I've only got ten minutes so ... we spent a lot of time covering "all pairs" testing today. Basically the idea is that if you have say X number of areas to test and each with Y, M, Q number of different possible options (imagine browser versions and video cards and types of network connections) that you can seriously reduce the number of cases you need to execute by just testing that you have each PAIR tested together rather than every single options. This can reduce say 300 test cases down to 30 or something much more reasonable. Per an examination of failures on pacemakers caused by software, testing for just PAIRS of possible failures caught 90% of bugs - you didn't even need to move on to testing A (condition1) B (condition2) C (condition1), just make sure to get each version of A+B and each version of A+C and each version of B+C. There are lovely tables called "orthoganal arrays" that explain it all, but I can't really do it justice in the time I have.
I actually have a "bonus session" starting in six minutes on how to give a good talk, so I'm going to run off and do that. Miss you all, even though it is really nice and warm here and I finally found a good place for Mexican food.