How a London tube strike accidentally improved thousands of commutes
And what it says about the processes you're about to hand to AI
You know that route you take every morning without thinking?
The one where your legs just carry you and your brain is somewhere else entirely?
You probably couldn’t tell me when you last considered whether there’s a better way to go. It works. You stick with it.
On the evening of February 4th, 2014, millions of London commuters had that choice taken away from them. Tube workers walked out on a 48-hour strike. 171 out of 270 stations closed.
Most people were furious. But something unexpected happened.
Researchers from Cambridge and Oxford got hold of anonymised Oyster card data from around 18,000 regular commuters - people whose journeys were completely predictable in the weeks before the strike. The same station in and out, every single morning.
During the strike, half of them started and ended their journey at a different station. Forced out of their routine, they improvised.
Here’s the interesting bit. After the strike ended and every station reopened, about 5% of the most habitual commuters permanently switched their route. They’d accidentally discovered something better - faster, less crowded, more pleasant - that had been available to them all along.
They’d been commuting the same way for years. Some of them, presumably, for decades. It took someone else shutting down their usual option to make them notice.
Herbert Simon, the Nobel-winning economist, had a word for this: satisficing.
It means settling on a solution that’s good enough rather than continuing to search for the best one. Most of the time, that’s the right strategy. Life is too short and too complex to optimise everything.
The problem is when “good enough” calcifies into “the only way we do things.” When nobody questions it anymore. Not because the process is bad. Because it’s functional. It works. Nobody’s complained. So nobody’s bothered doing anything to improve it either.
The belts and shafts problem
Last week’s piece was about understanding what a job really involves before you automate it.
This is the harder question underneath it: should the job exist in its current form at all?
In the 1890s, factories started replacing their steam engines with electric motors. On paper it should have changed everything. In practice, almost nothing changed for decades. Productivity barely moved.
Because the factories were designed around steam. One enormous engine in the basement, driving everything above it through a complicated system of belts, shafts and pulleys. When electric motors arrived, owners ripped out the steam engine and bolted in an electric one.
And the problem was they stuck with the same layout, the same belt and the same shafts.
The new technology was doing the old job.
The real gains came much later, when someone worked out you could give every machine its own small motor. That let you redesign the factory floor entirely - placing machines where they made sense, not just where the belts reached.
Productivity exploded.
Most AI projects right now are in the belts and shafts phase. Drop an AI tool into an existing workflow, get a marginal improvement, declare victory. But the process itself - designed for an era when humans did everything by hand - goes completely unexamined.
You’ve made a slightly quicker commute but on the wrong route entirely.
Manufacture your own tube strike
Now, ideally, you don’t want to wait for a crisis to rethink how you work. But you can simulate the conditions for one?
Pick one workflow that nobody's questioned for years.
Get the people who do the work in a room.
Map what actually happens.
Then ask: if we were starting this today - with the tools, the team, and the information we have now - would it exist, and if so, would we build it this way?
You might find that you would. Fine. At least you’ve pressure-tested it.
But in many cases, you’ll find a process - a way of doing things - that looks a lot like a commuter who’s been taking the Circle Line for fifteen years when there’s a pleasant walk from Liverpool Street that would have been better all along.
Thanks for reading,
Ollie
Ollie on Work is a weekly newsletter about what I’m learning from building with AI, advising leadership teams, and trying to bridge the gap between what technology can do and how businesses actually work. If someone forwarded this to you, you can subscribe here: