Recently, journals like @ASNAmNat and @Ecology_Letters have made changes to their data sharing policies to "give them teeth". This will force authors to make their data openly available if they want to publish. I have seriously mixed feelings about this.

Openly-accessible data are clearly better for science. It allows others to check the authenticity of data and confirm published results. It also allows others to use the data for new projects, as well as for syntheses and meta-analyses.

So much information has been collected and then lost as hard drives fail or as scientists retire or pass away. This is a huge waste of resources. Clearly, there are major benefits to enforcing data archiving and sharing policies.

On the other hand, datasets can take enormous amounts of effort to collect. That effort, I believe, should be rewarded with opportunities to participate in research that uses those data. I don’t think it should mean automatic co-authorships, but it should bring opportunities.

This is particularly important for students and scientists in low and middle-income countries. Here, data are one of the main bargaining-tools people have to develop projects, collaborations, and advance their careers.

Unlike in more developed nations, people in these countries have fewer resources to develop highly productive careers. For these scientists, making their data completely available means that others, who might have more time or money, can use them to produce studies they cannot.

Motivations to reach out to those who generate data for collaboration are reduced dramatically, just exacerbating inequalities in how different scientists can move forward in their careers across different parts of the World.

The problem is complex and I don't have lots of answers. However, forcing people to provide all their data without any restrictions seems like an unfair solution. The same goes for making publishing extremely expensive. Only those who can will benefit.

Maybe data should be immediately available only for the purpose of checking that they are not fraudulent or that there are no problems with the analysis. Medium to long embargoes should be allowed for other uses of the data, as long as they are deposited in a safe repository?