The pattern of technology failures over the last several years has been striking. Cellular networks have crashed during emergencies. Payment systems have gone down nationwide for hours. GPS has produced dangerous routing errors that put drivers into hazardous situations. Cloud services that millions of businesses depend on have had multi-region outages that knocked operations offline for an entire day. The lesson isn’t that technology is bad — it’s that the redundancy practices households used to maintain by default have eroded, and the failures hit hardest when the technology was the only fallback left.
Recent high-impact failures
The 2024 AT&T nationwide outage left millions of customers without cellular service for hours, including the inability to call 911 in many cases. The 2024 CrowdStrike incident took down Windows machines worldwide overnight, grounding airlines, halting hospital systems, and disabling business operations in dozens of countries. Multiple AWS regional outages have intermittently disabled large portions of the consumer internet over the years. The Optus outage in Australia took down a major carrier for nearly half a day. None of these were the result of attacks; all of them were ordinary infrastructure failures at industrial scale.
The redundancy that used to be automatic
A generation ago, households kept paper maps, paper address books, landline phones, cash, manual can openers, mechanical clocks, and physical bills with account numbers on them. Most of those redundancies have been lost without being consciously discarded. When the network goes down, the modern household often has no offline navigation, no way to look up a phone number, no functioning phone if the modem is down, no payment method for cash-only transactions, and no easy way to access account information that lives only in apps. The old redundancies weren’t held for paranoid reasons — they were just the default — and their disappearance means a meaningful percentage of basic functions are now single-point-of-failure on technology.
What reasonable redundancy looks like today
The realistic playbook isn’t to revert to 1985. It’s to reintroduce a few specific redundancies that recent failures have shown to matter. Some cash on hand for the day a payment system stops working. A printed list of important phone numbers. Offline maps downloaded for areas you frequently visit. Account numbers and login info for critical services backed up somewhere accessible without an internet connection. A flashlight that doesn’t depend on a phone. None of this requires preparing for the apocalypse — it just requires preparing for the kind of mundane technology failures that already happen regularly.
The institutional version
Businesses face a more complex version of the same problem. Operations that depend on a single cloud provider, a single payment processor, or a single internet connection are exposed to outages that can shut them down for hours or days. The redundancy patterns — multi-region cloud, backup payment processors, secondary internet connections — aren’t free, but they’re cheaper than the cost of being offline during a major incident. Many small businesses don’t have any of them, which is why outages produce the hardship stories they do.
Bottom line
Technology is reliable enough that people have stopped designing around its failures, and that’s the conditions under which the failures hurt the most. Reasonable redundancy isn’t paranoia or nostalgia — it’s the recognition that complex systems break, and that the people best positioned to handle the next outage are the ones who didn’t put all their fallbacks into the same system that’s now down.
Leave a Reply