For a computer scientist, open science is almost a given. For years, computer scientists have published datasets and source code of prototypes. Usually to challenge colleagues to do even better. A good example within data science are the so-called Kaggle competitions, where data is shared in abundance.

Sharing data is not really encouraged at TU/e. Instead, it is made increasingly difficult by mandatory data management plans and required cooperation agreements between universities when data leaves TU/e. Before such an agreement is reached, a PhD student is often already a doctor.

Society demands that science becomes increasingly open. Partly because public money should lead to public results, but partly also to guarantee fundamental principles of verifiability and reproducibility. Because that is one essence of open science: You share your results for free, publicly via open access journals, and at the same time you enable colleagues to validate those results by sharing your data and your tools.

Publication costs

But open science is not free. Sites, like Kaggle, have to be maintained, and with open access journals, the publisher's support organization is paid for by the author rather than by the collective readers. The exact publication costs (Article Processing Charges or APCs) vary widely, but count on a few thousand euros per article, and this is where the shoe pinches.

However, the TU/e does not reimburse these APCs, while subscriptions to journals are paid for. So I plead for refunding the APCs and not the subscriptions. If someone still wants a subscription, it should be paid from his or her own budget.

Boudewijn van Dongen is a professor of Process Analytics at TU/e. The views expressed in this column are his own.