One reason why the United States elected to place its nuclear weapons technology under the control of a civilian agency was that issues surrounding the weapons were seen to be too important to be left to a military clique. The resulting system does not allow information on nuclear weapons to pour out into the public domain. Important technical information is kept secret — but it is nonetheless made available to a fairly wide circle of security-cleared experts, who are able to use it to inform a relatively open public debate on weapons policy.

There are disquieting signs that the Department of Energy (DOE) is failing to adhere to this mode of operation, as it singlemindedly pursues its intended policy of science-based stockpile stewardship — a programme developed to maintain the viability of existing US nuclear weapons after the proposed implementation of the Comprehensive Test Ban Treaty.

Ray Kidder, a weapons designer with a long and influential track record in both theoretical physics and nuclear weapons policy, has been denied the opportunity to help the Arms Control and Disarmament Agency (ACDA) to assess maintenance options for nuclear weapons (see page 622). The denial came after the energy department declined to give him the necessary access to classified material. ACDA is not strong at the moment — it is about to be absorbed by the State Department — and it has not been able to stand up to DOE. The National Security Council may also have intervened to secure Kidder's exclusion.

But concern about the closed nature of the debate on stockpile stewardship does not stop with the treatment of Ray Kidder, who made his critical views on the issue known in Nature last year (Nature 386, 645; 1997). The formulation of the entire policy during the Clinton administration has been overly constrained by political concerns. No-one outside DOE has been allowed to compare stewardship — a crafty if expensive compromise solution — with alternative approaches, such as the remanufacturing option advocated by Kidder. Congress should insist on such a comparison, as it considers the administration's extravagant request of $4.5 billion for stockpile stewardship in 1999.