In a recent paper published in (2004 Class. Quantum Grav. 21 3803) Carlip used a vector–tensor theory of gravity to calculate the Shapiro time delay by a moving gravitational lens. He claimed that the relativistic correction of the order of v/c beyond the static part of the Shapiro delay depends on the speed of light c and, hence, the Fomalont–Kopeikin experiment is not sensitive to the speed of gravity, cg. In this comment we analyse Carlip's calculation and demonstrate that it implies a gravitodynamic (non-metric) system of units based on the principle of the constancy of the speed of gravity but it is disconnected from the practical method of measurement of astronomical distances based on the principle of the constancy of the speed of light and the SI metric (electrodynamic) system of units. Readjustment of theoretically admissible but practically unmeasurable coordinates of Carlip to the SI metric system of units used in the JPL ephemeris reveals that the velocity-dependent correction to the static part of the Shapiro time delay does depend on the speed of gravity, cg, as shown by Kopeikin (2004 Class. Quantum Grav. 21 3251). This analysis elucidates the importance of employing the metric system of units for physically meaningful interpretation of gravitational experiments.