"Access to supercomputers. The science is well ahead of our ability to implement it. It's quite clear that if we could run our models at a higher resolution we could do a much better job-tomorrow-in terms of our seasonal and decadal predictions. It's so frustrating. We keep saying we need four times the computing power. We're talking just 10 or 20 million a year-dollars or pounds-which is tiny compared to the damage done by disasters. Yet it's a difficult argument to win."