Favorable Odds: Startup Opens Eyes to Probabilities
You're more likely to be killed by a vending machine than a shark, say the researchers behind a soon-to-debut, semantic-based Web startup called Book of Odds. And, they add, you're more likely to die falling out of bed than win the lottery.
Of course, mileage varies if you're a vending machine repairman, a camera operator for Shark Week, or always sleep on the floor, but the idea behind this collection of 500,000 (and growing) odds statements "isn't to determine personal probabilities," says company founder Amram Shapiro. "It's much more valuable in the aggregate."
From the company's homepage, which features a short quiz you can take to get a feel for the content:
"Book of Odds will cover a wide range of topics including health, crime, politics, accidents, and relationships. Its consistent format will make it easy to understand. Any one odds statement may be used to better grasp another - the unfamiliar made more comprehensible by the familiar."
There's this disconcerting example from the company's Twitter feed: "Did you know that the odds a married man has ever had an affair are the same as the odds that a flight will be delayed?"
"It will be extensive: Over 50 staff-years of research have gone into creating a database of odds numbering in the hundreds of thousands, and over time it will expand into new topics, times, and geographies.
"It will be accurate and transparent so anyone can replicate or correct any odds statement.
"It will be engaging, fun to navigate, and full of surprising information and juxtapositions of facts."
Each odds statement will carry a "confidence level" notation derived from an assessment of the quality of the underlying data, Shapiro says.
Three years in the making, privately funded Book of Odds is expected to launch Oct. 14 but tomorrow will offer attendees at the Web Innovators Group meeting in Cambridge a look at the site and an invitation to join a closed beta.
"We felt that we could not only fill an important need, we suspected that it would be really, really fun," says Shapiro, who spoke with me Friday while readying for tomorrow's coming out party.
"One of the first companies to use the new Anzo Suite is Book of Odds Enterprises, Inc., a start-up company launching a one-of-a-kind Web site to make the odds of everyday life easily accessible. This groundbreaking site will leverage semantically linked and cross referenced content from hundreds of sources in order to create a fundamentally new reference source.
" 'The Cambridge Semantics' suite has allowed Book of Odds to build a rich semantic database linking content from a wide array of sources and formats in order to create a powerful statistical reference tool which will be accessible to the general public,' said Shapiro."
Before founding Book of Odds, Shapiro was a director at Arthur D. Little and Pittiglio, Rabin, Todd and McGrath. The startup's Chief Operating Officer is Louise Firth Campbell, another ADL alum. The company employs 20 and expects to derive revenue from a number of sources, advertising being primary among them.
I asked Shapiro to tell me the most common question he's received from those who have already seen Book of Odds.
"They ask how come no one did this before."
Here are some more sample odds statements via the company's twitter account:
The odds a person aged 13 or older will be diagnosed with HIV in a year are 1 in 4,389 (US, 2006).
The odds an accidental death will be due to a fall down stairs are 1 in 68.41 (US, 1999 - 2005).
Odds an adult using a public restroom was observed not washing their hands are 1 in 4.37.
The odds a child younger than 18 has ever been diagnosed with ADHD are 1 in 3.6 (US, 2007).
The odds a wildfire in Southern California is started by humans are 1 in 1.11.
The odds a person 18 or older will drink any type of coffee in a day are 1 in 1.75.
It's more likely that a pilot or crew member has nodded off during a flight than it is for a child to live with two parents.
Odds this site will be interesting and fun? 1:1.
Odds the company will be a success? We'll need more data to make that calculation.