To the Editor — It is obvious that data transparency and the availability of raw experimental data must be the future standard practice in the experimental sciences. While as an experimentalist I am barely affected by it, I do have reservations against forcing our colleagues in the numerical theoretical sciences to make all code available publicly, a push mainly enforced by funding agencies. This treats the experimental and theoretical sciences on an unequal footing, and in the long run may hurt the advancement of numerical science.

It is common practice to develop experimental capabilities in a laboratory and perfect it over many years. The effort to achieve superconductivity of sulfur hydride under extreme pressures is a good example of this. Despite the open publication of the design of the apparatus, typically the know-how gives those who develop it a competitive advantage to drive the science.

The situation in the theoretical sciences is completely analogous. Hard problems are tackled and advances in concepts, algorithms and computational means expand our understanding of physical systems. Yet, simply implementing policies enforcing such a ‘digital apparatus’ to be made available immediately will be detrimental to the development efforts, jeopardize the careers of young scientists developing such code, and most importantly will miss the opportunity to encourage code sharing in a way that is constructive for those who wrote it.

We as a community have to shape the cultural change towards completely open data and code. Critically, we have to redefine the way how credit is assigned for developmental efforts. Instead of forced publication, I believe that positive incentives are a more proper way to fairly ensure the sharing of code. Part of this lies with the prominent publishing houses. Will the Nature journals publish prominently released code on its own, separate from the scientific discoveries made by it?