Skip to main content
Philosophy, sociology and religion

Philosophy, sociology and religion

Discovering dark matter

02 Jun 2010 Robert P Crease

With the search for dark matter hotting up, Robert Crease proposes an experiment into the nature of discovery

Discovery channel

The discovery of dark matter – the mysterious, invisible substance believed to make up more than 80% of the matter in the universe – would be a key moment in 21st-century physics. Hardly surprising, then, that so much attention was given to a paper written last year by the members of the Cryogenic Dark Matter Search (CDMS-II) detailing their evidence for dark matter (arXiv:0912.3592v1). The CDMS-II collaboration is looking for evidence of collisions between weakly interacting massive neutral particles (or WIMPs) – a leading candidate for dark matter – and nuclei of germanium in a detector in a mine in Soudan, Minnesota. The detector is located 700 m underground to minimize background noise from neutrons produced in cosmic-ray collisions, which can mimic real WIMP signals.

The CDMS-II collaboration strongly promoted the paper, which it submitted to arXiv on 18 December 2009. Five days before, the group circulated an e-mail flagging the upcoming paper and announcing a pair of talks that it had scheduled for 17 December. The talks – one at Fermilab and the other at the SLAC National Accelerator Laboratory – were arranged to start simultaneously. One was broadcast live over the Internet. Given the unusual lengths the CDMS-II collaboration was going to create a record of who said what and when, it was – an outside observer might conclude – about to stake a claim for discovering dark matter.

“All the physics blogosphere is abuzz,” reported the physics blog Cosmic Variance in early December as the big day approached. A film crew making a documentary about dark matter recorded the event, which was also reported by the mainstream media, including the New York Times. “The excitement in the air is palpable,” wrote Cosmic Variance blogger JoAnne Hewett an hour before the seminar started. “It looks like a signal talk,” Hewett’s colleague confided in her as the talk began.

Got sigma?

Not for long. CDMS-II spokesperson Jodi Cooley revealed that the researchers had found only two events, compared with 0.5 expected from background, yielding a confidence level of about 1.3σ, or 21%. Physicists normally expect more – at least 3σ, or 99.73%. “The results cannot be interpreted as significant evidence for WIMP interactions,” Cooley admitted in her talk, “but we cannot reject the possibility that either event is signal.”

The blogosphere crashed. Many acted betrayed. “They should have brought in Geraldo Rivera to open the signal box,” laughed one, referring to the former US talk-show host known for his melodramatic style. Some bloggers suggested that the collaboration had hyped its results to secure funding for a planned upgrade of its detector. Others thought it did so to stake a discovery claim given that XENON100 – a more sensitive, xenon detector in the Gran Sasso lab in Italy – had already begun to report results.

The scale of the build-up and let-down was itself a “signal” that something unusual was happening. The CDMS-II episode, it seemed to me, could tell us a lot about the use of statistics in science.

“Big deal,” said one physics colleague to whom I excitedly mentioned my idea. “It’s part of the business. We grapple with this kind of thing every day.” Yet to philosophers and historians, “the business” contains interesting features that physicists usually take for granted. What elements shape the role of statistics in discovery? Do researchers in different disciplines seek different levels of confidence in “a discovery”? Does an astronomer, say, want firmer evidence than a psychologist? Does suspicion of experimentalists and their methods sometimes inflate the acceptable confidence level?

The critical point

In January I devoted this column to what I saw as ambiguities in the discovery of dark energy, announcements of which were published by two different groups at slightly different dates in 1997/1998. I used the episode to examine the connection between publication date and discovery, deciding that sometimes “discoveries are not simple, unitary events made by a specific person at a specific place or time”.

This month I would like to follow a suggestion by Adam Riess, a key member of one of the two dark-energy discovery groups. Historians usually discuss credit after a discovery is made, Riess pointed out to me. But thanks to the fact that WIMP search results will roll in with increasing data from different sources in the next few years, historians have a unique opportunity to assess a discovery as it happens, allowing them to test models and assumptions.

Suppose, for instance, that XENON100, super-CDMS or some other dark-matter search turns up evidence for WIMPs at a confidence level of 2σ, or 95% confidence; will that count as a discovery? What about 3σ or 99.73%; or 5σ or 99.9999%? If the latter, will the scientific community consider the findings at lower confidence levels to warrant a partial claim to having seen WIMPs? And on what grounds could it plausibly refuse?

Then flip it around: suppose the 5σ result refutes WIMPs consistent with the previous findings – will these earlier results then be viewed as statistical flukes, incorrect claims or errors? And if so, on what grounds?

Please send me your thoughts, and I will write a follow-up column about them. If a finding of better than 5σ for WIMPs is finally established, then we will be able to compare our results with the judgment of the scientific community.

Finally, I also welcome your examples of findings that looked like promising discoveries before vanishing as more statistics were acquired – as well as non-findings that grew into findings with more statistics. Such cases have interesting implications about the nature of discovery and about who, ultimately, deserves the credit.

• What result do you think would constitute a “discovery” of dark matter? How then should we view the CDMS-II findings? Do you know of “discoveries” that grew into non-discoveries with more statistics, or vice-versa? Send your responses to Robert P Crease at the e-mail below

• To find out more about the search for dark matter, don’t miss the following exclusive video interviews.

Deep exploits — the search for dark matter

Going underground — life inside the Boulby lab

Related events

Copyright © 2024 by IOP Publishing Ltd and individual contributors