Bayesian model selection is a popular method for comparing different models. The choice of prior can greatly impact the results, even with vague priors being informative for model selection. Improper priors can lead to undetermined results. The study discusses how prior sensitivity affects model selection and suggests practical solutions for designing objective priors. The connection between marginal likelihood and information criteria is also explored. The researchers provide numerical examples, including a real-world application in exoplanet detection.