Impositiion of selection and opportunity for censorship meust be regarded as aberrations of a communication system for science. Future historians might wonder why these faults evinced so little concern. Because editorial decisiions pre-empt scientific debate, editors and their advisers assume a heavy responsibility for nurturing fresh conjectures and for maintaining unbiased speedy communication. Evidently this responsibility has not always been honoured.Available evidence of inappropriate rejection confirms the expectable, if not adequately anticipated, tendency for reviewers to oppose innovation; so that although (...) the statistics of rejection might seem acceptable, there is an unfortunate selectivity working against new ideas. It follows that the perceptive and inventive must work harder to be heard. This is quite the wrong bias: the ingenious should be encouraged, not opposed.The strategy of ‘peer review’ is attractive to publishers and disbursers of funds because it is cheap and practical; but there are fundamental weaknesses in the notion of peer review: peer status is illusory, preception is fickle, and impartiality is unnatural. Peer opinion is not, ipso facto, scientific argument.Unchecked whimsical censorship, which is most likely to affect the more perceptive and innovative offerings, disregards the fundamental uncertainty of scientific knowledge and denies the value of general debate. ‘Peer’ review should not usurp that debate, but should rather act as a strictly preliminary filter atuned to acceptance of fresh ideas.The psychology of anonymous ‘peer’ review is inimical to science. Authority automatically accorded to reviewers' comment and defence of ‘reputation’ are antitheses of scientific method. Proper functioning of review as a preliminary filter depends entirely on the good sense and integrity of the few people involved. Unfortunately there is no guarantee that criticism will be rational, relevant and refutable. The accidental authoeity of the uncertain review process encourages obstruction and fraud.In order to safeguard communication, the burden of proof in refereeing should be on the criticisers. Disagreement is no reason for suppression. Reviewers' objections, which should be factual and specific, should always be tested. Failure to ebserve these elementary principles of scientific assessment, by admitting anonymous unsupported opinion as valid criticism, has caused and is causing constraint on communication and on exposure of new ideas. (shrink)