An analogy can be applied to scientific publications. As with individual bidders in an auction, the average result from multiple studies yields a reasonable estimate of a “true” relationship. However, the more extreme, spectacular results (the largest treatment effects, the strongest associations, or the most unusually novel and exciting biological stories) may be preferentially published. Journals serve as intermediaries and may suffer minimal immediate consequences for errors of over- or mis-estimation, but it is the consumers of these laboratory and clinical results (other expert scientists; trainees choosing fields of endeavour; physicians and their patients; funding agencies; the media) who are “cursed” if these results are severely exaggerated—overvalued and unrepresentative of the true outcomes of many similar experiments. For example, initial clinical studies are often unrepresentative and misleading. An empirical evaluation of the 49 most-cited papers on the effectiveness of medical interventions, published in highly visible journals in 1990–2004, showed that a quarter of the randomised trials and five of six non-randomised studies had already been contradicted or found to have been exaggerated by 2005 [9]. The delay between the reporting of an initial positive study and subsequent publication of concurrently performed but negative results is measured in years [10,11]. An important role of systematic reviews may be to correct the inflated effects present in the initial studies published in famous journals [12], but this process may be similarly prolonged and even systematic reviews may perpetuate inflated results [13,14].

More alarming is the general paucity in the literature of negative data. In some fields, almost all published studies show formally significant results so that statistical significance no longer appears discriminating [15,16]. Discovering selective reporting is not easy, but the implications are dire, as in the “hidden” results for antidepressant trials [17]: in a recent paper, it was shown that while almost all trials with “positive” results on antidepressants had been published, trials with “negative” results submitted to the US Food and Drug Administration, with few exceptions, remained either unpublished or were published with the results presented so that they would appear “positive” [17]. Negative or contradictory data may be discussed at conferences or among colleagues, but surface more publicly only when dominant paradigms are replaced. Sometimes, negative data do appear in refutation of prominent claims. In the “Proteus phenomenon”, an extreme result reported in the first published study is followed by an extreme opposite result; this sequence may cast doubt on the significance, meaning, or validity of any of the results [18]. Several factors may predict irreproducibility (small effects, small studies, “hot” fields, strong interests, large databases, flexible statistics) [19], but claiming that a specific study is wrong is a difficult, charged decision.

Constriction on the demand side is further exaggerated by the disproportionate prominence of a very few journals. Moreover, these journals strive to attract specific papers, such as influential trials that generate publicity and profitable reprint sales. This “winner-take-all” reward structure [24] leaves very little space for “successful publication” for the vast majority of scientific work and further exaggerates the winner's curse. The acceptance rate decreases by 5.3% with doubling of circulation, and circulation rates differ by over 1,000-fold among 114 journals publishing clinical research [25]. For most published papers, “publication” often just signifies “final registration into oblivion”. Besides print circulation, in theory online journals should be readily visible, especially if open access. However, perhaps unjustifiably, most articles published in online journals remain rarely accessed. Only 73 of the many thousands of articles ever published by the 187 BMC-affiliated journals had over 10,000 accesses through their journal Web sites in the last year [26].

Impact factors are widely adopted as criteria for success, despite whatever qualms have been expressed [27–32]. They powerfully discriminate against submission to most journals, restricting acceptable outlets for publication. “Gaming” of impact factors is explicit. Editors make estimates of likely citations for submitted articles to gauge their interest in publication. The citation game [33,34] has created distinct hierarchical relationships among journals in different fields. In scientific fields with many citations, very few leading journals concentrate the top-cited work [35]: in each of the seven large fields to which the life sciences are divided by ISI Essential Indicators (each including several hundreds of journals), six journals account for 68%–94% of the 100 most-cited articles in the last decade (Clinical Medicine 83%, Immunology 94%, Biochemistry and Biology 68%, Molecular Biology and Genetics 85%, Neurosciences 72%, Microbiology 76%, Pharmacology/Toxicology 72%). The scientific publishing industry is used for career advancement [36]: publication in specific journals provides scientists with a status signal. As with other luxury items intentionally kept in short supply, there is a motivation to restrict access [37,38].

Manuscripts are assessed with a fundamentally negative bias: how they may best be rejected to promote the presumed selectivity of the journal. Journals closely track and advertise their low acceptance rates, equating these with rigorous review: “Nature has space to publish only 10% or so of the 170 papers submitted each week, hence its selection criteria are rigorous”—even though it admits that peer review has a secondary role: “the judgement about which papers will interest a broad readership is made by Nature's editors, not its referees” [43]. Science also equates “high standards of peer review and editorial quality” with the fact that “of the more than 12,000 top-notch scientific manuscripts that the journal sees each year, less than 8% are accepted for publication” [44].

The publication system may operate differently in different fields. For example, for drug trials, journal operations may be dominated by the interests of larger markets: the high volume of transactions involved extends well beyond the small circle of scientific valuations and interests. In other fields where no additional markets are involved (the history of science is perhaps one extreme example), the situation of course may be different. The question to be examined is whether published data may be more representative (and more unbiased) depending on factors such as the ratio of journal outlets to amount of data generated, the relative valuation of specialty journals, career consequences of publication, and accessibility of primary data to the reader.

Some may accept the current publication system as the ideal culmination of an evolutionary process. However, this order is hardly divinely inspired; additionally, the larger environment has changed over time. Can digital publication alleviate wasteful efforts of repetitive submission, review, revision, and resubmission? Preferred publication of negative over positive results has been suggested, with print publication favoured for all negative data (as more likely to be true) and for only a minority of the positive results that have demonstrated consistency and reproducibility [54]. To exorcise the winner's curse, the quality of experiments rather than the seemingly dramatic results in a minority of them would be the focus of review, but is this feasible in the current reality?

There are limitations to our analysis. Compared with the importance of the problem, there is a relative paucity of empirical observations on the process of scientific publication. The winner's curse is fundamental to our thesis, but there is active debate among economists whether it inhibits real environments or is more of a theoretical phenomenon [1,8]. Do senior investigators make the same adjustments on a high-profile paper's value as do experienced traders on prices? Is herding an appropriate model for scientific publication? Can we correlate the site of publication with the long-term value of individual scientific work and of whole areas of investigation? These questions remain open to analysis and experiment.