> because who ever expected a case where 4G was available and 3G was not
I'd expect that to be everyone that thought about it for more than a few seconds. Both because eventual replacement was obvious and because sometimes you only have partial coverage.
You may or may not be familiar with some of the use cases here in Australia. 3G coverage was more widespread and reliable for low bandwidth use cases, right up until the end. It was 4G that would have partial coverage, not 3G. 3G was often forced by IoT vendors, or routers using cellular for out of band management, due to its superior ability to penetrate through buildings. Same with ATMs, payment terminals, emergency telephones in lifts or out on the highways. People in regional areas would also force set their phones to 3G, to stop them flapping between a poor signal 4G network and the consistent, but slower, 3G network.
Many DAS/Distributed Antenna Systems, essentially networks of antennas placed inisde buildings to extend cellular coverage indoors, are costly and must be approved on a per-operator basis. Some of the DAS solutions requirements I've seen to provide full coverage for Optus, Telstra and Vodafone in 4G only in a 30 story building or medium sized mall, wanted 18 full racks, 100 amp 3 phase power connectivity, 15kw minimum redudancy cooling capacity, 8 hours of battery backup for all racks and cooling, all located in the centre of the facility to minmize cable runs. As a result, high quality DAS systems don't exist in every large building or campus environment. Even in 2024, we see buildings nearing the end of construction before the developer bothers to consider an appropriate DAS system, and at that point they balk at the space and location requirements, and refuse to understand why a DAS can't just be installed in 4RU in a crammed comms room thats the size of a domestic bathroom.
> It was 4G that would have partial coverage, not 3G.
By the time 4/5G are built out well, I'd expect 3G to win some and lose some, becoming more of the latter every year.
Unless the plan was to switch all the 3G equipment over to 4/5G simultaneously, making 4/5G become reliable overnight?
> due to its superior ability to penetrate through buildings
That's a feature of low frequencies. I don't think 3G does better anywhere if you compare the same frequency. And a bunch of the frequencies are exclusive to newer protocols, aren't they?
Not necessarily, it depends on how the network operators laid out their bands. In theory I suppose it could free up adjacent bands for more network capacity.
> a bunch of the frequencies are exclusive to newer protocols
Depends on where you are in the world. In the US the FCC has been growing out the number of available cellular bands. Cellular technologies can work on just about any frequency, given the handset supports the appropriate bands.
2G/3G tend to be a little bit more power efficient than more modern technologies as _generally_ they require less processing overhead on the handsets. However, LTE/5G do better with error correction and lower signal strength. You also get higher capacity. Stuff like GSM uses a round robin time sharing mechanism which can quickly get saturated in dense environments.
2G is _mostly_ there for legacy applications. The reason why it breaks down for this case is because a lot of older handsets negotiate 2G first before switching over to LTE, and then 5G.
I'd expect that to be everyone that thought about it for more than a few seconds. Both because eventual replacement was obvious and because sometimes you only have partial coverage.