A research data sharing game

Abstract

7While reusing research data has evident benefits for the scientific community as a whole, 8 decisions to archive and share these data are primarily made by individual researchers. For 9 individuals, it is less obvious that the benefits of sharing data outweigh the associated costs, 10 for example time and money. In this sense the problem of data sharing resembles a typical 11 game in interactive decision theory, more commonly known as game theory. Within this 12 framework we analyse how measures to promote sharing and reuse of research data affect 13 individuals who do and do not share data. We find that the scientific community can benefit 14 from top-down policies to enhance sharing data even when the act of sharing itself implies a 15 cost. Namely, if (almost) everyone shares, many individuals receive benefits, as datasets in 16 our model can be reused to achieve a higher efficiency (i.e. more publications, higher quality 17 papers). Surprisingly, as sharing implies a cost, even sharing individuals themselves in a 18 community in which sharing is common can gain a higher efficiency than individuals who do 19 not share in a community in which sharing is not common. In addition to these findings, we 20 find that measures to ensure better data retrieval and quality can compensate for sharing 21 costs by further enabling reuse. Nevertheless, an individual researcher who decides not to 22 share omits the costs of sharing. Assuming that the natural tendency will be to use a strategy 23 that will lead to maximisation of individual efficiency, we see the average scientific 24 community efficiency in our model steadily drop as more individuals decide not to share. 25 With this in mind, we conclude that the key to motivate the researcher to share data lies in 26 reducing the costs associated with sharing, or even better, turning it into a benefit. 27 28

Simplified Methodology for Fatigue Analysis of Reinforced Asphalt Systems

“…If fatigue resistance is selected as the performance based on which the effectiveness of the network is assessed, the methodology and established criteria used to analyze the test results can have a signi cant impact on the outcomes. There are several criterions described in the literature to accomplish that: stiffness reduction criterion (classical criterion of 50% [32] or even 90% reduction [33]), phase angle criterion [34] and dissipated energy criterions [35][36][37]. Similarly to the case of conventional (unreinforced) testing samples [38], the application of different criteria on the same dataset can result in even more varied fatigue laws when the asphalt samples incorporate a grid reinforcement [39].…”