For a while, the giant North-American telecommunications company "AT&T" promoted a technology that they called “StarLAN.” StarLAN did use existing low-cost telephone cables and avoided the major expense of rewiring office buildings. The StarLAN developers knew that the cheap, existing telephone cable couldn’t be used to reliably detect collisions, so they came up with an innovative new hardware architecture that handled each separate telephone line in point-to-point fashion so there wouldn’t be any collisions on that cheap telephone cable. All of those point-to-point telephone wire segments came together in a central location: at a new box they called a “Hub.” StarLAN was still Ethernet at its heart, and it still used a collision domain, but the hub condensed the collision segment down to just a few inches within a small box.
In spite of its innovative architecture, StarLAN just couldn’t deliver speeds faster than one million bits per second because of the cheap, old cables strung throughout our telephone system on which it was based. The major innovation of StarLAN was to get the industry thinking about collapsing the size of those collision domains. StarLAN was a commercial failure, so commercially at least, they never quite met industry expectations.
But, engineers did find a way to adapt a higher-quality wire that still looked like a telephone wire, was still inexpensive, and they could tap it into an Ethernet segment.