In 1991, the London Science Museum switched on a mechanical device that hadn't worked for a century and a half. It worked. But it wasn't the machine everyone thought it was.
Let's start with the fact that Charles Babbage never built his Analytical Engine. This isn't conspiracy theory — it's evidence on record. By the time of his death in 1871, what remained were drawings, notebooks, scattered parts, and the reputation of a brilliant failure. The machine existed as an idea that changed from year to year: sometimes 1000 numbers of memory, sometimes more, sometimes less, sometimes support for indexed variables, sometimes abandoning them. The project lived in a state of perpetual refactoring.
From 1833 (or from 1837 — Babbage himself gave different dates for when work began) he designed a device whose architecture looks frighteningly familiar to anyone who's opened a programming fundamentals textbook. The "Store" — memory capable of holding numbers up to 40 decimal digits long. The "Mill" — an arithmetic-logic unit that could add, subtract, multiply, divide, extract square roots, and compare values. Three types of punched cards: operation cards (what to do), variable cards (what to do it on), and cards for loading and storing (where to put the result). Loops. Conditional branches. Everything needed for universal computation, only made from brass and steel.
But the evidence doesn't add up. Relations with engineer Joseph Clement ended in a quarrel back during the Difference Engine No. 1 phase — the previous project, which also wasn't completed. Government funding ran out. The parts demanded precision that 1830s–1840s industry could barely provide. Each gear was workshop craftsmanship, each clearance a risk of jamming. Babbage knew how to design, but didn't know how to negotiate. His main enemy turned out to be not physics, but logistics.
In 1842, Italian mathematician Luigi Menabrea published an article about Babbage's machine in French. In 1843, Ada Lovelace translated it into English, but the translation was the lesser part of the work. She added notes that tripled the original in volume. In Note G — an algorithm for computing Bernoulli numbers, step-by-step instructions for a machine that didn't exist in metal.
This isn't just a table of commands. Lovelace understood what many contemporaries didn't see: the machine could operate on more than just numbers. If notes, letters, or logical statements were represented by symbols, the device would become universal. She wrote about this explicitly, without hedging. Babbage designed a calculator. Lovelace saw the potential of a programmable system.
But there's a nuance. Babbage himself knew about the possibility of symbolic operations — his notes confirm this. Lovelace didn't open his eyes, she articulated this understanding for the public. Her contribution isn't in inventing the concept, but in being the first to write a working algorithm with detailed breakdown. Note G isn't a prediction of the future, but an engineering specification for a machine of the present that never got switched on.
Babbage had no money. He had no team. He had no stable design. The drawings changed because he found more elegant solutions, then found problems in the elegant solutions and started over. The Analytical Engine project resembled an investigation where the lead detective changes his theory of the crime every week.
Industrial precision — that's what killed the design. Babbage demanded clearances in fractions of a millimeter, while 1840s workshops worked with precision sufficient for locomotives, but not for logic machines. Each part was made by hand, each error multiplied across thousands of components. This isn't a question of the engineer's genius — it's a question of available technology. Babbage was designing a Ferrari in the era of horse-drawn transport.
Funding ended after the failure of Difference Engine No. 1. The government invested 17,000 pounds (at the time — the cost of two warships), got a pile of parts and broken relations between client and contractor. No one was going to fund the Analytical Engine. Babbage tried to find private investors, wrote letters, explained the principles. No results.
In 1991, the London Science Museum switched on a device built from Babbage's drawings. It worked. The press wrote: "Nineteenth-century computer finally assembled." This was half true.
They built Difference Engine No. 2 — a specialized device for computing polynomials by the method of finite differences. Not a universal computer. Not the Analytical Engine. The Difference Engine had no loops, no conditional branches, no programmable logic. It solved one problem, solved it brilliantly, but that didn't make it a predecessor to modern computers in the full sense.
The reconstruction proved that Babbage's mechanical principles worked. That his precision requirements were achievable — 150 years later, when CNC machines and metalworking appeared that he could only dream about. The device consisted of 8000 parts, weighed 5 tons, and performed calculations without a single error. But this wasn't proof that the Analytical Engine would have worked.
Since 2010, the Plan 28 project has been trying to figure out whether the Analytical Engine can be built from surviving drawings. The verdict so far is discouraging: the documentation is incomplete, contradictory, and key mechanisms — especially support for indexed variables — are insufficiently detailed. This doesn't mean the machine was impossible. It means Babbage didn't leave unambiguous instructions.
Modern historians of technology read his notes like detectives — cross-reference different versions, look for references in letters, try to understand which variant he considered final. The problem is that there probably wasn't a final version. Babbage died without finishing the investigation of his own idea.
Calling the Analytical Engine the "first computer" is technically incorrect. It wasn't built. It wasn't unambiguously defined. It didn't launch an industry. But it was conceptually close to what would later be called the von Neumann architecture — though von Neumann himself probably didn't think about Babbage. Ideas were in the air twice: once in the 1840s, when technology couldn't keep up, and again in the 1940s, when everything aligned.
Drawings. Lovelace's notes. A working Difference Engine No. 2 in the museum. And a question that can't be closed: could the Analytical Engine have worked if Babbage had had enough money, time, and patience to see it through?
The successful assembly of the Difference Engine confirmed that Babbage's mechanics weren't fantasy. But the Analytical Engine is a different level of complexity. The Difference Engine is a calculator. The Analytical Engine is a programmable system. The difference between them is like the difference between an adding machine and a processor. One can be built in a 19th-century workshop if you really try. The other requires not just craftsmanship, but an industrial base that simply didn't exist then.
Babbage didn't lose because he miscalculated. He lost because he was born 100 years too early for his machine to be possible. Sometimes being right isn't enough — you also need to be on time while you're alive.