We argue that the dynamic growth of the world-wide computer network has a revolutionary impact on scientific research and its spatial organization. Due to third generation Internet, scientific research can be done very far from real scientific centres. In those parts of the world where researchers have access to the fast Internet, scientific research undergoes democratization. The world-wide computer network made revolution not only in communication between scientists, but also in the way of approaching scientific discovery, creating virtual laboratories and infrastructure for computing 'on request', called grid computing. To illustrate these revolutionary changes, we use examples related to the Polish optical Internet owned by scientific community. We also present another, rather troublesome consequence of networking - the problem of the growing gap between data generation and data comprehension. To bridge this gap, a new discipline has emerged, called knowledge discovery from data, or machine intelligence. Knowledge discovery is a process of identifying true, non-trivial, potentially useful and comprehensible patterns in data. The patterns can be used for explanation of phenomena described by data and for prediction of their development. Patterns discovered from data may also become new scientific hypotheses, which is new comparing to the way of approaching scientific discovery before Internet.