Does Moby Dick predict murders?

Why a novel hides any name

In 1997 the journalist Michael Drosnin published The Bible Code. It built on a 1994 paper by Doron Witztum, Eliyahu Rips and Yoav Rosenberg, who read the Hebrew text of Genesis at equal skips and said the names of famous rabbis sat closer to their dates than chance allowed. Drosnin went further: he said the text held Yitzhak Rabin's murder, among many other events. In Newsweek he dared his critics to find a prime minister's assassination in Moby Dick.

Brendan McKay, a mathematician at the Australian National University, took the dare. In Moby Dick he found Indira Gandhi, Leon Trotsky, Martin Luther King, John F. Kennedy, Abraham Lincoln, Rabin himself and Princess Diana, each with words about their deaths nearby. "Once you learn Drosnin's rules (none)", he wrote, "you can find things like this anywhere." In 1999 McKay, Dror Bar-Natan, Maya Bar-Hillel and Gil Kalai showed in the same journal that the Genesis result came from the many free choices made along the way: which spellings, which names, which dates.

The count is the whole trick. A five-letter name can start at any of about 935,000 letters, at any skip up to 5,000, read either way: about 9 billion places to land. Then the searcher picks the grid that looks best and reads it. Our page looked through up to 400 grids for every name and kept the darkest one. Moby Dick did not help. Any long text does the same. In his book Evidence-Based Technical Analysis, the trader and teacher David Aronson put it plainly: the codes "predict with 20/20 hindsight", and they are "a by-product of the search method and not of the text being searched."

The same game in a trading room

Chart books teach that a pullback likes to stop at 38.2%, 50% or 61.8% of the move before it, the Fibonacci lines, and they print pictures where it did. Those pictures are picked like the grids above: from thousands of pullbacks, the one that landed on a line. Our coin-flip prices have no memory and no lines, yet every chart still shows a near-perfect turn, because three lines with a little room each catch plenty of turns by chance. Lines a few points higher or lower catch just as many.

Aronson wrote the same about Elliott waves, the wider school that the Fibonacci lines belong to: "that which purports to explain everything explains nothing." Its loose rules let the analyst fit wave inside wave to any chart after the fact, the way old astronomers added circle upon circle to keep the Earth at the centre. A rule you can test says where the turn will be before it happens, and then counts the misses.

Sources

1. Michael Drosnin, The Bible Code (Simon & Schuster, 1997). His challenge appeared in Newsweek, 9 June 1997.
2. Doron Witztum, Eliyahu Rips and Yoav Rosenberg, "Equidistant Letter Sequences in the Book of Genesis", Statistical Science 9, no. 3 (1994), 429–438.
3. Brendan McKay, Dror Bar-Natan, Maya Bar-Hillel and Gil Kalai, "Solving the Bible Code Puzzle", Statistical Science 14, no. 2 (1999), 150–173.
4. Brendan McKay, "Assassinations foretold in Moby Dick!", Australian National University web page (1997).
5. David Aronson, Evidence-Based Technical Analysis (Wiley, 2006), chapters 2 and 6.
6. The text is Moby Dick from Project Gutenberg, letters only, searched in your browser at every skip from 2 to 5,000 both ways. The chance count multiplies how common each letter is by the number of places. The prices are fair random walks with 1% daily moves; a turn is a 5% swing, and a line catches a turn within 2 points.