result, Knox was able to take control and win (The only reason for Knox's 9'th move was to reduce the number of edges in the largest region so that it could apply nimstring as quickly as possible.) Of course one example doesn't mean a whole lot. It could be the case that the modified opening strategy consistently results in Knox

getting into a bad position in which good opponents will be able to win most of the time. And that's the reason for testing this opening idea out; I don't know what's going to happen.