Here's my favorite example of this:
https://pgealerts.alerts.pge.com/outage-tools/outage-map/
When the power's out and broadband is down, if you are lucky the cell network is still up. However, everyone fails over to it simultaneously, so there's no way you'll get 2G speeds out of your 5G plan with 5 bars.
I throttled my browser dev tools to 2G and clicked reload. After 2+ minutes, it popped up a "It looks like you are on a slow connection" modal (sometimes this loads on top of the outage info, obscuring the data you want!), so I clicked "Use low bandwidth version". After 51 seconds, the debugger says page load is finished (for the low bandwidth one), but the page is just a white background, and the browser "loading" animation is still running.
After 2.17 more minutes, I get a form where you can type in an address + a bunch of irrelevant info. I typed an address, waited a minute, then typed this paragraph. After 1.5 minutes, it showed my address in the autocomplete menu, and I clicked it.
33 seconds later, it started loading a google maps ajax. 47 seconds after that, it displayed a header, with a white body. 1.33 minutes later, ignoring styling, it returned the string "power is on".
(No map, etc, at this point.)
Had this been a real outage in a storm, I'd be standing outside in the rain, or in a situation where getting cell coverage for more than 30 seconds at a time is impossible if you're holding the phone. I usually just put the phone on a carefully placed chair, then back away slowly.
Anyway it takes 9 minutes under ideal scenarios (no drops) to send 30 bytes of address, and an 11 byte response. 44 bytes / 540 seconds = 0.08 bytes per second good put.
For reference, voyager can send 20 bytes per second, so it's 250x faster than this. Morse code / telegraphs are typically sent at 1.5 words per minute, where a word is 15 characters, so 0.375 bytes per second = 4.6x faster.
Remember, these are simulated conditions for the PG&E site. My record best time getting it to load during a major outage is 45 minutes, not 9.
