It sounds like JSTOR's servers aren't really optimized for high article download rates, to the point that his one laptop accounted for a significant part of normal continental US load. They probably had some bottleneck in the system that they never noticed before — maybe their logging infrastructure was absurdly slow or something.