By Davide Castelvecchi
Claiming new discoveries about elementary particles requires a great
deal of care.
In that tiny, bizarre world, the laws of physics can never tell you
exactly what will happen—but they usually predict with great precision how
likely it is that things will go a certain way. Though very rare, freak
accidents, such as particles appearing out of nowhere, are possible.
In a typical BABAR
measurement, researchers look for very uncommon patterns of particle
decay. At the forefront of research, it is the rarest events that can
teach us new physics. Results come from months of painstaking data taking
and analysis, rather than from a single ‘Eureka!’ moment. "It’s usually
not the case that you look at one event and say ‘Wow!’" said Jeffrey
Physics Analysis Coordinator.
Rare particle decays can be compared to ‘special’ coins that don’t land
on tails or heads with equal probability. If you had a bag with a million
normal coins in it, and only one special coin—one that, say, always lands
on tails —how would you go about finding the special coin?
Say you pick a random coin out of the bag. If you toss it once, and it
lands on tails, that doesn’t say much, does it? What if it lands on tails
twice in a row? A regular coin has a 25% chance of doing that, so it’s not
too unusual. Such strings of deceptively favorable events are called false
positives, or, as particle physicists would say, ‘background’.
Now say you are a skeptic, and you toss your coin ten times—tails every
time. A little math shows that there was less than one chance in a
thousand for a regular coin to do that. Sounds more promising. But is ten
flips enough? At that point, if you stop looking you are vulnerable to
things you don’t anticipate as background.
To prevent themselves from jumping to conclusions too early, physicists
often employ a procedure called blind analysis. You decide ahead of time
how many coins you’re going to test—say, a million—and how many times you
want to toss each one—say 100. Even if a coin’s first 50 tosses give all
tails, you force yourself to keep on tossing until the end. After weeks
and weeks of tossing coins and recording the results, you finally look at
your records, and see something like this: most coins behaved roughly like
normal coins, some landed more often on tails than heads, some the other
way around, but just one coin landed on tails 99 times.
For a normal coin, that would happen about once in 10 billion billion
billion times, so a false positive is extremely unlikely. On the other
hand, you should expect even the special coin to land on heads once in a
while. Because of quantum accidents, there’s no such thing as absolute
For each BABAR
measurement, the moment of truth is scheduled to come at the end of the
blind analysis procedure, in what people often call the ‘unblinding
party.’ Unblinding parties usually involve a small group of graduate
students and postdocs, staying up late looking at someone’s laptop in the
Research Office Building kitchen. "If there’s a ‘Eureka!’ moment, that’s
it." Richman said. "You look at your final output file and it could be:
‘Bang! There is a big signal!’