Wizard of Odds: Rolling the Dice at the World Science Festival

“Probability – what’s hard about that?” opened John Hockenberry, journalist and moderator for the World Science Festival’s Wizard of Odds panel discussion on Saturday, May 30. As many skeptics understand, that’s a seemingly natural question that can cause a lot of problems.

People aren’t always good at thinking about numbers intuitively, and might not automatically account for newly provided information. Hockenberry noted some folks’ disbelief when he tells them he’s the father of two sets of twins.

“What are the odds of that?” they often ask. “In my house,” Hockenberry responds, “100 per cent.”

Hockenberry’s story alludes to the idea of Bayesian statistics, which was introduced to the crowd in a prior, humorous animation. Bayesian statistics incorporates prior probabilities and additional information to refine potential outcomes as time goes on.

Unfortunately, that’s not much of a help when it comes to quantum mechanics. The first panelist, Masoud Mosehni, explained his work at Google on quantum computing, which takes advantage of the inherent probabilistic nature of particles.

quantum-computing-chipA Google quantum computing chip, from Business Insider

“It’s unlike anything you’ve experienced,” he said. Mosehni referred to the fan-favorite Schrödinger’s Cat analogy when describing how a tiny particle can be in a state of “superposition,” between two outcomes, until it’s measured. If perfected, a quantum computer could offer greatly increased data-crunching speeds compared to conventional machines, by spreading the work out over all the possibilities.

Mosehni’s research specifically aims to force superposition onto macroscopic objects, like metal rings. This can be done by cooling them to temperatures just above absolute zero, the point at which all atomic motion stops. Hockenberry worried how this might affect future smartphones, asking what the chance was that Google devices could leak liquid nitrogen and “freeze our nipples.”

“I would say zero,” Mosehni replied.

Second panelist Leonard Mlodinow, physicist and author of the book The Drunkard’s Walk:  How Randomness Rules Our Lives, tried to bring Bayes back.

“The scheme that Bayes came up with is still used today,” he said, pointing out that Mosehni’s employer utilizes Bayesian algorithms to find advertisements you might be more interested in, based on your browsing history.

Mlodinow then moved on to more familiar stories of misunderstood probability, like the Monty Hall problem. Cartoon depictions of three doors were projected behind the panel, and audience volunteer Lauren guessed that the big prize was hidden behind door number three. When the second door opened to reveal a gag prize, in classic “Let’s Make a Deal” fashion, Hockenberry asked if Lauren would like to stick with her choice or switch to door number one.

monty

“I’m gonna stay with three,” Lauren answered, drawing groans from the crowd.

“You know too much!” Mlodinow told them.

As the audience realized, the odds are actually more in your favor if you switch. Think about it. Your chance of guessing right initially is one of out three. Of course Monty Hall will reveal a gag prize, so that means, with the additional information in this case, the probability that door number #1 hides the prize has rocketed up to 50%. Better than your initial guess. Most people stick with their first choice, and that’s how game shows stayed in business through the 1980s.

Physician and genomics researcher Robert C. Green was up next, to talk about Bayesian statistics applied to biology, A.K.A “prior probability.” He spoke of misleading data, such as in a public service announcement that claims one in eight women will develop breast cancer. Green emphasized that probability is applied for your entire lifetime, and increases the older you get. For instance, a woman in her 50s only has a one in 44 chance of developing breast cancer in the next 10 years, but a woman in her 70s has a one in 26 chance.

Hockenberry asked how much Green’s field of genomics can help a given patient. You can probably guess his response.

“A lot, but almost not at all,” Green said.

Green used his own sequenced genome as an example, showing how numbers without context can unnecessarily panic people. The testing results show he’s twice as likely to be afflicted with celiac disease as the average person. That sounds scary if you don’t realize the chance of anyone getting celiac disease is vanishingly small, and doubling a tiny number doesn’t increase the total risk much. Overall, Green has less than a 1% chance of having his wheatcakes taken away.

Some disorders are easily pinpointed – if you have the genetic marker for Huntington’s disease, you’ve got Huntington’s disease – but most are like celiac, meaning environmental factors have to be taken into account when assessing probabilities. Testing for everything probably isn’t practical, and could lead to false positive diagnoses.

Mlodinow jumped in to tell the story of how he was diagnosed with HIV in the ‘80s, even though he didn’t partake in risky behaviors. His doctor didn’t know the rate of false positives, and waiting for additional results to come in caused a very anxious couple of weeks for Mlodinow.

“That’s a nightmare,” Hockenberry said. “Let’s talk about another nightmare – driverless cars.”

The stage was thus set for electrical engineer Richard Alan Peters, who said that cars are deterministic devices operating in a probabilistic world.

“Or like my room as a teenager,” Peters said. “Things could be anywhere.”

Peters explained that a driverless car constantly collects data and updates its assumptions based on what’s actually observed – always narrowing the probability of what’s going on around it and adjusting.

dogsFrom the BBC

A video of some truly remarkable adjustments was shown, in which what appeared to be robotic dogs were forced to navigate difficult terrain, and were even kicked, without breaking stride.

Hockenberry was more amazed by what wasn’t happening, asking “why they don’t sniff each other’s butts?”

Molodinow had the last word of the afternoon, suggesting that in the face of these realities, all of us should probably (see what I did there?) recognize that our own successes and failures are often governed by cold numbers, too, but that’s not necessarily a bad thing.

“We should realize that, be humble, and just chill a little,” he said.

Watch the entire presentation right here!

Leave a Comment


NOTE - You can use these HTML tags and attributes:
<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>