Learn more physics!
They explain "Notice that there are fewer water molecules on the liquid side because the some of the water has been replaced by salt."
Then they elaborate: "It is important to realize that freezing point depression occurs because the concentration of water molecules in a solution is less than the concentration in pure water. The nature of the solute doesn't matter. One might expect from the diagram above that solutes with large molecules are better at blocking water molecules travelling towards the surface of the ice. The hypothesis that solutes with large molecules cause a larger freezing point depression than those with smaller molecules is not in accord with experimental data! The misconception arises because the diagram can't be drawn to scale; the size of the molecules is very small compared to the distance between them."
They elaborate the same theme further here:
Even for enormous solute molecules, which displace huge numbers of water molecules, the effect is proportional only to the number of solute molecules per volume, Even for the many solutes (MgSO4, CuSO4, NaOH, LiOH, Na2CO3...) which increase the concentration of water, the effect is still to depress the freezing point, opposite to what you would expect from their explanation. So their explanation is wrong. The freezing point depression is, for small solute concentrations, entirely due to a single factor. When a water molecule leaves the liquid, it takes away some space that had been available for the solute to run around in. That reduces the solute entropy. So the effect tends to hold the water molecules in the liquid. All the other terms that are talked about on that site (and, truth be told, in some of our own course material here) have no effect at low concentrations. Those other contributions can enhance or reduce the freezing point depression at higher concentrations, depending on details.
(published on 05/26/11)
Follow-up on this answer.