Read It Sideways

Before there was an internet there were bulletin boards inside single institutions, and one of the busiest belonged to the computer science department at Carnegie Mellon. People posted questions, arguments, notices and jokes to it, in plain text, on terminals with green characters on black.

In September 1982 somebody started a physics thread about what happens inside a lift in free fall. It ran on through helium balloons, pigeons and candles, and eventually somebody posted a warning that a lift had been contaminated by a mercury spill.

It was not true. It was a joke, made in the voice of an official safety notice, on a board where official safety notices also appeared.

The problem, stated exactly

People believed it. Rudy Nedved had to post to say it had been a joke, and the department was left with a question that is easy to state and was, in 1982, genuinely open.

On a shared channel carrying both serious announcements and jokes, in plain text, with no voice and no face, how does a reader tell the difference?

Neil Swartz proposed the obvious answer. Adopt a convention of putting a star in the subject field of any notice which is to be taken as a joke.

That is a perfectly sound engineering solution. It is out of band, it is unambiguous, and it does not disturb the text. A short argument followed about which character should carry the meaning. Somebody suggested a percent sign for bad jokes. Somebody suggested a bracket construction meant to resemble a pair of laughing lips.

11:44

At a quarter to noon on 19 September 1982, Scott Fahlman posted this.

I propose that the following character sequence for joke markers: :-) Read it sideways. Actually, it is probably more economical to mark things that are NOT jokes, given current trends. For this, use :-(

Two things about it are worth separating.

The first is the mechanism, and it is different in kind from every other proposal in the thread. A star in the subject line is a flag. It says this item belongs to the joke category, and you have to know the convention to read it. Fahlman's three characters are a picture. A colon, a hyphen and a bracket are a pair of eyes, a nose and a mouth, and once somebody has told you to turn your head you cannot unsee it. It does not require a lookup table. It requires a neck.

The second is that the message is itself a joke, in the second half, and it is a fairly sharp one about the tone of the board. Most people who quote this post quote the first line and stop.

What happened next

It spread at the speed of a good idea on a small network. By 10 November, James Morris at Carnegie Mellon was writing to Xerox PARC to explain it, and by then the community had already produced variants, including a form for a stupid question and a mark for an angry message.

That is the pattern of a living convention rather than a decree. Nobody standardised it. Nobody had the authority to. It propagated because each person who saw it could reconstruct the rule from one example.

Within a few months it had reached most of the universities and laboratories on the network, which at that point was a list you could print.

The twenty missing years

Fahlman did not keep a copy. Neither did anybody else, particularly.

By the 1990s the smiley was on every screen on earth and the original post was gone. It had scrolled off a board on a machine that had been decommissioned, and what survived of that machine was backup tape, in formats nobody had needed to read for a decade, in boxes.

In 2002 a search was mounted for it. Mike Jones, then at Microsoft Research, put up funding, and staff at Carnegie Mellon went through the old tapes. They found it, with the timestamp intact, and the date was confirmed as 19 September 1982.

The Guinness records recognise it as the first digital emoticon. Fahlman's own account of the episode is on his university page, along with the reconstructed thread, and he has been consistently careful about the claim, which is that he proposed a convention on one message board and it caught on, rather than that he invented the idea of drawing a face out of punctuation, which people had done in print before.

Why the problem is still with us

The reason to bother with this story is not the character sequence. It is the problem it was solving, because the problem never went away and has arguably become the defining difficulty of written communication at scale.

Text carries no tone. Every convention we have built since is an attempt to put the tone back: the emoticon, then the emoji, then the reaction, then the caps, then the ironic full stop, then the tag that says just kidding because the sentence alone will be read straight by somebody.

The mercury spill happened on a board of a few hundred people who mostly knew each other and shared a profession. The same failure now happens on platforms of a few billion people who do not, where the joke and the announcement look identical and the consequence of misreading is larger than a department's confusion.

Fahlman's fix worked because everybody on that board wanted to be understood and everybody agreed on what the mark meant. Neither of those conditions is reliably true any more.

The three characters survived. The conditions that made them work did not.

Sources

• Scott E. Fahlman, "Original bboard thread in which :-) was proposed", Carnegie Mellon University (the recovered thread of September 1982, beginning with the physics discussion of a lift in free fall involving helium balloons, pigeons and candles; the hoax warning about mercury contamination and Rudy Nedved's clarification that it was a joke; Neil Swartz's proposal of a convention of putting a star in the subject field of any notice to be taken as a joke; the subsequent suggestions of other characters; and Fahlman's message of 19 September 1982 at 11:44 proposing the character sequence for joke markers, telling readers to read it sideways, and adding that it is probably more economical to mark things that are not jokes given current trends).
• Scott E. Fahlman, "The Birth, Spread, and Evolution of the Smiley Emoticon", 2021 (his own account of the origin, the spread of the convention to other universities and laboratories on the network within months, the derivative forms that appeared, and his qualifications about what he does and does not claim to have invented).
• Carnegie Mellon University, "Smiley Lore :-)" (the university's account of the proposal, its date and time, and its dissemination, including James Morris writing to Xerox PARC on 10 November 1982 with the convention and the variants then in circulation).
• The Register, "First ever smiley found preserved" (the loss of the original message, the search of Carnegie Mellon's old backup tapes funded by Mike Jones of Microsoft Research, and the recovery of the post in 2002 with its date confirmed as 19 September 1982).
• Guinness World Records, "First digital emoticon" (the recognition of the 19 September 1982 posting as the first digital emoticon).