MIKE MOSS SAYS: Glenn, For many years, human observers estimated visibility based on how well a series of objects or lights of known distance could be seen, and related these distances to a reference map of the area that showed how far away these objects (towers, buildings, hilltops, bridges, etc) were known to be. Most airports now rely on automated weather systems that include a "forward scatter sensor." This instrument produces pulsed flashes of blue light, some of which is scattered at an angle toward a nearby detector. Visibility is estimated from the intensity of scattered light, which increases when there are more fog droplets, smoke or haze particles, raindrops or snowflakes in the beam. These sensors report a visibility based on one minute samples averaged over the past ten minutes leading up to each observation.

As it relates to your question then, the current method of obtaining visibilities for most locations is based on the "particle measurement" principle, while in the past (and in a supplementary role at a few present locations, mainly military bases) it was based more on the "actual visual distance" as estimated using a pair of eyeballs.