Why “First” Is a Slippery Word
Ask which machine was the first computer and you will get different answers depending on who you ask. The reason is not that historians are careless. It is that “computer” bundles together several properties, and the earliest machines had some of them but not others. Does it have to be electronic? Does it need to be programmable? Must it be general-purpose? Must it keep its program in memory? Each definition points to a different machine.
Before the 1940s, none of this was settled. Several groups in different countries, often unaware of one another because of wartime secrecy, were building automatic calculating devices at roughly the same time. The result is a set of overlapping stories, not a single starting line. Most 1940s computers were one-off projects, built by small teams for a specific purpose, and not products that anyone could order.
The Main Candidates
Here is a brief tour of machines that are often discussed, with the caution that dates refer to different stages (design, working prototype, completion) depending on the source.
Zuse’s machines in Germany. Konrad Zuse built a series of calculating machines in the 1930s and early 1940s. His Z3, from around 1941, used electromechanical relays and read its instructions from punched tape. It is often described as an early programmable, automatic digital computer. It was not electronic, and the original was later destroyed during the war.
The Atanasoff-Berry machine in the United States. John Atanasoff and Clifford Berry built a special-purpose electronic device in the late 1930s and early 1940s at Iowa State College for solving systems of linear equations. It used electronic components for arithmetic but was not programmable in the general sense. Its influence on later machines became a subject of legal and historical dispute.
Colossus in Britain. The Colossus machines, built during the Second World War for codebreaking, used vacuum tubes and are often described as among the first large-scale electronic digital machines. They were designed for a specific cryptanalytic task, though they could be configured with switches and plugs. Because the project stayed secret for decades, its role was long overlooked.
The Harvard Mark I. Completed in the early 1940s by Howard Aiken’s team in collaboration with IBM, this large electromechanical machine could execute long sequences of operations read from punched paper tape. It was a major automatic calculator, though it was built from mechanical and electrical parts and not from electronic ones.
The ENIAC. Built in the United States and completed in the mid-1940s, ENIAC was a large electronic machine using thousands of vacuum tubes. It was general-purpose in the sense that it could be set up for many different problems, but it was programmed by rewiring cables and setting switches, not by loading a program into memory.
Sorting by Definition
A table can show how the choice of definition changes the picture:
| If “computer” means… | Machines often mentioned |
|---|---|
| An automatic digital calculator | Z3, Harvard Mark I and others |
| An electronic digital calculator | Atanasoff-Berry machine, Colossus, ENIAC |
| A programmable electronic machine | Colossus (in a limited way), ENIAC |
| A stored-program electronic machine | Machines of the late 1940s, discussed below |
Even this table simplifies matters, since historians argue about the details of each machine. It should be read as a guide to the reasoning, not a verdict.
Stored Programs Come Later
Many of the machines above were programmed by physical set-up. The next step was storing instructions in the machine’s memory alongside the data, so that changing the program meant changing memory contents. Discussions of this idea in the mid-1940s involved several people, and an influential written description circulated in 1945. Experimental machines in Britain, including one at Manchester in 1948 and another at Cambridge in 1949, are generally counted among the first to run stored programs. This design, usually called the stored-program concept, became the pattern for nearly all later general-purpose machines, and within a few years the first commercial computers were being offered to customers. Those who define “computer” by this standard point to the late 1940s.
Beyond the Usual Names
Most accounts concentrate on Britain, Germany and the United States, but work elsewhere also deserves mention. Researchers in other countries built or designed machines in the 1930s and 1940s and the years after, and global computing history is more complicated than a short list can capture. It is also easy to overlook the engineers, mathematicians and technicians, many of them women, who operated and programmed these machines but were often not credited.
What to Take Away
A better question than “which was first?” may be “first at what?” Each of these projects contributed something: automatic sequencing, electronic speed, programmability, or stored instructions. None sprang up in isolation. They drew on earlier mechanical calculators, punched-card tabulators and telephone relay technology, and they influenced one another directly or indirectly.
That makes the history more interesting and less tidy, and it is why a single computer inventor is so hard to name. When you see a headline or a museum label calling something “the first computer,” treat it as a claim with a particular definition behind it. Asking which definition is in use usually clears up most of the apparent disagreement, and it is a habit that serves well across the history of technology.