Yesterday we wrote about why the conspiracy theories around this month's outages are false. That piece did not answer the harder question: why do they keep happening?
The answer is duller than a plot and considerably more worrying.
Different causes, one shared weakness
Each of the big outages of the past week appears to have a different root cause. Forensics takes time and the exact cause of each is not yet known, but early indications rule out a coordinated attack.
What they share is not a culprit. It is a condition: the technologies underneath the internet were never built for a world this interconnected, or for one where computers can find and exploit weaknesses faster than humans can fix them.
"These events do not fill me with confidence about the resilience of our digital infrastructure," said Matthew Green of the Johns Hopkins Information Security Institute.

The Jenga tower, and the man in Nebraska
There is a well-known xkcd cartoon that draws "all modern digital infrastructure" as an enormous Jenga tower resting on one thin vertical block, labelled: a project some random person in Nebraska has been thanklessly maintaining since 2003.
It is a joke that stopped being funny this month.
One of those blocks is PHP, a programming language most people have never heard of, which powers about 70% of the world's websites. Researchers using AI agents found a way to exploit a PHP flaw that could let an attacker read passwords, steal entire databases or take over sites. It was patched last month — we covered that patch and what to do about it at the time.
The fix is not the point. The point is that a decades-old tool holding up three-quarters of the web carried a serious flaw that nobody had found until an AI went looking.
"We're dependent on so many things we've never even heard of behind the scenes that keep society running," said Stuart Madnick, professor of IT security at MIT. "In that regard, we've become tremendously vulnerable to things we don't even know about until something goes wrong."
The Titanic comparison is the more precise one
The ship's watertight bulkheads were not defective. They were designed against a worst case that turned out not to be the worst case.
Rebecca Krauthamer, chief executive of the cryptography firm QuSecure, made the point that even well-designed systems fail against problems their designers could not have imagined. That is a different kind of failure from negligence, and it is much harder to fix, because the fault is in an assumption rather than a component.
Why AI makes this both better and worse
AI has given security researchers a far better ability to find weak supports and strengthen them. That is genuinely good news, and the Java flaw we reported yesterday was found exactly that way.
But attackers have the same tools. Agents work faster and for longer than people, which makes them very good at finding flaws — regardless of who is running them.
And there is a second-order effect worth naming. The race to patch has pushed some companies into shipping security updates that were not ready. A rushed fix is its own outage risk, which means some of this month's disruption may be caused by the response to the threat rather than the threat.
What this means for the rest of us
Do not expect it to stop soon. The useful response is not to find the hidden hand, because there probably isn't one. It is to know your own dependencies: which single provider, if it failed tomorrow, would take your business, your payments or your records down with it — and whether you have any answer at all for that day.




