Until now, Nature Neuroscience has been a monthly journal, appearing in print and online toward the end of each month. Within the next few weeks, however, we will introduce Advance Online Publication (AOP), allowing us to publish papers on our web site before they appear in print. We believe this will be a valuable service to both authors and readers; although the Nature titles are committed to rapid publication and have a policy of avoiding long backlogs, AOP will now allow us to publish even faster. In the case of Nature Neuroscience, authors can expect their papers to appear online within less than 3 weeks of returning their corrected proofs.

New papers will be uploaded to the AOP section of our web site each Monday, but AOP will supplement, rather than replace, the monthly print publication cycle. We expect that most research articles will be published in the AOP section before they are transferred to the regular monthly issue, but the remaining items—news and views, reviews, book reviews, editorials and so forth—will continue to be published monthly, with web publication coinciding with distribution of the print journal.

Some journals publish incomplete or uncorrected proofs on their web sites, but the AOP versions of our papers will be definitive in every respect. They will be laid out in their final published form, and will include both PDF and HTML versions (the latter complete with web links). Because they appear before the print issue has been assembled, however, AOP papers will not have the traditional volume/issue/page citation information. Instead, they can be cited using a digital object identifier (DOI), a unique character string that can be used to identify any type of electronic content. DOIs can be associated with URLs (or other metadata) through a central clearing-house, allowing publishers to make automatic links to each other's electronic content. The DOI system is being widely adopted by science journals, and is expected to become the industry standard in future, supporting not only cross-linking of references but also electronic commerce and other web-based transactions. Other scientific abstracting and indexing services such as Medline and the Institute for Scientific Information (ISI) are now storing DOIs, and although they are not yet using them as the basis for searching or linking records, it seems likely that this will happen soon.

For better or worse, a large infrastructure—academic credit, impact factor calculations and so forth—has been built around the existing citation conventions, and they are unlikely to disappear overnight. We will therefore continue to use the conventional citation format for now, although we will also allow DOI-based citations. The standard may change in the future, and that would be welcome if it provides a more meaningful indication of actual publication dates.

What is the 'official' date of publication? The answer is less clear now than it was when journals were only published in print. For legal purposes such as patenting, it seems likely that the first date—that of electronic publication—will be the one that matters; authors who are filing patent applications should keep this in mind. We also expect that the AOP date will determine 'newsworthiness' in the eyes of the popular press; therefore, embargos on our press releases will be lifted at the time of electronic rather than print publication. The rules and customs for assigning academic credit for priority are less clear. Presumably authors who are anxious to scoop their competitors to important discoveries will point to their online publication date as evidence of precedence. Similarly, we shall consider online publications (but not electronic preprints) when evaluating the novelty of new submissions. But what about the citation record? If a paper is published online in December 2001 but does not appear in print until January 2002, traditional indexing services would record it as a 2002 publication. Authors are often acutely aware of this, and most editors have received calls in early December from anxious would-be contributors, asking whether there is any chance of a same-year publication date for the manuscript they are frantically preparing to submit.

Perhaps some ambiguity is not such a bad thing if it helps people to appreciate the absurdities of the traditional system. Despite the obsession with print publication dates, they are often little more than marketing ploys. Many news magazines carry a print date that is many days later than when they are distributed, thereby misleading readers into thinking they are reading up-to-the-minute news. Conversely, some science journals pull the opposite trick, distributing their monthly issues after the stated publication date in order to lull authors into thinking that they can scoop the competition. We do not expect the AOP system to dampen authors' competitive instincts, but it will at least ensure that the competition is based on publication records that are accurate and precise.