Greetings, and welcome to Friday, Nov. 13, 2026. It's Friday the 13th, the second one this year after February. Watch out for black cats, don't walk under any ladders, and whatever you do, don't open your property tax bill today. Some things are scary enough without help from the calendar.
My calendar says it's Friday the 13th, Indian Pudding Day, and Sadie Hawkins Day. Throw a pinch of salt over your shoulder for luck, then warm up with a bowl of Indian pudding, the old New England dish of cornmeal and molasses. And ladies, it's Sadie Hawkins Day, so go ahead and ask him out.
Today in History:
1775: Continental Army troops under Gen. Richard Montgomery capture Montreal during the American invasion of Canada.
1849: California voters approve a state constitution that bans slavery, paving the way for statehood the next year.
1927: The Holland Tunnel opens under the Hudson River, connecting Manhattan and Jersey City.
1942: President Franklin Roosevelt signs a law lowering the minimum draft age from 21 to 18.
1956: The Supreme Court rules Alabama's bus segregation laws unconstitutional, a victory for the Montgomery Bus Boycott.
1971: NASA's Mariner 9 reaches Mars and becomes the first spacecraft to orbit another planet.
1974: Karen Silkwood, a nuclear plant worker and union activist, dies in a car crash in Oklahoma while on her way to meet a reporter.
1982: The Vietnam Veterans Memorial is dedicated in Washington, D.C., bearing the names of more than 58,000 Americans who died in the war.
1985: The Nevado del Ruiz volcano erupts in Colombia, and the mudflows that follow bury the town of Armero, killing more than 23,000 people.
2015: Islamic State gunmen and suicide bombers attack a concert hall, a stadium, and cafes across Paris, killing 130 people.
Birthdays Today include: Edwin Booth, the great Shakespearean actor of his day, who once saved Abraham Lincoln's son Robert from being crushed by a train, before his brother John Wilkes assassinated the president; Robert Louis Stevenson, author of Treasure Island and The Strange Case of Dr Jekyll and Mr Hyde; Louis Brandeis, Supreme Court justice; Garry Marshall, creator of Happy Days; Joe Mantegna, actor from Criminal Minds; Whoopi Goldberg, actress from Ghost, Star Trek; Vinny Testaverde, NFL quarterback for the New York Jets; Jimmy Kimmel, late-night host; and Gerard Butler, actor from 300. If today's your birthday, too, happy birthday from all of us here. Having your birthday on Friday the 13th, you've already used up your bad luck for the year.
* * *
I have never signed on to Microsoft 365. For the same reason, I've gone out of my way to keep anything I care about out of anybody's "cloud." I write my columns using open-source equivalents of Office. Installed on my own boxes. Maintained by me. This week's Microsoft meltdown is exactly why.
Here's the point, up front: centralization buys you efficiency on the way up, and it buys you exactly the same efficiency on the way down. When everything runs through one system, everything fails at once. That holds for your email, your commuter train, and your federal government. There’s a poster over my desk that reads “Computers fail efficiently.”
At 12:54 p.m. Eastern on Thursday, Microsoft's status account admitted that it was "investigating reports of users experiencing issues accessing Exchange Online mailboxes." First it was the Americas. Within hours, Microsoft conceded that the mess could hit users in any region. Outlook, Teams, SharePoint, and Copilot went dark or close to it, and Downdetector lit up with complaints about Azure, OneDrive, and the Microsoft Store, too.
Microsoft's explanation, according to ABC News, came down to a "number of servers being unexpectedly unavailable," with the load piling up on what remained across multiple regions. So the company froze deployments, moved traffic off the sick hardware, and rebooted. That's the trillion-dollar version of "have you tried turning it off and on again?" Around 6 p.m., Microsoft said some servers were back and stable. Then it warned they might keel over again once traffic returned. As I write this Friday morning, the outage is in its second day, and Microsoft HQ in Redmond, Wash., is pointing to "encouraging recovery signals." I'll believe it when the "encouraging" part makes it to the people trying to get to their inboxes.
The damage didn't stay in corporate America. New York City's Emergency Management Department warned that city agencies relying on Microsoft might be slow to answer email or run online services. Over in Antalya, Turkey, the outage hamstrung delegates at the COP31 climate summit. I'll leave it to you whether that one counts as a loss.
And it's hardly the first time this year. In January, a Microsoft 365 outage dragged on for nearly 10 hours. In July, another one knocked out Teams and SharePoint for North American users. On August 31, a bad "core authentication configuration" took down services across the suite, and that outage ran into a second day as well. Somewhere along the way, Redmond turned "five nines" into nine fives.
Some of you will know I worked in IT for a number of years. The first thing any competent network admin asks about a system is where the single point of failure sits. The second is how big the blast radius gets when that point goes. The cloud sales pitch answers the first question by insisting there isn't one. (Yeah, right.) This week answered the second.
Certainly, there are efficiencies to be had. One vendor, one bill, one login, one support line. The problem is that the act of centralizing makes the crash and burn every bit as efficient as the billing.
Anyone remember CrowdStrike? In July 2024, one defective security update bricked roughly 8.5 million Windows machines. That was less than 1% of the Windows universe. Microsoft itself conceded at the time that the broad economic and societal damage came from how many enterprises running critical services used that one product. Delta Air Lines alone scrapped around 5,000 flights over several days because its crew-tracking tools ran on the same Windows boxes. Meanwhile, Southwest and Alaska, which didn't use CrowdStrike, canceled fewer than a half-dozen flights each. Diversity of systems isn't inefficiency; it's insurance.
There are parallels that cannot be ignored in the non-IT world as well. The trains, the planes, the buses, and of course, government.
Yes, you're reasonably safe on a train, a plane, or a city bus, right up until you're not. And when you're not, your only consolation is that plenty of other people are stuck, hurt, or dead right there beside you.
Consider the air traffic system, which has been as centralized as anything in America for decades. On January 11, 2023, one damaged database file in the FAA's NOTAM system forced the first nationwide ground stop since 9/11. One file. More than 1,300 flights got canceled, and another 10,000 got delayed. The FAA just recently decided (about a week ago, I think) that it should stop using floppy disks in its ATC systems.
Then came Newark. On April 28, 2025, controllers at the Philadelphia facility that handles Newark's approach traffic lost radar and radio for about 90 seconds. Picture a room full of controllers staring at black screens with planes in the air and no way to talk to them. Two weeks later, it happened again. Here's the kicker: The year before, the FAA had relocated Newark's airspace control from the New York TRACON on Long Island to Philadelphia. Consolidation, they called it. The airspace stayed in New Jersey. The single point of failure moved to Pennsylvania.
Rail works the same way. The Northeast Corridor carries some 820,000 trips every weekday between Boston and Washington, all running on one spine of track and overhead wire. In May 2024, a single signal wire touched the overhead catenary, blew out the power, and stopped Amtrak and NJ Transit cold, leaving thousands stranded at Penn Station. Then-Gov. Phil Murphy called it an "unmitigated disaster" and pointed out that NJ Transit pays Amtrak $100 million a year to maintain that corridor. One wire, one corridor, everybody stuck.
Compare that with freight on the highway. Know why trucking has been the choice for most shippers? Anyone who has driven commercially (as I have) knows the system works because nobody runs it. Hundreds of thousands of independent carriers, dispatchers, and drivers each make their own calls. When one truck breaks down on the thruway, the load still gets there, just on another truck. Nobody in a central office can ground every rig in America because a database file went bad.
Which brings me to the most centralized system Americans depend on: Washington.
One year ago, on November 12, 2025 (a year to the day before Microsoft's servers fell over), the longest government shutdown in American history finally ended after 43 days. Whoever you blame for it — and you know where I come down — look at the blast radius. SNAP benefits stopped for households that depended on early-month payments. The FAA ordered flight cuts at 40 major airports because unpaid controllers stopped showing up. White House economic adviser Kevin Hassett put the cost to the economy at roughly $15 billion a week.
A fight among a few hundred people in one city froze food aid in Montana, air traffic in Atlanta, and museums on the Mall. That's not a bug in centralized government; that is centralized government. A living, breathing, potent example of why we should be avoiding centralized anything.
Handing day-to-day governance to Washington looks cheaper at first. Then one day, you find the people running it can't hear you, don't answer, and don't much care about your local needs or your local voice.
The Founders understood blast radius long before anyone coined the term. In Federalist No. 45, James Madison promised that the powers of the federal government would be "few and defined," while those left to the states would be "numerous and indefinite." Justice Louis Brandeis later described the states as laboratories of democracy. Think about that metaphor the way a network admin would. When one lab blows up, the other 49 keep working. When the only lab blows up, you've lost the building.
My files live on drives I can put my hands on. When one dies, it's my problem and my fix, and nobody at a climate summit in Turkey ever hears about it. Microsoft would call that inefficient. I call it a firewall.
I say it's time to start moving in that direction again.
= = =
Thought for the Day: Superstitious people avoid ladders and black cats on Friday the 13th. The rest of us know the scariest thing out there is a politician who says, "I'm from the government, and I'm here to help."
VIP members: Have you been hit by this Microsoft outage?
Take care today, gang. I'll see you here tomorrow.
Recommended: Thune: 'What Could We Have Done Differently?' Glad You Asked, Senator.