“…Leaning on a type of 'Kolmogorov Extension Theorem' for convex expectations [9,Theorems 4.6 and 5.6], he shows that any 'convex Q-operator' corresponds to a 'convex Markov chain', which is a convex expectation on the bounded measurable variables with respect to the product 𝜎-algebra generated by the canonical process. However, both of these frameworks have crucial shortcomings: that of Škulj [1] and Krak et al [2] only deals with lower and upper expectations of variables that depend on the state of the system at a single time point or at a finite number of time points, respectively, while that of Nendel [3] only deals with bounded variables that are measurable with respect to the product 𝜎-algebra. For applications, this implies that for both of these frameworks, key inferences like (lower and upper) until probabilities, expected temporal averages, expected occupancy times, expected hitting times -also called expected first-passage times -and the expected number of (selected) jumps are not included in the domain.…”