With this week's proposal from DESY for a 500-giga-electron volt linear accelerator (see page 397), the time has arrived for high-energy physicists to start thinking seriously about how, when and where such an accelerator should be built.

Three principle efforts to design it are under way: DESY's superconducting TESLA, and the more conventional Next Linear Collider, being developed in the United States, and Japan Linear Collider. The question of choosing the right design should be made independently of the deeply political decision of where to locate it. Sadly, in the world of big science, such distinctions are rarely clear-cut.

For US physicists, still smarting from the cancellation of the Superconducting Super Collider (SSC) in Texas in 1993, decision time is perhaps arriving a couple of years too soon. The expenditure of several billion dollars on the SSC before its cancellation has left an unfortunate gap between the likely cost of the Next Linear Collider — estimated at $7.9 billion in 1999 — and the amount the US government will realistically support. Meanwhile, Japanese physicists, who deservedly have held high hopes of hosting a large linear collider, face a changing political climate in which funding may be increasingly difficult to obtain.

In Europe, where official cost estimates for a linear collider may conceal an actual price (of the sort required under US law, which includes the costs of labour and operations) similar to that ascribed to the Next Linear Collider, it is unlikely that sufficient support can be gathered from the German government or through a European collaboration, not least given the continuing demands of CERN.

The way forward is to develop an effort that is genuinely global from the outset. There are at least two prerequisites for successful collaboration, neither of which yet exists: robust political support at the highest levels, and an acceptance by Europeans that, with CERN's Large Hadron Collider, they have had their turn. The latter could help high-energy physicists worldwide acquire the former.