What the Hell Is Going On? America's Systems Are Failing, and the Bigger Picture Is Getting Hard to Ignore

2 hours ago

14 min read

A pilot in a cockpit holding a mug reading WHAT THE HELL IS GOING ON, surrounded by burning infrastructure: an air traffic control tower, an exploding airliner, a burning Microsoft building, a capsizing cruise ship, and a Friday the 13th calendar

Microsoft goes down. The FAA grounds the nation. A tanker strikes a cruise ship. Aircraft emergencies dominate headlines.

And Washington is rolling out new artificial intelligence systems while investigators warn that our critical infrastructure is dangerously vulnerable. Is this coincidence, incompetence, cyberwarfare, or something more fundamental?

I'm going to ask a question that I suspect quite a few Americans are asking right now.

What the hell is going on?

Seriously.

On Thursday, November 12, Microsoft experienced a massive outage affecting businesses, government operations, and millions of people who depend on its services.

On Friday, November 13, the Federal Aviation Administration ordered a nationwide ground stop. Not a couple of airports because of weather. Not an airline having a bad day. A nationwide halt to departures because of communications problems affecting the country's regional air traffic control centers.

Meanwhile, a tanker struck a docked cruise ship on the Mississippi River in New Orleans, rolling the passenger vessel onto its side with roughly 129 passengers aboard. On the night of November 9, a passenger aboard United flight 2174 from Newark to New Orleans allegedly tried to force his way into the cockpit while claiming he had a bomb.

And on November 1, a private jet crashed in Palm Beach County, Florida, killing nine people.

Now, before somebody starts connecting all of this with red string on a bulletin board, let's establish something important.

There is currently no verified evidence that these incidents are connected.

But I'm also not particularly interested in pretending that the larger picture isn't worth examining. Because the story here may be considerably bigger than whether somebody hacked the FAA. It may be about the increasingly fragile technological infrastructure upon which almost everything in modern society depends. And that should concern everybody.

First, America grounded its aviation system

At 3:54 p.m. Eastern on Friday, November 13, the FAA issued an extraordinary nationwide ground stop. The alert directed that all departing traffic remain on the ground "until further notice," citing "extraordinary circumstances and for reasons of safety." Inbound international flights, the agency said, would be managed on a case by case basis. Aircraft already in the air were not ordered to immediately land. That's an important distinction from some of the more dramatic social media reporting circulating Friday afternoon. Still, this was an exceptionally rare event.

The last comparable nationwide departure halt came in January 2023, following a failure of the FAA's Notice to Air Missions system. Before that, the defining nationwide aviation shutdown followed the terrorist attacks of September 11, 2001.

The national airspace system handles more than 44,000 flights and roughly three million airline passengers every day. You don't simply stop that operation without an extraordinarily serious reason.

Homeland Security Secretary Markwayne Mullin said there was "no indication as of yet" that anything nefarious caused the halt, and described it as a telecommunications outage. He said President Trump had been briefed and that officials were "not ruling out anything at this point" as investigators worked. He called the event unprecedented. Transportation Secretary Sean Duffy said FAA technicians were investigating and promised immediate updates.

Aviation safety expert Jeff Guzzetti called the outage extraordinary and said he didn't know what would cause it. He also pointed out something worth holding onto: pilots are trained to work through communications failures, and there are redundancies for exactly this. The immediate problem, in his telling, was shifting from a safety issue into a logistics problem, with aircraft stacking up and airports running out of pavement to park them on.

Reagan Washington National Airport was reported to be at capacity. Chicago O'Hare was nearing it.

And here's the detail that matters more than any of the official statements.

Shortly before the nationwide alert went out, both major Chicago airports, O'Hare and Midway, were already listed on ground stop alerts attributed to an "equipment outage."

So: did a failure begin in Chicago and propagate through interconnected systems? Did multiple centers lose access to a shared communications dependency at roughly the same time? Or did the FAA recognize a broader failure and progressively impose restrictions as the magnitude became apparent?

Those are three very different technical scenarios. And until investigators reconstruct the actual failure timeline, we shouldn't confuse the sequence of public announcements with the sequence of technical failures. Mullin's answer is reassuring. It doesn't answer the more important engineering question.

How does a telecommunications problem become serious enough to stop departures across an entire nation?

Then there's Microsoft

Thursday's disruption hit Microsoft 365 services: Outlook, Teams, Exchange Online, and the connected products that ride on them. Government and institutional operations were affected along with everyone else.

Microsoft's technical updates described elevated infrastructure load, servers failing health checks and becoming unavailable, traffic redistributed to what remained, and repeated attempts to restore stability.

Think about your car for a moment. You know how your check-engine light can come on because a sensor is malfunctioning, even though the engine itself is perfectly fine? Now imagine that sensor isn't merely warning you about a problem. Imagine it has the authority to shut down your engine. And then imagine thousands of engines using the same monitoring system.

Welcome to one of the fundamental problems in modern automated infrastructure.

Health monitoring exists to protect a service by detecting bad components and pulling them out of rotation. But what happens when the monitoring mechanism itself starts making incorrect decisions? What happens when an automated process removes too much capacity, shoves the traffic onto the survivors, and overwhelms them in the process?

Suddenly something designed to prevent failure helps amplify it.

Two details from Microsoft's own disclosures deserve attention here.

First, the company said affected servers "can become unresponsive again once they begin receiving traffic." That's not the profile of a false alarm. A badly calibrated sensor doesn't get worse when you use the equipment. That's the profile of a load-dependent fault, which is a different and more interesting problem than a bad health check.

Second, Microsoft identified the overwhelmed component as the process "responsible for establishing secure connections." Not a vague core process. The one that handles secure connections.

And there's a contradiction sitting in the public record that nobody has resolved. Microsoft initially said the health-check failures "may be related to a recent change." Later, it said it hadn't identified any recent changes that correlate with the start of impact. Those two statements can't both be the whole story, and the difference matters enormously. One describes a deployment problem, which is embarrassing but bounded. The other describes a system that drifted into failure on its own, which is a much harder thing to fix and a much worse thing to own.

Microsoft reported no indication of malicious activity, and said a change applied to improve serviceability wasn't behaving as expected while recovery work continued. Which makes all of this potentially more concerning, not less. Because if these systems can disrupt themselves without any attacker, we need to rethink how we're designing them.

Three days earlier, the FAA expanded an AI system

Here's a detail that deserves serious attention.

On September 21, Transportation Secretary Duffy and FAA Administrator Bryan Bedford unveiled the Strategic Management of Airspace, Routes and Trajectories, or SMART. It's an AI-supported, cloud-based platform that centralizes roughly 200 data streams including weather, flight paths, traffic flow, and controller staffing, then projects where congestion will build and recommends how to route around it. The goal is straightforward: fewer delays, less time circling, more efficient use of airspace. The FAA began using it that same day in a limited mode in the airspace around Washington, D.C.

On November 10, it expanded to all 22 air route traffic control centers.

Three days before the nationwide ground stop.

Now, this is exactly the kind of problem artificial intelligence should be helping us solve. Air traffic demand genuinely does exceed capacity, and humans genuinely do benefit from better forecasting. The FAA's own framing of the problem is correct. But introducing sophisticated software into critical infrastructure creates new engineering obligations. What systems does it connect to? What network resources does it share? What happens if the application fails? What happens if a configuration change has consequences somewhere else? And can the platform be isolated without disrupting what controllers need to operate safely?

The FAA has been explicit about the boundaries. Its announcement stated that SMART "cannot be used to replace the critical role air traffic controllers play or take over control of an aircraft," that all of its recommendations are reviewed by FAA staff, and that local leadership at air traffic control facilities can accept or pass on those scheduling suggestions. Duffy has repeated that characterization since: it's a recommendation tool. I'll take that at face value. I'll also point out that "advisory" is a description of authority, not of risk. A recommendation engine that's wrong in a busy system doesn't need to control an aircraft to make things worse. It only needs to shape the decisions people then make under time pressure.

There is no evidence that SMART caused Friday's communications outage. It isn't the equipment that carries controller instructions to pilots. The failure was in communications, not in traffic planning.

But the timing makes its deployment history a legitimate subject for investigation rather than an inconvenient coincidence to be dismissed.

Washington was warned

Here's what makes this more frustrating than confusing.

On September 21, the Government Accountability Office published a report on aviation cybersecurity, and it's worth reading slowly. GAO found that the FAA has identified spectrum-related threats to the national airspace, including spoofing and jamming, but hasn't completed the risk and mitigation assessments or updated the security documentation needed to address them. It found that the FAA didn't have a defined, real-time monitoring and detection capability for those threats. And it put the consequence in plain English: there are tools that monitor for such threats in real time, but the FAA doesn't currently have them, so it can only investigate incidents after they've been reported.

It also found something worse. The text-based communications that controllers and pilots use to exchange information, including ACARS and Controller Pilot Data Link Communications, are vulnerable to interception and spoofing because of limits in authentication, encryption, and protocol design. A malicious actor, GAO wrote, could transmit fraudulent clearance cancellations, possibly leading to flight delays or safety issues.

GAO made nine recommendations. The Department of Transportation, responding for the FAA, concurred with all nine. All nine remain open.

Then there was the other warning, and this one came from Congress.

On October 9, Don Beyer of Virginia led a letter to Transportation Secretary Duffy and FAA Administrator Bryan Bedford, with co-leads April McClain Delaney, Alexandria Ocasio-Cortez, Kweisi Mfume, Suhas Subramanyam, and Sarah Elfreth. They asked the FAA to pause SMART at Reagan National, Dulles, and BWI until the agency answered questions about safety testing, implementation, and oversight. Read what they were worried about, because it reads like a checklist of everything that became interesting later. They wanted to know how the system performed during severe weather or outages, and whether any safety review requirements were waived during the contracting process. They wanted to know what training frontline controllers had received, with the letter raising concerns that safety workers got only days of notice or none at all before deployment. And they wanted to know who is responsible if a recommendation contributes to a safety incident.

They also noted that the FAA had laid out plans to expand nationwide within 90 days.

Beyer's stated standard wasn't exotic. Before deploying an AI-enabled air traffic tool, he wrote, the FAA "must demonstrate, through air traffic controller feedback and independently verified testing, that this technology is safe and reliable, and it must clearly communicate those findings to the public."

That was October 9. The expansion to all 22 centers came November 10, well inside that 90-day window. The nationwide ground stop came November 13.

Does that prove anything about causation? No. A letter from six members of Congress doesn't diagnose a telecommunications failure, and the people asking for a pause had reasons that predate this week's events entirely.

But it does demolish one comfortable explanation. The concern wasn't invented on social media Friday afternoon. It was documented in a federal audit and in a congressional letter, in writing, with dates on them, before anything broke.

And the underlying fragility isn't new either. The FAA has been working to replace aging telecommunications equipment, obsolete copper connections, and legacy systems. Congress approved $12.5 billion toward modernization in 2025. Duffy has since said the administration will seek another $30 billion on top of it, which tells you what the people closest to the problem think $12.5 billion buys.

Friday's event doesn't establish which weakness failed. It makes the case for addressing them considerably harder to wave off.

Ships are colliding and airplanes are making headlines

Now let's widen the lens, carefully.

At about 12:30 p.m. Friday, the tanker Atlantic Asphalt struck the American Heritage, an American Cruise Lines river vessel docked at the Port of New Orleans, and rolled it onto its side. Roughly 129 passengers were aboard, plus crew.

The American Heritage had docked about two hours earlier to begin disembarking passengers. People who had already gotten off were being moved into the port's administration building when the ship went over. Thirty-two people were hospitalized, none with critical injuries, and New Orleans Mayor Helena Moreno said all passengers and crew had been accounted for. Governor Jeff Landry, heading to the city, called it "a literal miracle on the Mississippi."

That outcome is the most important fact in this section, and it's the one that got the least attention.

Investigators are examining what happened. A source told the Guardian it was possible the tanker lost power before impact, but that isn't confirmed, and vessel tracking showed an uneventful track until the ship began drifting to starboard just after passing the Crescent City Connection bridge.

Could a vessel lose propulsion or steering and cause a catastrophe? Of course. Could navigation be disrupted? Technically, yes. But nothing publicly established about this collision shows a hacker anywhere near it.

Then there's United 2174: a Boeing 737 MAX 9 with 46 passengers and six crew, bound from Newark to New Orleans on November 9. A passenger allegedly stood up, told the crew he had a bomb in his backpack, and tried to get into the cockpit. The pilots squawked the hijack code. The aircraft landed safely. A 34-year-old man was arrested, and authorities said the bomb claim was false. That is a legitimate aviation security event. It is not evidence of a communications-system attack.

And the November 1 crash of a Learjet 60 in Palm Beach County, which killed all nine aboard after departing Austin for North Eleuthera, remains under NTSB investigation. Reports of icing and turbulence don't establish a final cause. We should also be careful about claims of an unprecedented surge in aviation security incidents. A list of events in Chicago, Milwaukee, Nashville, and Atlanta isn't meaningful without flight numbers, dates, and circumstances. One incident generates headlines in several cities. And even a genuine increase in reported incidents has to be measured against historical rates before anybody can call it a statistical anomaly.

But here's my point.

When multiple serious events land in rapid succession, they create a picture of disorder. And whether or not those events are connected, the individual failures expose vulnerabilities in systems people depend on. That's worth examining without pretending the connections are already proven.

Is this a cyberattack?

Let's address the elephant in the room.

Could hostile actors target American transportation, telecommunications, and digital infrastructure? Of course. Nation-state operations, criminal intrusions, ransomware, and attacks on industrial control systems are well-established national security concerns. But saying something could happen isn't the same as demonstrating it did.

As of this writing, federal officials haven't identified a coordinated cyberattack connecting Friday's FAA disruption with Thursday's Microsoft outage, the maritime collision, or the other incidents circulating online. The preliminary assessments point away from malicious activity in both the FAA and Microsoft cases. Those assessments may sharpen as investigations continue. The absence of a confirmed attack doesn't mean there's nothing to worry about. It means we need to examine the vulnerabilities that exist whether or not anybody exploited them.

Because a system that fails catastrophically under ordinary operating conditions is a national security concern in its own right. An adversary doesn't have to create a weakness. Sometimes the weakness is already there, and all an adversary has to do is wait for the right Tuesday.

The real problem: complexity has become its own vulnerability

For decades, the technology industry has operated on a simple assumption. Connect everything. Centralize everything. Automate everything. Move it to the cloud. Make it faster, cheaper, easier to manage. And for the most part, that strategy delivered extraordinary results.

But somewhere along the way we built interdependencies that even experienced engineers can't fully trace. One platform depends on another. That platform depends on an authentication service. The authentication service depends on network connectivity. The network depends on routing equipment, carriers, software configurations, monitoring systems, and automated recovery processes. All of which have dependencies of their own.

What looks like redundancy on an architecture diagram is often several supposedly independent systems sharing the same vulnerable resource.

That's a common-mode failure, and it's one of the more dangerous problems in complex systems engineering. You can have five backup generators. If all five depend on the same faulty control panel, you don't have five generators. You have one.

This is why the FAA investigation has to examine more than whichever component reported the first error. Investigators need the whole dependency chain. And there's a distinction worth holding onto. A nationwide ground stop may actually demonstrate that aviation safety procedures worked exactly as intended. Controllers and regulators recognized unacceptable uncertainty and stopped adding flights to a system they couldn't trust. That's the system doing its job.

Why the infrastructure became unreliable enough to require that decision is another matter entirely. The safety response may have worked perfectly. The infrastructure may not have.

Are we entering an era of technological chaos?

Maybe. But let's define terms.

We don't have enough evidence to conclude that this concentration of incidents represents a statistically unprecedented wave of failures. News travels faster than ever, and social media puts unrelated industries side by side, which makes an ordinary run of bad days look like a coordinated national emergency.

That effect is real. So is the complexity underneath it, and I think the second issue gets a fraction of the attention it deserves.

We're introducing AI, automated decision-making, cloud orchestration, and tightly coupled software into systems designed around very different assumptions. We're doing it while replacing decades-old infrastructure. And we're asking the same organizations to keep operations running smoothly while rebuilding the technology underneath them.

That's a hard engineering problem under ideal conditions. It gets dangerous when redundancy, monitoring, testing, and change management don't keep pace with the complexity they're supposed to govern.

The solution isn't to stop modernizing. It's the opposite. We need better systems.

But we need systems designed to fail safely, degrade gracefully, and preserve essential operations when pieces break. And we need to stop assuming that more automation automatically buys more resilience. Sometimes it does. Sometimes it adds a dependency nobody fully appreciates until the night it matters.

So, what the hell is going on?

Here's where I land.

We have a major Microsoft service disruption with an unresolved contradiction in its own account of what changed. We have a nationwide FAA communications emergency that grounded the country. We have a serious maritime collision with a genuinely remarkable outcome. We have disturbing aviation security incidents and a fatal crash. And we have federal auditors and members of Congress documenting weaknesses in the relevant systems, in writing, before any of it happened.

That's an extraordinary collection of headlines. But extraordinary headlines aren't proof of an extraordinary conspiracy. The evidence supports several serious incidents, not a demonstrated coordinated campaign.

What concerns me is the possibility that we're watching a slower, broader failure emerge in plain sight. We spent decades making everything smarter, faster, more centralized, and more connected. Now we get to find out whether we made it resilient enough. Because there's an uncomfortable possibility here. What if nobody attacked the FAA? What if nobody hacked Microsoft? What if the failures that disrupt millions of people increasingly originate inside the very systems designed to prevent disruption? That's less dramatic than cyberwarfare. From an engineering and national security perspective, I'm not sure it's any less troubling.

We can defend against an enemy we can identify. How do we defend against the unintended consequences of complexity we created ourselves?

Maybe that's the real story from this Friday. Not that America is under attack. But that the machinery running America may be considerably more fragile than we want to believe. And if that's what we're discovering, we've got a much bigger conversation ahead of us.

Oh, and one more thing.

Happy Friday the 13th, America. Apparently even the infrastructure decided to get into the spirit.

Editor's note: This analysis reflects reporting available on November 13, 2026. The FAA ground stop, the New Orleans collision, and the other investigations discussed remain open. No evidence has established a coordinated attack or a causal connection between these incidents. Where a claim rests on a single source or remains unconfirmed, it is described as such in the text.

Sources

Rich Washburn is a technologist, strategist, and Founder & Chief AI Architect of ARIA AI Labs, working at the intersection of AI, infrastructure, communications, and capital. He also serves as Managing Partner and Chief AI Officer at Eliakim Capital.