A measure of predictive sharpness for probabilistic models (2025)
Here is a recent preprint on predictive sharpness. In this article, I propose a unified method for quantifying predictive sharpness that applies to deterministic and probabilistic predictions and also extends to multidimensional cases.

arXiv link (https://doi.org/10.48550/arXiv.2509.03309)
code (https://github.com/psyrjane/Predictive-sharpness)

Dimensions of predictive success (2024)
Consider a theory or a model that makes predictions. What is it for the theory/model to be predictively successful? In this paper, I sketch answers to this question by distinguishing between various dimensions of predictive success. Logical dimensions of predictive success include accuracy, specificity, scope, and depth. Contingent dimensions of predictive success include forecasting, use-novelty, theoretical novelty, and temporal novelty. I analyze interrelations between the dimensions, as well as their relevance to theory confirmation and theory development.

Syrjänen, P. (forthcoming). Dimensions of Predictive Success. The British Journal for the Philosophy of Science. https://doi.org/10.1086/733193.

Accepted draft (pdf)

Novel prediction and the problem of low-quality accommodation (2023)
Novel prediction is often thought to have special value in theory confirmation. One argument in favor of this view is that if theorists do not make novel predictions but rather use evidence that they already know (i.e. they ‘accommodate’ evidence), the theory building process may become more unreliable. In effect, when the evidence is already known, theorists have an opportunity to use various questionable methods to ensure that a theory that fits the evidence is produced. I investigate the idea that the problems of accommodation result in an advantage of novel prediction over accommodation in light of scientific practice (including Questionable Research Practices, preregistration) and reach the conclusion that despite its intuitive appeal, it may not hold water in the sciences. Novel prediction also has certain disadvantages, and accommodation provides advantages that are often overlooked.

Syrjänen, P. (2023). Novel prediction and the problem of low-quality accommodation. Synthese, 202, 182. https://doi.org/10.1007/s11229-023-04400-2.

The epistemic role of prediction in science (dissertation) (2022)
My dissertation examined the prediction versus accommodation problem in the philosophy of science. The dissertation contains discussion on the history of the prediction versus accommodation debate, evaluates arguments both in favor and opposed to predictivism, critiques predictivist arguments in favor of scientific realism, and develops a logical approach to predictivism. The dissertation can be found here (pdf).