First, there are not that many games of this type, but the user will see this as a "game". Most games don't use any bandwidth, and network games are often fairly lightweight in bandwidth, while some use a lot more. I doubt most games use anywhere close to the bandwidth used by modern bloated websites. If this game is different from the usual expectations of a game, it should include a warning and estimation of it's bandwidth usage.
> browsing facebook mobile
Why are you comparing playing a game to browsing a website? These are entirely unrelated activities with different expectations.
...that said, Pokémon Go could sure use a lot more caching. It seems that (didn't confirm with a packet sniffer, but latency says it all), for example, it redownloads all Pokéstop data whenever you visit it. That's IMO utterly ridiculous - a Pokéstop is pretty much a permanent fixture (an ex-Ingress portal), it doesn't change, and the user itself is expected to mostly stay in the same geographical area. I don't see any good reason not to cache at least the visual stuff.
Which brings me to the general problem of the hipstery approach to development - download and redownload everything from Internet ad infinitum, as if everyone had broadband and electricity was free.
Most of the people playing Pokemon have probably never heard of Ingress.
I agree that it could use a lot more caching, though!
I am comparing it to Facebook because that is what the linked document in question does.
I suspect you are being overly pedantic as a way to avoid my core point, which is that this application doesn't consume much/any more data per hour than most other activities people use a smart phone for. Even the linked document shows that.
The trouble seems to stem from the amount of time people are using the app, which is above the average for other apps.