On 11 February 2016, the Laser Interferometer Gravitational-Wave Observatory (LIGO) and its sister collaboration, Virgo, announced their earthshaking observation of Albert Einstein’s ripples in spacetime. LIGO had seen the death dance of a pair of massive black holes. As the behemoths circled each other faster and faster, the frequency and amplitude of the spacetime waves they produced grew into a crescendo as the black holes became one. Then the new doubly massive black hole began to ring softer and softer like a quieting bell. The escalating chirp and ringdown is also a metaphor for public information flow about the discovery. It could have unfolded differently.

When scientists make a discovery, they must choose how to disseminate it. A big decision they must make is whether to reveal the results before or after peer review. Reveal before peer review—sometimes even before the paper is written—and the community can use the results right away, but there is an increased risk that problems will be found in a very public way. Reveal after peer review, and the chance of such problems decreases, but there is more time for a competitor to announce first or for rumors to leak. At Physical Review Letters (PRL), where I am an editor, we allow authors to choose when they want to reveal their results. The LIGO collaborators chose to wait.

Just before LIGO’s experimental run began in September 2015, the team held a vote on which journal they would pick if they made a discovery. They picked PRL. Five days after the vote, LIGO’s detectors seemed to hear the universe sing out for the first time.

Had LIGO just confirmed a 100-year-old prediction made by Einstein? Had they discovered the first black hole binary? Had they opened a new era of astrophysics? With the stakes so high, the collaborators wanted to keep their results secret while they determined if the results were real. It was unfortunate that some onlookers chose to publicize vague rumors when the internal vetting had just begun.

 

By early December the collaboration was convinced that the results were real, and LIGO spokesperson Gabriela “Gaby” González let me know that we would be receiving a paper from the group in mid- to late January. When she told me that they had convincingly observed gravitational waves, that it was not a test, and that the source was the merger of two huge black holes, my jaw dropped.

Gaby stressed LIGO’s desire for strict confidentiality, so for a month I told only one other person in the world: my fellow editor Abhishek Agarwal. By mid-January we had to bring others into the loop to prepare for the paper’s arrival, to review it, and eventually to publish it. To avoid information slipping out from a casual conversation or a glance at a screen, we used the code name “Big Paper.” (The code name for the second LIGO, announced in June, discovery was “Big Two.”) To the best of my knowledge no information leaked from us. Inside the LIGO team, for similar reasons, the discovery was referred to as “The Event.”

Big Paper on The Event arrived at PRL on the evening of 21 January 2016, and we immediately sent it out to experts for anonymous peer review. The referees, like everyone involved, were sworn to secrecy. Informed, unbiased advice is central to picking which papers are published and to improving those that are. In this case it was clear that the paper was important and interesting enough for PRL. As expected, the reviews were very favorable and conveyed the message that the paper would be an inspiration to physicists and astronomers alike.

As the time for the announcement drew closer, the rumors increased. In one case, a preprint was spotted on a printer, then a physicist emailed his whole department about the results, one tweet quoted the email, and a science reporter based an entire story on that tweet. The information was incomplete, though correct—except for the journal where the paper would be published. That reporter learned at the press conference that PRL would publish the paper and sheepishly congratulated me.

Meanwhile, we continued to protect the information from leaking. My son, who is a budding science reporter, texted me a few days before the announcement, asking if I’d seen the rumors. That led to an awkward phone call—I still couldn’t tell him about the discovery. When we ordered a celebratory cake for the editorial office, we avoided any mention of the result on the frosting, lest it lead to an information leak. It turned out that we were not being overly cautious: A tweet containing a picture of a cake at NASA’s Goddard Space Flight Center on the morning of the announcement did leak news of the discovery!¹ Confidentiality requires vigilance.

Everyone at LIGO’s press conference was given access to the PRL paper hours beforehand, on condition that they not publish their stories until after the announcement was made and the news embargo lifted. Actually, it might have been better in some ways had the press had access to the paper a little earlier, but that also would have increased the risk of the paper leaking prior to the announcement.

We had an agreement with the LIGO team to publish the paper online on 11 February at 10:45am Eastern Time, 15 minutes after the press conference was to begin. But I learned that morning from the reporters around me that the embargo was being lifted at 10:30am, and they planned to publish their stories then, which would create 15 minutes of pent-up demand for the PRL paper. So I found the spokesperson minutes before she went to the microphones and asked her if we could publish at 10:30. Gaby smiled and simply said yes.

After a few frantic emails, all the plans were changed, and at 10:30 we published the LIGO paper.² It didn’t help: The demand for the paper was still so great that our site crashed under a load of 10 000 hits per minute.³ After we added a slew of servers, our site came back up, and the paper was downloaded an unprecedented quarter of a million times on the first day.

The LIGO researchers had chosen to maintain confidentiality because they wanted their results carefully vetted before they went public. They also wanted the information to come from them, not from rumors. Although some of the information leaked before the announcement, they still did get the glory of presenting the full results to the world. And the ringdown phase has been impressive, as news of the result continues to spread far and wide.

Authors may have good reasons to announce their results prior to the completion of peer review—reasons that include competition from other groups, hope for informal community feedback, and desire to control the announcement and avoid weeks of rumors. But if authors choose that path, they should consider the possibility that peer review will turn up problems they did not think of, and they should tailor their announcement accordingly. Authors may instead choose to wait for the completion of peer review, especially when they have no concerns about competition. In such cases it is an even greater pity when rumors leak, because the leakers provide disincentive for such patience.

For LIGO, although much of the information leaked before the press conference, the researchers still had much to announce, probably in part because they had emphasized confidentiality. Announcing early makes sense in some cases, but the LIGO group made the right choice to wait.