From your quote, it actually sounds like the author was referring to "bouncing" rather than bussing. Both procedures entail combining ("condensing" to use his term) two or more tracks into one.
The difference is that with bussing you're just giving tracks a common destination, such as a patch point, aux track or bus in SONAR. The tracks themselves remain active and untouched. With bouncing, you are also combining those tracks but you're physically writing that combination to a fresh new track, and afterward you'd normally mute the original tracks because you won't be using them anymore.
If you were a producer in the 60's, you'd likely have been writing your data to a 2-, 4- or 8-track tape recorder. Being limited in the number of discrete tracks (meaning stripes on the magnetic tape), it was impossible to store more than 2, 4 or 8 individual recordings there. Bouncing would let you combine 2 or more of those tracks and re-record the combination onto another track, freeing up the original tracks to record more stuff. That's most likely what the writer you quoted was alluding to.
Nowadays, we no longer have such severe track limitations. A SONAR project can have as many tracks as your computer can handle, and even on a modest computer that number is large. But even today we sometimes max out what our CPU can handle, so bouncing remains a useful technique - not for allowing more storage but for reducing CPU usage and disk I/O.
Merging tracks onto a bus, patch point or aux track can be similarly used to conserve CPU cycles, because multiple tracks can then share processors (e.g. reverb). More often, this technique is just a convenience. If 10 tracks share a common reverb plugin, for example, it's much easier to adjust that reverb for those tracks all at once. It can simplify mixing by allowing you to adjust multiple related instruments' levels as a group (e.g. drums). But it's different from bouncing because the result is not a new track, just a summation of existing ones.