Back when I was a postdoc, when you wrote a paper you would send the manuscript to a few friends for feedback before submitting it for publication. New preprints could often be found in the seminar room or the lounge of physics departments.

In the fall of 1989, there was a string theory conference at Rutgers University. I couldn’t go, so I asked a colleague, Paul Ginsparg, to share some of the papers with me. He emailed me some TeX files.

I thought others would be interested in the papers too. So I emailed them to a bunch of people. Most of us in the field knew each other.

Someone wrote thanking me and asked if I happened to have a recent paper by Ed Witten, a leader in the field. I didn’t have it, but I walked down the hall to Witten’s office and relayed the request. Witten said, ‘Sure,’ and he emailed me the file.

I started sending whatever preprints I had to everyone. When I asked, people sent me their papers to circulate. Then colleagues started sending me their papers on their own. And when people asked me for papers, I said yes and asked them to send me theirs too. I thought it should be a group effort, an exchange. I learned how to create an email list—it grew to include a couple hundred people. That was a big number at the time.

Many people were eager for leaders in the field, such as Witten, to see their work. As a result, the leaders were up-to-date on developments. By sharing preprints, people who were younger, less well-connected, lower on the totem pole, at less prestigious institutions, or in other countries got access more quickly than they otherwise would have. The rapid sharing was democratizing. I was motivated by a sense of community responsibility. And I was inspired in part by seeing this ethic in many of the people I worked with, especially senior colleagues.

Obviously, we were all busy calculating stuff too. Working in string theory was exhilarating and terrifying at the same time. It was like a freight train was barreling through, and as a researcher you were just trying to stay on the train. You didn’t want to wait months for a paper to arrive in your department seminar room, or longer for it to be published. The atmosphere in string theory was both close-knit and competitive.

There was this hunger, this interest in finding out what was going on and being able to use it and build on it. I knew what was going on down the hall, and I wanted to know what people were doing in other places. I would read every single paper before posting it. Being quick was important, and priority—establishing who was first—was important- Insights were easy to understand but terribly difficult to come up with.

Some funny things happened, too. For example, as emails were relayed between computers, you’d get messages reporting where they were. In Princeton, when it rained, the link to the outside world would sometimes go down. When it dried out, I would get tons of messages saying my emails had just gone out. I had to send longer papers out in two parts to fit them into the email length limits; otherwise they’d bounce back and I’d log on in the morning to find hundreds of error messages.

During the 1991 Gulf War, the link to Israel worked only at certain times of day. So all of a sudden I’d get messages when my emails arrived. Some physicists from the recently collapsed Soviet Union contacted me and asked for papers. I sent them many at once, and apparently it crashed their server; I had to send the papers more slowly.

In the summer of 1991, I was chatting with Paul [Ginsparg] at a conference in Aspen. He asked me about automating my distribution list. I thought it sounded like a cool idea, but I didn’t know much about programming. I told him to feel free.

A day or so later, Paul came back to me with some scripts he had written. Later that summer, when he was ready to launch, I mailed the group and told them to move over, and I transferred my list to him. After that, starting in August 1991, it was Paul’s thing. And it’s become amazing.