Current norms around hypothesis testing combine Fisher’s paradigm and Neyman-Pearson paradigm.
Fisher was focused on evidence and introduced the p-value as a measure of evidence against the null. Neyman and Pearson, on the other hand, were focused on decision-making. Current practice is a combination of these two perspectives. We make decisions and report the evidence. For instance, we reject the null and report the p-value.
This is odd, as observing compared to is not decision-relevant in the Neyman-Pearson setting. Or, at least, it should not be decision-relevant. You should not do anything different upon observing a smaller or larger p-value, as long as both are less than your predetermined significance level, .
But intuitively, a smaller p-value is more evidence against the null, and it seems like that should be useful in some way. Part of the draw of replacing p-values with e-values is that more extreme e-values can be used to make more extreme decisions (see post-hoc hypothesis testing and e-values enable post-hoc hypothesis testing).