Harvard professor Peter Galison (he’s actually one of only 24 University Professors, a special honor) is opening a conference on author attribution in the digital age.

He points to the vast increase in the number of physicists involved in an experiment, some of which have 3,000 people working on them. This transforms the role of experiments and how physicists relate to one another. “When CERN says in a couple of months that ‘We’ve found the Higgs particle,’ who is the we?”

He says that there has been a “pseudo-I”: A group that functions under the name of a single author. A generation or two ago this was common: The Alvarez Group,” Thorndike Group, ” etc. This is like when the works of a Rembrandt would in fact come from his studio. But there’s also “The Collective Group”: a group that functions without that name — often without even a single lead institution.” This requires “complex internal regulation, governance, collective responsibility, and novel ways of attributing credit.” So, over the past decades physicists have been asked very fundamental questions about how they want to govern. Those 3,000 people have never all met one another; they’re not even in the same country. So, do they stop the accelerator because of the results from one group? Or, when CERN scientists found data suggesting faster than light neutrinos, the team was not unanimous about publishing those results. When the results were reversed, the entire team suffered some reputational damage. “So, the stakes are very high about how these governance, decision-making, and attribution questions get decided.”

He looks back to the 1960s. There were large bubble chambers kept above their boiling point but under pressure. You’d get beautiful images of particles, and these were the iconic images of physics. But these experiments were at a new, industrial scale for physics. After an explosion in 1965, the labs were put under industrial rules and processes. In 1967 Alan Thorndike at Brookhaven responded to these changes in the ethos of being an experimenter. Rarely is the experimenter a single individual, he said. He is a composite. “He might be 3, 5 or 8, possibly as many as 10, 20, or more.” He “may be spread around geographically…He may be epehemral…He is a social phenomenon, varied in form and impossible to define precisely.” But he certainly is not (said Thorndike) a “cloistered scientist working in isolation at his laboratory bench.” The thing that is thinking is a “composite entity.” The tasks are not partitioned in simple ways, the way contractors working on a house partition their tasks. Thorndike is talking about tasks in which “the cognition itself does not occur in one skull.”

By 1983, physicists were colliding beams that moved particles out in all directions. Bigger equipment. More particles. More complexity. Now instead of a dozen or two participants, you have 150 or so. Questions arose about what an author is. In July 1988 one of the Stanford collaborators wrote an internal memo saying that all collaborators ought to be listed as authors alphabetically since “our first priority should be the coherence of the group and the de facto recognition that contributions to a piece of physics are made by all collaborators in different ways.” They decided on a rule that avoided the nightmare of trying to give primacy to some. The memo continues: “For physics papers, all physicist members of the colaboration are authors. In addition, the first published paper should also include the engineers.” [Wolowitz! :)]

In 1990s rules of authorship got more specific. He points to a particular list of seven very specific rules. “It was a big battle.”

In 1997, when you get to projects as large as ATLAS at CERN, the author count goes up to 2,500. This makes it “harder to evaluate the individual contribution when comparing with other fields in science,” according to a report at the time. With experiments of this size, says Peter, the experimenters are the best source of the review of the results.

Conundrums of Authorship: It’s a community and you’re trying to keep it coherent. “You have to keep things from falling apart” along institutional or disciplinary grounds. E.g., the weak neutral current experiment. The collaborators were divided about whether there were such things. They were mockingly accused of proposing “alternating weak neutral currents,” and this cost them reputationally. But, trying to making these experiments speak in one voice can come at a cost. E.g., suppose 1,900 collaborators want to publish, but 600 don’t. If they speak in one voice, that suppresses dissent.

Then there’s also the question of the “identity of physicists while crediting mechanical, cryogenic, electrical engineers, and how to balance with builders and analysts.” E.g., analysts have sometimes claimed credit because they were the first ones to perceive the truth in the data, while others say that the analysts were just dealing with the “icing.”

Peter ends by saying: These questions go down to our understanding of the very nature of science.

Q: What’s the answer?
A: It’s different in different sciences, each of which has its own culture. Some of these cultures are still emerging. It will not be solved once and for all. We should use those cultures to see what part of evaluations are done inside the culture, and which depend on external review. As I said, in many cases the most serious review is done inside where you have access to all the data, the backups, etc. Figuring out how to leverage those sort of reviews could help to provide credit when it’s time to promote people. The question of credit between scientists and engineers/technicians has been debated for hundreds of years. I think we’ve begun to shed some our class anxiety, i.e., the assumption that hand work is not equivalent to head work, etc. A few years ago, some physicists would say that nanotech is engineering, not science; you don’t hear that so much any more. When a Nobel prize in 1983 went to an engineer, it was a harbinger.

Q: Have other scientists learned from the high energy physicists about this?
A: Yes. There are different models. Some big science gets assimilated to a culture that is more like abig engineering process. E.g., there’s no public awareness of the lead designers of the 747 we’ve been flying for 50 years, whereas we know the directors of Hollywood films. Authorship is something we decide. That the 747 has no author but Hunger Games does was not decreed by Heaven. Big plasma physics is treated more like industry, in part because it’s conducted within a secure facility. The astronomers have done many admirable things. I was on a prize committee that give the award to a group because it was a collective activity. Astronomers have been great about distributing data. There’s Galaxy Zoo, and some “zookeepers” have been credited as authors on some papers.

Q: The credits are getting longer on movies as the specializations grow. It’s a similar problem. They tell you how did what in each category. In high energy physics, scientists see becoming too specialized as a bad thing.
A: In the movies many different roles are recognized. And there are questions of distribution of profits, which is not so analogous to physics experiments. Physicists want to think of themselves as physicists, not as sub-specialists. If you are identified as, for example, the person who wrote the Monte Carlo, people may think that you’re “just a coder” and write you off. The first Ph.D. in physics submitted at Harvard was on the Bohr model; the student was told that it was fine but he had to do an experiment because theoretical physics might be great for Europe but not for the US. It’s naive to think that physicists are Da Vinci’s who do everything; the idea of what counts as being a physicist is changing, and that’s a good thing.

[I wanted to ask if (assuming what may not be true) the Internet leads to more of the internal work being done visibly in public, might this change some of the governance since it will be clearer that there is diversity and disagrement within a healthy network of experimenters. Anyway, that was a great talk.]

Categories: science, too big to know dw