We can learn from IBM's successful history that you don't have to have the best product to become number one. You don't even have to have a good product.

The most valuable thing you can make is a mistake - you can't learn anything from being perfect.

With bundled machines you can throw away the hardware and keep the software, and it's still a good buy.

Learn to say 'no' to the good so you can say 'yes' to the best.

My main goal is to become the most successful weightlifter in history.

I take what I see work. I'm a strict believer in the scientific principle of believing nothing, only taking the best evidence available at the present time, interpreting it as best you can, and leaving your mind open to the fact that new evidence will appear tomorrow.