They're not hacking the brain. Regex s/cyberspace/electronic and it would make a lot more sense.

Ah, typical Euro-centric thinking where they think because you live in Paris, you can just usurp free wifi because it's there. Never mind that it's on shared cable modem bandwidth and that typically bandwidth in Europe is shit. The ISM band is so saturated that you could charge your phone just through induction because to live there you're living 3-5 stacked on top of each other (in the city). So when the lights go out, so does the Wi-Fi network. If you were relying on the mobile network instead of dedicated 2-way radios, then it usually means that the mobile network will overload instantly with people calling their friends asking "are the light out where you are?" And I don't like the idea of an Emergency "switch" of any kind backdooring my router. Cisco already does it with their 'Vault' devices and the idea that I would now have to rent my router that I paid full price for from the retailer makes me want to smash the device with a hammer. Don't try relying on it outside of the city tho... it goes from city one block to farm land just like that. And it would never work here because of the suburbs.

Certainly the airline could do better but you know who else is at fault? The parents for sending their child off like luggage instead of being responsible parents. It was unfathomable 20 years ago to send a child off alone but now it's more common place. Parents need to take some responsibility in this.

Amazon is directly connected and multi-homed with several backbone carriers. It wasn't an ISP issue.

One should backup anyways. But typically you have 48 hours to know if a certain track will hit your area. Not 5 days. Weather forecasts are only good to 90% up to 3 days.

Wrong... because if your storage is on one continent and you fail over to another continent/region, suddenly your app slows to a crawl as your storage has to go over a WAN link (even if it's within Amazon's network). This leads to timeouts in queries/connections and generally bad things that will happen.

Bad service architecture... Not multi-homing to multiple servers in multiple service regions.

Not to mention that the app has to be architected to take advantage of different zones/regions.

Problem is that most companies that use these services don't load balance across the datacenters. They usually have most of their core in one given region and nothing or load balanced in others... Mostly it's because of cost having extra instances up and running... but it's also in software design and app architecture. If you don't have multiple cores in multiple regions, your site will go down when there's an outage in that cloud region. So cost wise, if you have to do this... where's your savings again? The Cloud is bullshit.