View original NYT version   (current day only -- requires connection)

February 8, 2000

ESSAY

Point, Counterpoint and the Duration of Everything


Related Articles
  • Science Times Columns

    Forum

  • Join a Discussion on Science in the News
    By JAMES GLANZ

    Like many journalists who write about science, I have received what might be called the delta-t lecture from Dr. J. Richard Gott, a folksy and engaging professor of physics at Princeton. The lecture, which revolves around survival times, or delta-t's, explains Dr. Gott's theory that the likely future duration of anything, from the human race to a musical playing on Broadway, can be predicted. All you need to know is the time that has passed since the entity in question came into being until the moment when you asked, "How long will this last?"



    Laura Pedrick for The New York Times
    How long till the White Sox win another World Series? Not in the next two years, but before 5077, says Dr. J. Richard Gott, a physics professor who created a predictive formula.
    Near the end of the lecture, which lasted until the small hours of a summer night in Princeton in 1996, I asked Dr. Gott to predict how long it might be until the Chicago White Sox won another World Series, knowing that the last time they had done so was in 1917. After a quick calculation, he said there was a 95 percent chance that the following statement was true: the White Sox would not win a World Series in the following two years, but they would win one before the year 5077 if the team still existed then.

    That was an awfully wide delta-t, even for a Sox fan. And despite his having published the theory in the journal Nature in 1993, I worried that there might be something amiss in his reasoning and never wrote about it.

    Now I wish that I had. Last July, Timothy Ferris, a science writer, published a lucid piece in The New Yorker, titled "How to Predict Everything: Has the Physicist J. Richard Gott Really Found a Way?" The piece was read by Dr. Carlton M. Caves, a longtime New Yorker subscriber, who also happens to be a physicist studying information theory and statistics at the University of New Mexico.

    Motivated by the article, Dr. Caves did a remarkable thing: he wrote a mathematical rebuttal to the theory, which will be published by a British academic journal, Contemporary Physics. Although Dr. Gott does not accept many of Dr. Caves's conclusions, the new calculation does, at least, begin to reveal both the practical limitations and the strengths of a theory that can sometimes appear incomprehensibly general and hence difficult to believe.

    Dr. Gott's theory is based on the Copernican principle, which can be boiled down to his oft-repeated mantra, "you're not special." The astronomer Nicolaus Copernicus showed that the Earth is not at the center of the universe, that it occupies no special place, and the same idea works for more mundane things like addresses or names in a phone book.

    In an interview, Dr. Gott pointed out that neither Caves, Copernicus, Ferris, Glanz or Gott fell outside the fat 95 percent of the phone book -- before Anno or after Wilson, say. Dr. Gott could have predicted that that simple fact would probably be true "without having to go and do extensive research on the size of phone books all over the world."

    Then he extends the Copernican principle to time. Reasoning, after he visited the Soviet Union in 1977, that he was there not at a special time but somewhere in the middle 95 percent of its existence, he could figure out the Soviet Union's likely survival time from how old it was at that moment, 55 years. It's like guessing the total thickness of a phone book by counting the pages backward from a typical name, like Gott.

    Simple math shows that the current age divided by 39 (based on an area of probability of a bell curve) is the shortest likely survival time, and the age multiplied by 39 is the longest. His formula predicted that the Soviet Union would endure 17 months to 2,145 years. It fell 14 years later. Humanity itself, he predicts, is likely to be around from 200,000 to 8 million years longer, but not indefinitely.

    "The formula is extremely useful and maybe the best estimate you have in cases where your prior information is weak," Dr. Gott said. "The duration of intelligent species is one."

    But according to Dr. Caves's calculation, which is based on a rigorous formalism called Bayesian statistics, it is exactly the availability of this "prior," or supplemental, information that makes Dr. Gott's theory incorrect in a wide variety of circumstances. Only if one had no information about the politics, history, leadership, industrial base and institutions of the Soviet Union and all other countries would Dr. Gott's prediction be anything but a lucky guess, according to the Bayesian analysis.

    "Put succinctly, he rejects as irrelevant the process of rational, scientific inquiry, replacing it with a single, universal rule," Dr. Caves wrote. "That has to be wrong."

    In the paper, Dr. Caves highlighted his point by compiling a notarized list of 24 dogs owned by faculty and staff members and students in his department at the University of New Mexico. He selected the six dogs that were more than 10 years old, a procedure that required prior information, and pointed out that Dr. Gott's formula predicted that each should survive to twice its present age with a probability of 50 percent.

    Then Dr. Caves proposed $1,000 bets, at two-to-one odds in Dr. Gott's favor, that each of the six dogs would not survive that long. Dr. Gott replied that his formula would apply to the survival of all 24 dogs, but that the selected sample was unfairly chosen. "By the way," he added, "I just don't do bets."

    Dr. Caves replied that "it is inescapable that he doesn't believe his own rule in the case of the dogs."

    And what of the White Sox? They have not come close to the World Series since 1996, fulfilling the first part of Dr. Gott's prediction. But Dr. Caves says that "prior information," like the natural time spans of ballplayers' careers, cycles of ownership and the changing fortunes of other teams would allow him to predict a Series victory well before 5077.

    "Even if the White Sox are wretched for 10 years or 20 years," Dr. Caves said, "I think it's a real mistake to assume they'll be wretched for 3,000 years."