As the computer networks grew, I became a frequent flyer. The LEP experiments at CERN became the largest sources of scientific data in Europe, if not in the world. Networks are built piece by piece by human beings and require close technical cooperation to make sure that links and routing systems are configured correctly at both ends, not to mention paid for at both ends. A peculiarity of most international data links was that they had to be bought as two ‘half circuits’, one from the telecommunications operator at each end. Even as the national telecommunications monopolies were progressively replaced by competing companies through the years, this ‘half circuit’ payment regime survived in most cases. Thus, as well as coordinating technical matters, we had to coordinate the financing of links, so there had to be rules about whose traffic was allowed on which link. CERN’s policy was that we would only pay for network capacity used by our own staff. The physicists from each country were to pay for their own network access to CERN, and for access back to their home institutes during the part of the year that they spent at CERN. The European countries, even though they were the main contributors to the CERN budget, generally accepted this model as fair. However, the National Science Foundation in the USA, responsible for funding international connections for NSFNET, couldn’t understand why CERN refused to pay its half. Our position was that we connected to the NSFNET for the benefit of American physicists, so they should pay. In the end, they did so, through budgets assigned to high-energy physics, not to networking. However, this dispute was why IBM’s help in funding the first high-speed transatlantic link was so crucial. Once that link had shown the physicists what they could get from such a connection, there was no going back when the IBM funding ended, and the money was found.