Every email address on the planet is built around a symbol that, before 1971, had no particular connection to computing at all. The @ sign had existed for centuries as a commercial shorthand — used in accounting and invoices to mean "at the rate of," as in "10 units @ $5 each" — and it earned a permanent spot on typewriter and computer keyboards for exactly that reason. Nobody designed it with networked communication in mind, because networked communication as we know it didn't exist yet.
A practical, unglamorous choice
The person who put it to its now-famous use was Ray Tomlinson, an engineer at Bolt Beranek and Newman (BBN), a Massachusetts research firm doing contract work on ARPANET, the government-funded precursor to the modern internet. In late 1971, Tomlinson was building on an existing program called SNDMSG, which let users on the same computer leave messages for one another, and adapting it so a message could be sent to a user on a completely different machine elsewhere on the network.
That created an obvious problem: how do you write an address that identifies both a specific person and the specific machine they're on, in a way a computer program can parse unambiguously? Tomlinson needed a character to separate the two halves of the address — the username and the host computer's name — and it had to be a character that would never plausibly appear inside a person's actual username, or the whole system would break the first time someone's name collided with the separator.
Looking at his keyboard, Tomlinson noticed @ wasn't being used for anything in the systems he was working with. It wasn't part of any name. It had a clear, pre-existing meaning — "at" — that happened to fit perfectly: a message addressed to user@host reads naturally as "this user, at this host." By his own account, it was a practical decision made quickly, not a symbolic or philosophical one.
A choice that never had to change
Tomlinson sent the first message across a network using this addressing format between two machines at BBN, testing the system on himself before anyone else used it. He later admitted he couldn't actually remember the content of that first networked message — by his own description, it was something entirely forgettable, since he had no idea at the time he was doing anything significant. He was just testing whether the program worked.
There was no committee sign-off behind any of this, and no marketing plan. ARPANET at the time was a small, informal community of researchers free to build and test small utilities like this on their own initiative, and Tomlinson's networked mail program spread simply because other people on the network found it useful and started copying the same convention themselves. Within ARPANET's relatively small user base, the username-at-host format became a shared, unofficial standard almost immediately, well before anyone wrote a formal specification down.
What's remarkable isn't the invention itself so much as its durability. Half a century later, and across billions of email accounts and countless usernames on social platforms that later adopted @ as a way to mention or tag someone, Tomlinson's quick, practical fix for a parsing problem has never needed to be replaced. No competing standard ever displaced it, no redesign ever proved necessary. A single symbol chosen mainly because it happened to be idle on a keyboard in 1971 became, arguably, the most universally recognized punctuation mark on the internet — familiar in essentially every written language and every country with an internet connection, doing the exact same simple job it was picked for on a Tuesday afternoon at BBN.