There's also a huge tendency to design for fast, high quality connectivity. Try using any Google product on airplane wifi. Even just chat loads in minutes-to-never and frequently keels over dead, forcing an outrageously expensive reload. Docs? Good luck.
I wish software engineers cared to test in less than ideal conditions. Low speeds, intermittent connectivity, and packet loss are real.
I call this "Designed in California" like some fruity company proudly says on their devices.
For software this means designed on top of the line hardware, with fast low latency internet. TFA describes the consequences.
For hardware it means designed inside climate controlled dust free offices and cars for people with long commutes to work on straight roads where you don't have to pay much attention.
Think phones shutting down if you have a real winter. Think smart turn stalks that can't signal a left turn on a crossroads that's not at 90 degrees. Think ultra thin laptops where the keyboard is so dust sensitive it lasts 3 months if you use them outdoors. Think a focus on audiobooks and podcasts because you're stuck in traffic so much.
The last decade of my life has been a speedrun in "less than ideal conditions" for computing. CGNAT, 5mbit dsl, spotty "fixed wireless" and my latest debacle: starlink, although that seems to be getting better slowly; used to drop 15/60 seconds, now it drops more like 4/200 seconds. Constant power issues and lightning strikes - i only have 1 computer that has a working NIC, because evidently tiny power fluctuations are enough to send most chipsets into the graveyard. I had to switch to full fiber between all compute sites on my property, and a wifi backup, because copper is too risky.
Yes, and it works, too. But i have outbuildings with servers and networking gear in them and metal conduit between buildings on/underground. Voltage potentials don't care, if there's a wet extension cord or something that's a less resistive path to start flowing and some gear is on that circuit or adjacent, it'll go.
Overall switching to fiber is cheaper than aggressive lightning protection, and i moved all the network gear to a commercial UPS, and the interconnect between the "modems" and the switches is media converted to fiber for 3 feet. any time i have to run networking further than 6' or so i run fiber and put a media converter or a single gbic switch there. I'm hoping i futureproofed enough to upgrade to 10gbit in a year or so. My backup NAS has 10gbit but nothing else is connected at that speed yet.
edit: One time lightning hit a pine tree in the back of the house, and it used my dipole antenna to reach a tree 80' away, and apparently there was an extension cable near there, which went back into the house, and it went all the way around the house, to reach the telco CPE box where DSL lived. the telco box and my mains earth are roughly 1 meter apart. That surge took out my main desktop computer, a washing machine (singed the dryer where it arced between it and the washer), the toaster oven, a microwave, my NAS, and my router connected to telco. It went two different paths inside the house, along both outside walls, one via mains copper and the other via cat5e copper. That was quite an expensive misadventure.
Developers are expensive, so we give them fast connections and fast computers. Then we act shocked when modern software/web requires fast computers.
Unless it's somehow regulated that people test less than ideal conditions it won't happen, yet most people (myself included) don't really want that either.
I live in some hills and some days I need to fully drive out of them to get google maps to load the map. The map I am using half a gb to cache locally on my phone already. Whats even the point of that cache? Same thing with spotify. Why is there latency searching my downloads library in offline mode?
I wish software engineers cared to test in less than ideal conditions. Low speeds, intermittent connectivity, and packet loss are real.