Hello,
You actually have a very valid and wise question, and a one that's not entirely easy to answer.
When Ethernet was developed, it was a LAN technology only, inteded to be used inside a single building. The transmission properties of the physical wiring for a LAN technology were much more tightly specified, and they were also much easier to maintain, compared to telco lines. Whether you were using Ethernet with the thick coax, thin coax, or TP cabling, there were strict rules on the length and the electrical properties of the cabling (think TP cable categories), and Ethernet cables were manufactured to comply with these rules. As long as your cabling was installed according to valid norms and standards, your entire LAN system complied with the Ethernet technical specifications, and so it was guaranteed to work. Of course, it would not hurt if Ethernet had some kind of a loop test, but in most cases, because the entire LAN cabling was relatively small in its size and readily accessible in a building, it was also fairly easy to diagnose if a problem appeared.
If you think how telco cables are installed and run then you'll immediately see how much more difficult it is for a local telco loop to keep its properties: Running out there, either in free air on poles, or underground, with very variable length, through connection boards that oxidize over time, enduring the heat in summer, cold in winter, air moisture, etc. Telco cabling is much more unpredictable over time, and so it was helpful to have a method for functional testing.
Moving to more technically involved nature of the issue, a loop test is basically performed by short-circuiting the Transmit and Receive leads in a cable, so that a device sending a test data pattern can hear it back practically in the same time. Note, however, that Ethernet was originally conceived as a half-duplex technology: A device can either send or receive data but not both at the same time. A loop test requires exactly that: Listening to your own data at the same time you are sending it. For original Ethernet, however, having the bus wires in the coax, or the Tx/Rx pair in the TP short-circuited would result into a permanent collision. Simply put, with half-duplex Ethernet, loop test with a device sending data and receiving it back simultaneously is not technically possible.
Also, loop tests are used to verify connectivity over a point-to-point physical medium. However, early Ethernet designs were based on a bus topology, and with a bus, loop tests do not make much sense because if you short-circuit the bus then it affects all connected stations, not just a single host.
With TP cabling and newer Ethernet variants that support full-duplex operation, doing loop test would be actually possible. This would make most sense when Ethernet is used as a WAN access technology, because again, isolating a fault on a LAN is done fairly easily even without loopback testing. Well, you might be interested that there is a big family of protocols and technologies called the Ethernet OAM (Operation, Administration, Management) that was created exactly because Ethernet as a former LAN technology was missing some of the supervisory features needed for MAN/WAN operators to maintain their networks efficiently. One of the sub-areas of Ethernet OAM is the Connectivity Fault Management (CFM), and within CFM, there is a Loopback test that effectively works similar to a ping but provided by pure Ethernet means - you send test frames to a defined test point, and the test point returns the frames back to you.
So in fact, there is a loopback test in Ethernet, even though it is implemented mostly on devices that run Ethernet as a WAN access layer (Metro Ethernet, Carrier Ethernet).
I hope this helps a little.
Best regards,
Peter
