An important contribution to the foundations of probability theory, statistics and statistical physics has been made by E. T. Jaynes. The recent publication of his collected works provides an appropriate opportunity to attempt an assessment of this contribution.

The article responds to some of the points raised by B. van Fraassen concerning probability kinematics and direct inference within the framework of the approach to the revision of probability judgment proposed by Levi in The Enterprise of Knowledge. In particular, the critical importance of the question of direct inference is emphasized and explained.