Everybody is talking about data. Experimental scientists live and breathe data. Theorists are challenged by data. Funders are wondering how to make the data produced with their support more accessible without stretching their budgets. Research communities are seeking new data repositories, and standards to support them. And scientific publishers are wondering how to host data and provide quality control.

Scientific Data is a new journal, launched by Nature’s publishers this week, that will help to address some of these challenges. By publishing formal descriptions of data sets — Data Descriptors, the publication’s main article type — it will render the data more visible and give originators explicit credit for those data, rather than for the papers that use them. The journal is peer-reviewed and online-only. Authors pay a charge on publication: this ensures that the final, published versions of their contributions to the journal are immediately freely accessible to all. The content is licensed under one of three Creative Commons licences, and machine-readable metadata are released with every article to maximize reuse.

To quote Scientific Data’s launch editorial: “The question is no longer whether research data should be shared, but how to make effective data sharing a common and well-rewarded part of research culture”. When it is feasible to do so, many journals, including all those in the Nature family, have long insisted that data are deposited in repositories where available, before publication. For other areas of research, we at Nature have significantly increased the figure limits in our papers. In Nature Protocols, there is a place for more-specific methods descriptions than is conventional in scientific papers.

Now, in Scientific Data, there is space for researchers to formally describe a data set and the techniques used to derive it, and to refer readers to research papers that have already incorporated the data.

Crucially, the journal’s descriptors, being peer-reviewed and citable, provide a way to assign credit to the originators of reusable data sets. In other words, the delivery and sharing of data becomes as credit-worthy, in principle, as publishing conventional research papers. It is important that the assessment of research and reward of researchers does more justice to this essential component of science.

The journal’s first publications include articles describing previously unpublished data sets — demonstrations that Scientific Data can help to motivate scientists to share valuable data. The journal’s editors highlight work by Zengchao Hao and colleagues detailing data sets that track drought around the world (Z. Haoet al.Sci.Datahttp://doi.org/sww;2014). Using the Data Descriptor, anyone can download the data, generate their own maps (past or future) for any area of the world and even use the authors’ source code to recalculate the drought metrics.

Another article, by Graham Edgar and Rick Stuart-Smith, provides an example of a Data Descriptor that builds on previous publications (G. J.EdgarandR. D.Stuart-SmithSci.Datahttp://doi.org/sxv;2014). It is based around the data produced by the Reef Life Survey, a citizen-science project that uses volunteer divers to help to survey biodiversity on the world’s reefs. Analyses of these data, which are relevant to our understanding of reef ecology and to conservation, have been published in a number of research papers. The data are given in full in the Data Descriptor, along with the authors’ descriptions of the survey procedures and data standardization — crucial information for other scientists interested in using these data.

Beyond its significance for data buffs, the journal is a further step in Nature Publishing Group’s drive to enhance research reproducibility. The more researchers take steps to make their data available and discoverable, the more a core principle of science — that others can replicate the work — can be fulfilled, in an era in which such replication is often beset by obstacles. For that reason alone, we at Nature welcome Scientific Data.