The 1970s "old joke" referred to above was told to this poster in the NASA / Jet Propulsion Lab cafeteria in about 1975-1976. He worked in the Digital Maintenance group in the JPL Space Flight Operations Center from 1974 to 1978.
The story / joke was a classic regularly used at JPL to explain ping time, and differentiate bandwidth from latency (and, by the way, the need to document where your cables ran, and that you needed to distribute your data circuits across multiple cables in different trenches - or somehow via multiple paths).
The NASA Deep Space Network tracking station at Goldstone is just outside of Fort Irwin, just east of Barstow, California. When you leave the highway you have to go through Fort Irwin to get to any of the Goldstone facilities. Depending on the highway route taken, and which Deep Space Network dish at Goldstone you are driving to (or starting from) it was about 160-185 miles (255-298 km) from JPL. At freeway speeds (65 mph, about 100 km/h) it was a minimum of three-and-a-half hours, usually four, and frequently more, depending on the traffic. If you ignored the speed limit while out in the desert (risky) you could get closer to three and-a-half hours. This distance and speed also explained how the "ping time" was 7 to 8 hours. Several of the freeways now in existence were not there then.
At the time (early 1970s), the data links from JPL to Goldstone ranged from as low as 1200 and 2400 bps (several of each) to 9600 bps (one or two). The 9-track magnetic tapes of the day recorded at a maximum density of 6250 bits per inch (but some older drives were limited to 800 or 1600 bits per inch). The tape reels were made in different sizes, the largest held about 2400 feet of tape, but due to the data being written in records, with gaps between the records, the maximum data capacity of a 2400 foot reel, blocked at 32,767 bytes per record and recorded at 6250 BPI was 170 megabytes per reel.
As the story that your contributor heard went, one day a plumbing contractor's backhoe dug up and broke the underground cable that carried ALL of the JPL-to-Goldstone data and voice lines through Fort Irwin, and it would take at least a day, maybe longer, to repair. So someone was designated to drive two boxes of 12 reels each of magnetic tape down to JPL, and quickly. The first available vehicle was a white NASA station wagon. Hence the punch line: "Never underestimate the bandwidth of a station wagon full of magnetic tapes hurtling down the highway".
Rounding off the numbers, twenty-four reels of tape at 170 megabytes each is 4080 megabytes. Three and a half hours is 210 minutes. 4080 megabytes divided by 210 works out to about 19.4 megabytes per minute, or 32.3 kilobytes per second (258.4kilobits per second) - over 100 times faster than a 2400 bps data circuit of the time. Note that the incident above involved only 24 reels - which didn't come anywhere near filling the station wagon, in fact the two boxes of tapes didn't even fill the front passenger seat. (as an aside, a station wagon is known as an estate car or estate in other parts of the world). Incidentally, that conversation was the first time your contributor ever heard the term backhoe fade used to describe accidental massive damage to an underground cable (compare it to the term rain fade used to describe a fade-out of a point-to-point microwave radio path due to the absorptive effect of water in the air).
Great story!
Ah yes, the early days of networking. We used the term sneakernet to describe how it was faster to carry a floppy disk across the room or building than it was to send the same data over the internal network.