In 2006, Garth Sundem and John Tierney published an equation in the NY Times that attempted to predict celebrity marriage crackups using a few metrics: age, fame, sexiness, etc. The pair recently modified the equation based on the evidence of the last five years and surprisingly, the equation is simpler.

What went right with them — and wrong with our equation? Garth, a self-professed "uber-geek," has crunched the numbers and discovered a better way to gauge the toxic effects of celebrity. Whereas the old equation measured fame by counting the millions of Google hits, the new equation uses a ratio of two other measures: the number of mentions in The Times divided by mentions in The National Enquirer.

"This is a major improvement in the equation," Garth says. "It turns out that overall fame doesn't matter as much as the flavor of the fame. It's tabloid fame that dooms you. Sure, Katie Holmes had about 160 Enquirer hits, but she had more than twice as many NYT hits. A high NYT/ENQ ratio also explains why Chelsea Clinton and Kate Middleton have better chances than the Kardashian sisters."

Garth's new analysis shows that it's the wife's fame that really matters. While the husband's NYT/ENQ ratio is mildly predictive, the effect is so much weaker than the wife's that it's not included in the new equation. Nor are some variables from the old equation, like the number of previous marriages and the age gap between husband and wife.