The technical details of how the network attacked itself has been described with tremendous accuracy by Bruce Sterling.
Basically, a bunch of AT&T’s long-distance switches were running recently upgrades code with a tiny bug which caused the switches to reset themselves if two uncommon things happened at the same time.
When the switch starts up, it sends a signal to nearby switches saying, “Hello! I’m here now, and can take calls.” Its neighbors note their new companion. However, here’s the problem: if two phone calls come into switch in the same millisecond, while it is logging a “Hello!” from a new neighbor, it hits the bug and resets itself.
While the switch resets, the phone-call load on its neighbors goes up, making them more likely to get two calls at the same time. And when the crashed switch comes back up, it sends a “Hello!” into the more heavily loaded switches — which crashes them, increasing the load on their neighbors, making those further neighbors more likely to crash when the original switch’s neighbors come back up.
And the wave propagated quickly through the telephone network, crashing switches in far-away cities from the first crash, in Manhattan, even echoing back to crash switches that had previously crashed and restarted. And this went on for some time.
It was a mess. AT&T took out a full-page newspaper ad to apologize for the problem, it was that bad.
At the time, though, this kind of an explanation, even as simple as I’ve tried to make it above, was hard to approach and even harder to accept by the phone-company security people and the law-enforcement authorities of twenty-three years ago. They knew the story: hackers said they would crash the network, the raids on the Atlanta hackers turned up evidence that they knew how to crash the network, and now the network had crashed. Who crashed that first switch that caused all this nonsense? Had hackers stolen the source code, figured out the bug, and sent “Hello!” signals to a bunch of switches until one of them finally fell over?
I don’t know that we ever figured it out. Decades later, coming out of a hacker culture where the only currency was bragging rights, it’s odd that no one ever claimed responsibility for the snowballing phone switch crash of 1990. Either someone is so mortally embarrassed that they won’t talk about it, or were playing around and have no idea what they accidentally did — or hey, maybe it was really a systemic accident.
It’s likely that a set of recently upgraded switches, heavily loaded with holiday traffic, would hit the bug on their own and crash on their own. It doesn’t require the invention of evil hackers.
Evil hackers will be invented later. Don’t worry; we’ll get there.
But instead, the authorities attacked the hackers. At the time, it must have felt great, taking out hacker after hacker and finding little behind the curtains but really geeky people, mostly, powerless in the real world.
There were other hackers, however, who in the previous ten or so years had generated a lot of currency that was not solely bragging rights, but actual, real-world currency, that you could take out of a bank and strike someone over the head with.
Two months after the crackdown saw the hacking world strike back in the most unexpected way. Or actually, maybe the authorities could’ve expected it. After all, hackers like nothing more than subverting a system by using its own tools against it.