The fact they used a 35m radio telescope as the receiver station here, and transmission powers way over legal limits for normal LoRa-like applications is a pretty good hint. You lose a crazy amount of signal with moonbounce, the effort is always easier to put into different methods instead. AFAIK it's only non-experimental use predates human-made satellites (US military, as a further channel to reach far-away war ships even if terrestrial propagation conditions are bad). If you can't use satellites, and can't have a receiver station within a few thousand miles of your transmitter, then it maybe is sort-of an option, but that's not really a situation you have anymore.

Amateur radio operators still do it, and with purpose-made codecs and really slow speeds it's not that out of reach (i.e. at least for those you don't need a radio telescope dish), but that's just because it's an interesting challenge.

Background noise - that's one way to put it! The path loss is around 250 dB, so a practical amateur moonbounce station makes use of a very powerful RF amplifier feeding an array of yagi antennas, all mounted on an altitude/azimuth rotator - in other words, it's a radio telescope, and the receiving side hopefully has something similar. You can do a google image search for "moonbounce," the setups are kind of amazing. A lot of investment mostly to exchange morse and specialized low-bandwidth digital modes.

The upshot is that you can communicate with any station that can also see the moon - somebody on the other side of the planet, 12,000km away. Plus there is the unique experience of hearing your own transmission echo, two seconds after you cease.

This kind of setup is well within reach for a commercial enterprise but will never be practical for IoT owing to the laws of physics. On the other hand, there is SNOTEL, a system that bounces snowpack telemetry off of the ionized trails left by meteors...

Because the moon is 238,900 mi away. Nor does it stay locked in position. Nor is it perfectly reflective (albeit quite a high albedo) nor a perfect sphere.

Path loss.