CDMS rumors: What they'll find and what it'll mean

Trying to squash a rumor is like trying to unring a bell. -Shana Alexander

Around the internet, blogs are all abuzz that an experiment searching for dark matter, CDMS, has cancelled all of their upcoming announcements and will be holding a special press conference on the 18th (this Friday!) to release their latest findings.

Here's what you can expect. First off, here's how it works. They take a bunch of hockey-puck shaped detectors,

shield them at the bottom of a mine shaft deep beneath the Earth (in Soudan, MN),

and try to measure these very rare events of dark matter particles (which can pass right through the Earth) interacting with these detectors.

Now, other things can interact with these detectors besides dark matter, including neutrinos, radioactive particles from inside the mine, cosmic rays that get through the Earth despite the shielding, etc. Everything that isn't dark matter is what we call "background," and can mimic the signal that dark matter would give. Everything relies on your dark matter signal being noticeable above this expected background.

Well, they've already announced that their expected background, at this juncture, is 0.5 events. So let me tell you what the possible outcomes are, and what they'll mean if you're a responsible scientist.

Let's say you have a lottery where you have one-in-a-million odds of winning. If you buy a lottery ticket for 500,000 lotteries in a row, you'd expect, on average, to win one half of one lottery. In reality, though, each time you play, you have a one-in-a-million shot, and you have 500,000 shots. Here's how that breaks down after you've played all 500,000 times.

  • 60.65% of people who play 500,000 times never win.
  • 30.33% of people who play 500,000 times win the lottery once.
  • 7.58% of people who play 500,000 times win the lottery twice.
  • 1.26% of people who play 500,000 times win three times.
  • 0.16% of people who play 500,000 times win four times.
  • 0.02% of people who play 500,000 times win five times or more.

This is basically what CDMS did; they played a lottery where the odds of getting any single event is miniscule, but they played it so many times that they've got a good shot now. Of course, they're hoping to find something that is significantly different than what you'd expect if all you had was this background.

So the key question to ask is how many events will they announce? If they've found no events, that's extremely boring. If they've found between one and four events, that's not scientifically conclusive of anything other than your expected background! But if they've found something like 10 or 15 events, that's very interesting, and indicates that something significantly above your expected background is going on.

Regardless of what they announce, there are going to be reports that CDMS has found dark matter. I hope -- after reading this -- you'll ask yourself whether CDMS' findings were significant, or whether they're simply inconclusive results that are consistent with a background of stuff we already know is there, and nothing else is necessary. Check back on Friday, and we'll look at what they've found with a critical eye.

More like this

If I might make a meta-comment, I believe that every one of the rumor posts traces back to the original resonaances post. So there's really only one voice in the echo chamber.

By Andrew Foland (not verified) on 14 Dec 2009 #permalink

Not quite, to be fair.
It is more like resonaances let everyone else start speculating wildly by pointing back the him...

The CDMS crowd, in retrospect, went rather quiet around thanksgiving - stopped chatting about what was up etc.
I also heard independently they were, or were going to, submit to one of the glossies - clearly not Nature, though... ;-)

But, unless they found a new way to clean up background, or a new understanding of real signals, it is very hard for them to have more than three events, without violating previous upper limits.

So... most likely guess is they have a hint of a signal, probably in the 50-60GeV window if I understand their sensitivity, but that it is barely 3 sigma. Almost certainly not as high as 5 sigma.

We'll see soon enough.

0.5 events averaged over what time period? Radioactive decay alone is quite sporadic and I presume events will be governed by Poisson statistics, which would mean an awfully long wait to get any significant information out of the system. Ah, the fun of rare events - give me cosmic ray counting at sea level any day.

By MadScientist (not verified) on 14 Dec 2009 #permalink

How does the experiment distinguish between a neutrino impact and a wimp?

What if they found between 5 and 9 events?

This whole rumor attached to a publication in Nature on 18 Dec 2009 is odd since Nature publishes on Thursdays: 17 Dec 2009.
Maybe they meant Science which publishes on Fridays.

[quote]What if they found between 5 and 9 events?[/quote]

I'm personally hoping for 42.

Isn't it typical for p-values to be below 5%, not below .05%? That would indicate that three or more events might be scientifically interesting. Even if three events is deemed a little high for a p-value, certainly 4 or more events seems impressive. Perhaps there is something I don't understand about the way physicists interpret p-values.