The Two-Century Argument About What Machines Can Do
Three cards, fifty-four columns. Read each column downwards and it is one character of the sentence Ada Lovelace wrote about the Analytical Engine in 1843, in the eight-bit code every computer now uses. A hole is a one.
In 1843 Ada Lovelace wrote that a machine can do only what we know how to order it to perform. We have been arguing with her ever since.
Ada Lovelace was twenty-seven, translating a French paper about a machine that did not yet exist, when she wrote the sentence that has never stopped mattering. The Analytical Engine, she said, has no pretensions whatever to originate any thing. It can do whatever we know how to order it to perform. In 1950 Alan Turing gave her claim a name, Lady Lovelace's Objection, and argued, politely, that she was wrong.
This book is the story of the people who believed her, the people who disagreed, and the people who kept building machines that forced the question back open. It runs from Babbage's brass wheels to the transformer, through Bletchley and the Philadelphia basement where six women programmed the first electronic computer, through Bell Labs, Apollo, Xerox PARC, the garage, the network, the phone and the data centre.
It is a book of people, not of technologies, and most of them are not the ones who got the credit. The cobbler's son who turned logic into algebra. The Post Office engineer who built Colossus with his own money. The women who wove the Moon landing's software out of copper wire. The researcher who designed the language every database speaks and died at twenty-seven. Every technical idea is built from first principles, one handhold at a time, so that by the last chapter the reader has climbed from gears to neural networks and can decide for themselves whether Ada was right.
Each time a machine does something new, somebody explains why that still is not thinking. The explanations are set out here in order. The rows in heavy type are the two that started it.
| Year | Who | What the machine had just done | What was said |
|---|---|---|---|
| 1843 | Ada Lovelace | Nothing. Babbage's Analytical Engine existed as drawings and a few trial pieces. | “The Analytical Engine has no pretensions whatever to originate any thing. It can do whatever we know how to order it to perform.” |
| 1950 | Alan Turing | Run stored programs on a few thousand valves in Manchester. | He names her claim Lady Lovelace's Objection and answers it: “Machines take me by surprise with great frequency.” |
| 1997 | After Deep Blue | Beaten the world chess champion over six games in New York. | It is only following rules. |
| 2012 | After AlexNet | Won the ImageNet contest by more than ten percentage points, on two gaming cards in a bedroom. | It is only matching patterns. |
| 2022 | After ChatGPT | Written essays, code and sonnets for a million people in its first five days. | It is only predicting the next word. |
| Now | You | Whatever was announced this week. | Each claim sounds correct until the next machine arrives. Then it sounds like a retreat. |
Before 1940. Three people describe the computer before anyone can build one.
The 1940s to the 1960s. War pays for the machines. The people who build them are mostly left out of the photograph.
The 1960s to the 1980s. The computer leaves the institution and ends up on a desk, by way of a garage.
The 1980s to the 2000s. The machines are joined together, and nobody is in charge.
The 2000s and 2010s. Search, the phone, the feed and the cloud. The computer moves into the pocket and starts watching.
The 2010s to now. A technology nobody believed in starts to work, and Ada's sentence is back on the table.
The book is being serialised on Substack in reading order. Each chapter's page here carries its opening and its sources; the full text is on Substack, and subscribers get each new chapter by email.