Radio astronomers are fond of sharing the story that WiFi was invented by radio astronomers. The tale is true, although the “special sauce” they brought to the problem was not radio astronomy itself but the expertise in signal-processing they developed while doing radio astronomy.

This is not a small win – the economic value of WiFi may exceed $3.5 trillion US dollars1, far more than global expenditure on radio astronomy since its invention.2 That said, when radio astronomers recount this tale it is often about the value of radio astronomy. But the key lesson is that when science pushes technological envelopes to achieve its goals the tools developed in the process are often useful in entirely unexpected settings.

So sometimes it can be a little like a Lotto shop bragging about selling last week’s winning ticket – if the lottery is honest, nothing can be gained from making a trip across town to buy a ticket there. The difference between science and Lotto is that a losing Lotto ticket is just a crumpled piece of paper whereas funding science always gets you the science you funded – but every so often it delivers exceptional, unforeseeable returns.

A more niche example is provided by string theory. Depending on who you listen to, string theory is either our best bet for “theory of everything” or a mirage chased by fundamental physics for the best part of forty years. But either way string theory has already revolutionised much of science – not the science itself but the ways in which many scientists communicate with each other and, often, the wider public.

Scientists have been publishing “papers” in “journals” since the 1600s and we still do. However, it takes months (and often years) for results to work their way through the editorial process, and in response preprints – drafts of papers3 – have long circulated inside the community. Once upon a time particle physicists would send pre-publication copies of new manuscripts to key institutions in the field who typically put them on a shelf in the library or the lunchroom. But that process moved at the speeds of paper mail and worked best for people at places that were likely to be on everyone’s mailing lists.

One of the hallmarks of string theory is its pace – particularly in the heady days of the 1980s and 1990s. New ideas launched into this tight-knit field spawn follow-ups in a matter of weeks and sometimes days, giving the field a particularly strong incentive to find efficient mechanisms to circulate new results.

In 1989 Joanne Cohn set up an email exchange for papers in her particular niche of string theory, sending electronic copies of preprints to a list of interested readers. In 1991 a colleague, Paul Ginsparg, automated the process, launching what has become known as the ArXiV in the process. From there it quickly branched out into the rest of fundamental physics and then to astronomy and astrophysics, maths and computer science and it is now (somewhat slowly) making inroads into biology and medicine.

Number of submissions to ArXiV per month – https://arxiv.org/stats/monthly_submissions

Today, the ArXiV adds around 250,000 preprints a year – roughly 5% of the 5 million or so articles that appear in journals or in conference proceedings. The difference