Nico Formánek

The relationship between computing power and model selection, following a scenario suggested by Humphreys (and others), is investigated. Against the worrisome conclusions of this scenario it is argued that using computing power to tune models has not been the only goal of science so far. Other reasons for model selection were always in play. It is discussed if modern computing methods, supplied with contemporary compute, make pursuing model selection on the grounds of error reduction alone more appealing. The conclusion is largely negative.

Trust is a central pillar of the scientific enterprise. Much work in the philosophy of science can be seen as coping with the problem of establishing trust in a certain theory, a certain model, or even science as a whole. However, trust in science is threatened by various developments. With the advent of more complex models and the increasing usage of computer methods such as machine learning and computer simulation, it seems increasingly challenging to establish trust in science. How and on what basis can an appropriate trust in science be built? We are interested in how trust is established in such cases of increasing complexity (of models and communication) and what could be appropriate measures to alleviate doubt.

Trust is a central pillar of the scientific enterprise. Much work in the philosophy of science can be seen as coping with the problem of establishing trust in a certain theory, a certain model, or even science as a whole. However, trust in science is threatened by various developments. With the advent of more complex models and the increasing usage of computer methods such as machine learning and computer simulation, it seems increasingly challenging to establish trust in science. How and on what basis can an appropriate trust in science be built? We are interested in how trust is established in such cases of increasing complexity (of models and communication) and what could be appropriate measures to alleviate doubt.

Several philosophical issues in connection with computer simulations rely on the assumption that results of simulations are trustworthy. Examples of these include the debate on the experimental role of computer simulations :483–496, 2009; Morrison in Philos Stud 143:33–57, 2009), the nature of computer data Computer simulations and the changing face of scientific experimentation, Cambridge Scholars Publishing, Barcelona, 2013; Humphreys, in: Durán, Arnold Computer simulations and the changing face of scientific experimentation, Cambridge Scholars Publishing, Barcelona, 2013), and the explanatory power of computer simulations :277–292, 2008; Durán in Int Stud Philos Sci 31:27–45, 2017). The aim of this article is to show that these authors are right in assuming that results of computer simulations are to be trusted when computer simulations are reliable processes. After a short reconstruction of the problem of epistemic opacity, the article elaborates extensively on computational reliabilism, a specified form of process reliabilism with computer simulations located at the center. The article ends with a discussion of four sources for computational reliabilism, namely, verification and validation, robustness analysis for computer simulations, a history of successful implementations, and the role of expert knowledge in simulations.

Philosophy of computer simulation has for long tried to answer the question: What makes computer simulation special? Answers normally proceed by giving a demarcation criterion, a property which only holds for computer simulation. The definition by demarcation has also been attempted in different fields, maybe most famously by Popper for science. So it might not come as a surprise that philosophy of mathematics has tried to demarcate computer aided proof. What is remarkable is that the demarcation criteria for computer simulation and computer aided proof are rather similar. As the debate in philosophy of mathematics is more advanced, I will show how its arguments can be translated for the philosophy of computer simulation. I argue that even in the light of computer methods human justificatory capabilities still remain central. In closing I point out, that the focus on justificatory methods like verification/validation in software engineering challenges the traditional philosophical account of engineering.