The Gambler's Fallacy Versus Russian Roulette
The Gambler's Fallacy Versus Russian Roulette
Ockham's Razor: It is vain to do with more what can be done with less.
—William of Ockham
Design or Many Worlds? Both have power to explain the type of phenomena that interests us. How to break the tie? We start with consideration of the power of prior knowledge in judging explanatory sufficiency.
Consider the following example. Suppose you are sitting on a jury during a murder trial. The evidence appears overwhelming. The DA presents DNA evidence that apparently links the defendant with the crime scene. An expert witness testifies that the probability that the DNA match could be a false positive is 1 in 1 followed by eighteen zeroes (1018). This evidence alone should be enough to convict.
Suppose that the defense comes forward with a novel theory. This is not the only world, the defense attorney says. There are, in fact, 1018 parallel worlds out there. Probability shows, therefore, that there is at least one person in the universe who, at this moment, is on trial for this murder but is definitely innocent. This is reasonable doubt. How can you convict?11
Before you decide, we note that the trial is not yet over. The DA presents additional evidence. The defendant is the only one with a motive, he says. The defendant lacks an alibi, he says. An eyewitness indicates the defendant was close by near the time of the murder.
On further consultation with my expert scientists, says the defense attorney, it turns out there are more than 1018 parallel worlds out there. In fact, there are enough worlds to cover the DA's circumstantial case. If the DA comes forward with evidence tomorrow that shows the defendant is an additional 10100 more likely to commit the crime than anyone else, we can certainly “discover” 10100 new worlds.
As one can see from the apocryphal story, there is room for the Many Worlds theory to be vastly abused. The Many Worlds advocate is engaged in a problem called the Gambler's Fallacy. Suppose I flip a coin repeatedly. Heads, you pay me a dollar. Tails, I pay you a dollar. I win 10,000 times in a row. Should you be upset? Not to worry. I assure you that there are sufficient Many Worlds out there where coin flipping is going on, and those copies of yourself must surely be winning so as to even the odds. Feel better? I'll give you double or nothing on the next flip.
But let us make a contrast to bring home an important point. Suppose Earth will be destroyed by an unavoidable natural disaster. We have time to build one spaceship big enough to take one family. A raffle is arranged where there are one billion families with an average size of six. One billion tickets are issued. Should we be surprised that someone wins the raffle? No; it was guaranteed. Thus should the winner be surprised, similar to the example of the firing squad victim? No.
What is the difference between the raffle example, where Many Worlds works, and the Gambler's Fallacy, where it doesn't? In the former example I have explicit prior knowledge. The raffle tickets, which are analogous to Many Worlds, are known to exist. It is known that someone will definitely win. So if I can directly observe Many Worlds, or I can show inferentially that some mechanism will definitely extrapolate a life-amenable universe with a life site, then I have something. If I am simply inventing Many Worlds, then I am engaged in the fallacy.