by Mark Bowen
The more convincing of the two events was named the Peacock event (see photograph 14). As Francis wrote in an essay that was included in Best American Science Writing 2000,
Hardtke’s screen showed a faint blue line streaking diagonally across columns of black dots. Most of the dots, each of which represented a photomultiplier sunk in the ice, were small and black. But a few, clustered along the line, were blue or green or red, and two, near the beginning of the line, were bright orange and very large. At five in the morning on October 12, 1997, the diagram told us, a neutrino—one of nature’s smallest and most elusive elementary particles—had entered the earth in the middle of the Pacific Ocean, between Midway Island and the Aleutians, hurtled straight through the planet, and collided head-on with a proton on the underside of the Antarctic ice. Two kilometers beneath the surface, our grid of photomultipliers had picked up a subatomic spark from that collision as it flew upward through the ice and flared past them for about a microsecond.
Gary was first to “leak” the news. That afternoon, he e-mailed Robert Schwarz, who was about to witness his second sunrise in a row at Pole:
Hi Bert, well I wanted you to know asap—looks like we found real up-going neutrino events in the B10 data we took last winter … so it works!… Francis was really worried that we were never going to see any good events, and what he was going to tell NSF today. Well [he] saw them for the first time during the showing of the events to NSF—so he is very excited, and the NSF chap was very impressed …
About half an hour later, Albrecht sent an e-mail with a link to an image of the event to the entire collaboration—and was met, strangely enough, with resonant silence. There were only two responses, one from John Jacobsen, which was somewhat skeptical (Gary disabused him of that), and the other from the normally skeptical Christian Spiering, who went the other way, writing “Boah—das sieht ja wirklich irre aus,” which Albrecht translated for the rest of the Madison group as “Woooow—that looks really crazy [incredibly good].”
* * *
On that very day, amazingly, Fred Reines died. Francis received the news by e-mail from two of the great man’s Irvine protégés and forwarded it to the rest of the collaboration.
* * *
Two weeks later, on September 11, the collaboration received an e-mail under the title “Have a look at this” from Eva Dahlberg, a grad student in Stockholm. It directed them to an image of a near dead-vertical muon that had zoomed up through the detector, right next to one of the strings, lighting up each in succession, at 97 percent of the speed of light. That would be 25 percent faster than light itself could have traveled through the ice, given its index of refraction.
Eva was also met with silence. Only Christian responded. He allowed himself one moment of ecstasy: “Oh oh oh oh! Eva, this is the absolute hammer,” and swiftly reverted to his habitual skepticism: “… or could it be cross-talk?”
Cross-talk was an irritating byproduct of AMANDA’s analog electronics in which the electrical pulse produced when light hit an optical module lower down on a string sometimes “talked” to the higher modules on the same string as it passed by them on its way to the surface. This caused them to send out secondary pulses that were hard to distinguish from the real thing. Gary Hill observes that “cross-talk would come to dominate our lives later, but this event was surely not cross-talk.” The fake, cross-talk signals moved up the string more slowly than muons did, while Eva’s event traveled at just the expected speed. The fake pulses also had a different shape than those produced by the light emitted by a muon.
* * *
The watershed moment for the collaboration as a whole came on the first day of the next collaboration meeting, which took place in Madison and began on a Sunday, November 1. Everyone was anxious to see the latest results, so they scheduled the B10 analysis session for Sunday afternoon.
By now, the Madison group had processed sixty-one days’ worth of data (all odd days, the even days remained blind). They presented the results in three back-to-back talks.
In those days, the resident sceptics in the collaboration were Steve Barwick and Ralf Wischnewski, Christian Spiering’s right-hand man in Zeuthen. Christopher Wiebusch was also quite vocal, but his motivation seemed to be less about skepticism than about demonstrating how smart he was. The dynamic during science talks was that these three would sit, or even stand, in separate corners of the room and hector the speaker—especially if he or she had good news to report. Although the haranguing could be prolonged and tedious at times, most of the members of the collaboration agreed that it was valuable. Skepticism is the very basis of good science, after all. Steve was especially sharp.
Albrecht gave the first talk. He presented a loose set of automated cuts that netted nineteen events, which he had then scanned by eye to identify nine high-quality neutrino candidates. Steve and Ralf groused a bit as Albrecht spoke: it isn’t kosher to cherry-pick data by eye, because it allows you to find what you want to find.
Ty walked to the lectern next, and as he began to speak a rapt silence fell upon the room. Even Steve and Ralf were rendered speechless. Ty presented an entirely automated analysis with a stricter set of cuts than Albrecht’s, yielding the same nine events. The Peacock event cropped up in the first thirty days and the Eva event in the second thirty-one. Since his approach was run entirely by computer—no biased human eyes involved—it was the first indication that they had a working instrument in their hands.
There was some nitpicking, but most of the questions were respectful and complementary, pointing out small concordances, for example, that supported Ty’s conclusions.
Gary closed by presenting a Monte Carlo predicting that the instrument should have seen, on average, thirteen events in sixty days: reasonable agreement. It was a complete package, and the three talks were notable for their clarity and cohesion. I have noticed over the years that lucidity and simplicity of argument are hallmarks of the presentations from Madison.
“In my mind,” writes John Jacobsen, “these were the first absolutely compelling high-energy neutrino events found in AMANDA—compelling in that you could just look at them and see what they were without having to squint too hard, mathematically or otherwise. Getting our act together to turn the 246 raw 1997 data tapes into those gold-plated events required not just the larger B10 detector, but also for everyone to ‘level up’ both technically (software/analysis) and to work better as collaborators.” In view of the intense competition and politicking that would follow, it is important to keep the truth of John’s last point in mind. The Madisonians were not working in a vacuum. They had synergized tools and insights from around the collaboration. In a large science enterprise of this sort ideas are always “in the air.” It is never easy to pinpoint exactly where a breakthrough comes from—although scientists themselves don’t always remember that.
Over dinner in an unpretentious Italian restaurant on the evening of the presentations, Francis told me that he was not only thrilled at the new state of affairs, he was greatly relieved. “I can tell you I spent a lot of sleepless nights in the last ten years thinking … it was clear who was going to take the blame in Washington if this failed. There should be some fairness, right? I mean … I would have never gotten a cent anymore.”
He then went out of his way to give credit to “the original DUMAND people.” “I mean the vision they had!… You know, every sane experimentalist would never touch this subject as being too hard and too extravagant, too risky. They took the decision that they were actually going to do this, and that was very brave. Everybody after that is kind of a follower, right?”
Synchronistically perhaps, John Learned was passing through Madison just that week. “This is the first time that there’s data that is really, I think, convincing from AMANDA that shows that it’s probably going to work,” he told me. “So this is a major breakthrough. There’s no real science in what they’ve seen, but it’s a technology demonstration, and … they’re, you know, manfully trying to go
ahead and get out a little bit of science from it. But the importance is that—it’s like the bit about Dr. Johnson’s dog; it’s not that it speaks badly, but the fact that it speaks at all that’s important.… I think some of us have been convinced that it was gonna work for a while, but there’s nothing like seeing the cards on the table.”
* * *
You might be surprised to hear that some physicists are actually afraid of discoveries. They find them upsetting. A breakthrough changes the way they’ve been doing business. It knocks them out of their set pattern of plodding along and perfecting things, finding fault, being skeptical. I’m not sure that this is more prevalent in high-energy physics than it is in other fields, but I suspect that it might be. For these people are dealing with exceedingly esoteric concepts and their results are usually expressed in statistical terms. The first glimmer of a discovery usually reveals itself as a small effect that has risen ever so slightly above a sea of noise. You have to be careful not to fool yourself. After the initial excitement, a fear of exposure sets in. What if you got it wrong? You don’t want to embarrass yourself in public. And this has happened many times.
There was a range of responses inside the collaboration. On one end were the Madisonians, who were quite sure that this was the real thing, and at the other were the Germans. Once Ralf and Christian got over the shock of hearing such good news, they dug in their heels and refused to accept it. At a basic level, it seemed to be a kind of existential pessimism. “It’s not paranoia,” suggests Francis, “it’s that they feel that they don’t deserve it when good things happen, so they get very excited about bad things.… We know things must be going well when the Germans get excited.”
Ralf, in particular, seemed to revel in bad news. This is not to say that he didn’t make important contributions. His primary responsibility was the data acquisition system down on the Ice, and in that role he made sure that the instrument was capable of taking data every year. (On the other hand, he was about as excitable as Steve Barwick, and at the tail end of the season he and Steve were usually at Pole together, trying to complete their respective tasks—Steve’s being calibration—so the last few days before station closing tended to be excruciating for the winterovers.) But his continual harping on everything that was wrong with the instrument crossed the line into sheer negativity at times.
Competitiveness was also a factor, and there was probably an admixture of arrogance as well, since the Germans had been working on Baikal for so many years that they tended to think of everyone else in AMANDA as newcomers. Christian has also made enormous contributions over the decades, but it’s funny how it has always seemed unclear exactly where his allegiance lies. While he and Ralf seemed to have no problem accepting Baikal’s up-going events, which had been found with a smaller, eight-string detector, and while they were okay with the questionable events from AMANDA-B4—in fact, Christian took the lead in writing the B4 journal article—they were absolutely convinced that most of the “Madison events,” as they called them, were fake. And it seemed that Christopher was incensed that another group had stolen the lead again.
Some collaborations work on the dictatorship model (Carlo Rubbia comes to mind), but AMANDA, in its anarchism, was inherently democratic. And Francis, furthermore, had no interest in dictatorship. He has an overriding belief in the power of diplomacy and getting along. This can sometimes be taken as superficiality, but it has also worked wonders in the collaboration over the decades. The harsher and more aggressive people have slowly calmed down or wandered away for one reason or another, and nowadays most everyone in IceCube says it’s the friendliest collaboration they’ve ever worked in. Francis points out that it’s the only collaboration he knows of “where everybody can have dinner with anyone else in the evening.… In most collaborations there are people who explicitly won’t sit at the same table.”
Besides, disagreement has its uses. The journal article announcing a discovery will always be more solid and fully realized if it’s been vetted internally, all views have been aired, and consensus has been forged. After ten years of dogged effort, Francis knew better than most that detecting a ghostly particle in the presence of an enormous down-going background with a detector they couldn’t even see was a genuinely difficult business. A sanity check would not only help, it would serve constructive political ends. Everyone realized that the Germans were bound to run their own analysis again anyway, so Francis, ever the peacemaker, beat them to the punch. He suggested they go ahead.
This “adversarial scrutiny” approach (Ty DeYoung’s phrase) is common in complicated physics experiments. It’s basically about making sure that results are reproducible. This ideal is not always reachable in one-of-a-kind experiments that cost millions or billions of dollars. Francis points out that the Nobel Laureate Samuel Ting, who is Chinese by heritage, always has a Chinese group do one analysis and “the rest of the world” do a second, and instructs the two groups not to talk to each other. The physicists who discovered the Higgs boson went a step further and built two entirely separate detectors managed by separate collaborations, consisting of thousands of people. This is a Machiavellian tactic, which ought to be managed benevolently, however; and Francis and Christian did not do that. They set up competing analyses, not separate ones.
The reality was that the Germans couldn’t possibly run a separate analysis, because they were using many of the same tools as the Americans and communicating with them almost every week. The dynamic that arose when they got underway was that the Americans, mainly Ty and Gary, were put in the position of defending a claim that they thought was correct, while the Germans, mainly Christopher, were trying to tear it down. Now that emotions have cooled, Ty will admit that the scrutiny from Zeuthen helped a bit. But the truth is that it caused him, Gary, and Christopher enormous anguish for quite a few years. Only recently have Ty and Gary been willing to talk about it, and Christopher still hasn’t opened up.
Probably because he was the instigator. Christopher’s criticisms took on a personal tone, they were aimed at getting under the Madisonians’ skin, and they continued, actually, long after this particular bone of contention had been chewed to bits. For several years, he took pot shots at Gary and Ty in every possible forum: collaboration meetings, e-mails, phone calls, and at first they responded defensively. Without going into the details, let’s just say that there was regrettable behavior on both sides and that it is surprising that neither Francis nor Christian put a stop to it. Eighteen years later, they still chuckle at the pain their little management experiment caused, while the scars on their students remain just below the surface.
Once the Germans made progress and began dishing out their criticism, Gary and Ty got their dander up and began competing, too. And while this did spark a tremendous burst of creativity, it also had the ironic effect of making the two analyses only more intertwined. Ty says, “There was a lot of cross-pollination, since everyone was forced to adopt every new tool as soon as it came up or fall behind in the arms race.”
By January 1999, the Madison group had run all the odd days from 1997 through their constantly improving automated procedure and found seventeen neutrino candidates, one of which they identified by eye as being fake. By adding an automated improvement that had been developed in Zeuthen, they showed that they could eliminate the fake and thus improve the method. By summer they had fifty neutrino candidates.
Astoundingly, for more than a year, the Germans insisted on ignoring the progress in Madison and continued to carp about the seventeen events that had been revealed in January. They attempted to take the high ground by choosing not to scan events visually or even use real data to develop their methods! They relied entirely on Monte Carlos, using a simulated instrument with a simulated atmospheric neutrino beam, and applying it to the real data without looking at the data in advance. This way, even though they were aware of the Madison events, they could argue (speciously) that they were keeping themselves blind and could look at both the odd and even
days.
In retrospect, the Zeutheners were demonstrating their lack of experience in high-energy physics. Since the mainstays, Christian and Ralf, had come of age behind the Iron Curtain, it seems that they were unaware of the standard techniques in the field. The Madisonian way of doing things is the accepted approach in most high-energy physics experiments and in IceCube today. When the collaboration has a new set of data or a new question they would like to ask of an old set, they un-blind a so-called burn sample, a small fraction of the dataset, and develop their analysis methods with that. Then they freeze the methods and run them on the rest of the data, which has been kept blind. A discovery so obtained has the highest possible statistical value, because the question has been asked only once without a previous look at the data. If the method is changed and run a second time on previously un-blinded data, any discovery will have less statistical significance. This gets into the realm of a priori and ex post facto statistical inference, a slippery subject about which the experts in IceCube still argue until the cows come home.
* * *
No one argues that Christopher is not a fine physicist. He and two young Zeuthen colleagues produced a solid competing analysis by fall. Running it on both the odd and even days, they found 116 neutrino candidates and—crucially, one would think—confirmed the nine-month-old Madison analysis that had come up with seventeen.
But Christian and Ralf continued to dig in their heels, and Ralf remained obsessed with the seventeen events. At a meeting that took place in February or March of 2000, five or six months after Christopher and his colleagues had confirmed the Madison analysis, Ralf dragged an audience through a minute examination of the seventeen and explained how each could have been fake. Some of his arguments were so over the top that even Christian didn’t buy them. He was so buffaloed by the Eva event, for example, which was probably the most unassailable neutrino candidate in the sample, that his imagination failed him and he ascribed it to random noise.