Bias and the Bible Code
The question of bias often arises when discussing the Bible code. This seems inevitable. The Bible code proponents claim to have scientific evidence that information about recent events has been encoded into the book of Genesis (and possibly other parts of the Hebrew Bible, depending on who you listen to). In other words, scientific proof of a miracle.
The main proponents of the Bible code are either Orthodox Jews or Christians. So there's an undercurrent of suspicion among many skeptics that these believers are biased. They want to find a code, because that would validate their religious beliefs.
On the other hand, a good many of the believers seem to think that it's the skeptics who are biased. If the codes were real, then they'd have to accept that God wrote the Bible, and they can't do that. (Note, however, that plenty of the skeptics, are Orthodox Jews and Christians.)
So bias is important. Note that there are two kinds of bias -- intellectual and emotional. I'm sure you know the difference. If a tiger is fighting a house cat, your brain tells you to expect the tiger to win, even if your heart goes out to poor Fluffy.
It's relevant for me to discuss my own biases here. First off, I'm a Christian. I believe in God. I deeply respect the Bible. I think that God has the ability to encode hidden messages in the Bible if He so chooses. I'm also a physicist. If someone makes scientific claims, then I want to see scientific evidence. I started my investigation of the Bible code with an intellectual bias. Let me state it here:
The Bible code sounds too good to be true. It makes an extraordinary claim, one undreamed of by most people until recently. Extraordinary claims require extraordinary proof. When I began my investigation, my intellectual bias was that probably these claims would fizzle.
OK, and what about my emotional bias? That, I believe, was essentially neutral. If the codes turned out to be true, well, that'd be very interesting -- they'd make a powerful apologetic tool. That would be good. But if the codes turned out to be true, then there'd be a serious problem of people misusing the codes for their own devious ends -- trying to break the stock market, predict the Second Coming, and lots of other goofy things. That would be bad. So, on an emotional level, I didn't have a strong desire either way -- to prove or disprove the codes.
Despite my intellectual bias, I felt the subject worth investigating further -- just in case. I explained my reasoning at the end of Chapter 4 of my book, Who Wrote the Bible Code? If there really were codes, and if I could prove it, I'd have a story to knock yer socks off, dude. And if there really weren't any codes, and if I could prove that, then we could all move on to something more important.
In my book, I devoted Chapter 5, "The Bias Protection Plan" to the question of bias and how to eliminate it. Here, I'm referring to intellectual bias, of course. I don't know how to eliminate emotional bias. When somebody wants to believe something, they'll find a way.
You eliminate intellectual bias by using scientific method. Let's review that.
In the scientific method, you go through the following steps:
- You make observations.
- You develop hypotheses to explain these observations.
- You suggest experimental ways to test these hypotheses.
- You predict the results of the experiments.
- You run the experiments.
- You compare the results of your experiments to your predictions.
- You publish all your results.
Of course, you may well loop back from step 6) to step 1) a few times before you get to the publication stage. However, in step 7), you're supposed to publish all your results, not just those that happen to agree with what you expected. This is the scientific protection against bias.
Unfortunately, it's very easy to cheat when searching for Bible codes.The reason is that we have a limited supply of Bible (they aren't writing any more of it), and it's widely available (so people can peek before they predict). Therefore, it's easy to do step (5) first -- run the experiments before making predictions -- and many Bible coders do this without apology. They'll go hunting around, "experimenting" by looking for this or that word, without really saying what they're looking for. Then when they find something, bingo! -- they publish only that result, without telling how many similar "codes" weren't there.
Now most everyone agrees that you're going to find a lot of words by random chance. So it's all too easy to find "Yeshua" and "false" and "Messiah" and conclude, "Yeshua is a false Messiah!" Or, if you don't like that conclusion, it's all too tempting to leave out the "false" and conclude, "Yeshua is the Messiah!" Neither of these conclusions is at all scientific. But this is the sort of thing that many Bible coders do all the time. Michael Drosnin, Yacov Rambsel, and Grant Jeffrey come to mind here.
However, there have been some real attempts to prove the Bible code scientifically. If you've read Jeffrey Satinover's book, you know all about the "Great Rabbis Experiment" published by Doron Witztum, Eliyahu Rips, and Yoav Rosenberg in the journal Statistical Science in August of 1994. In that article, Witztum, Rips, and Rosenberg (everybody calls them "WRR") argue that they followed precisely the usual steps of the scientific method.
The Great Rabbis Experiment looks good on paper, but the devil is always in the details. It's now widely known that WRR's procedure had a number of holes that made it possible to bias the results. I stress the word possible. Nobody can prove that WRR actually did bias their experiment. But it is very clear that they could have, and a great deal of circumstantial evidence indicates that something went awry. In any event, their experiment does not follow scientific method and it can not be used as scientific evidence that the Bible has real codes.
How do you use the scientific method correctly? Let's walk through the steps of the method and discuss what WRR did right or wrong at each step in the Great Rabbis Experiment:
- Make observations. WRR correctly observed that you can "often" find clusters of "related" words at various skips in the Hebrew text of the Bible (or Moby Dick, or War and Peace, etc.)
- Develop hypotheses. WRR made the plausible hypothesis that the author of Genesis selected clusters of words and cleverly wove them into the text at various skips, making sure to encode more data than could reasonably occur by random chance.
- Suggest experimental tests. WRR say that they chose pairs of related words in a purely objective way (to avoid the suspicion that they jumped ahead to step 5 first). They also say that they defined in advance how to measure the "closeness" of word pairs and how to estimate statistical significance. This is where they ran into problems, as I'll discuss below.
- Predict results of the experiments. WRR predicted that the names of rabbis would be found "near" their birth dates and death dates.
- Run the experiments. WRR measured the closeness of the rabbis' names to their dates.
- Compare the experiments to predictions. WRR, showed that the rabbis' names were in fact much closer to their birth dates and death dates than one would expect by random chance.
- Publish results. WRR published their results.
The main problem with the Great Rabbis Experiment came in step (3). WRR have been unable to prove that they defined their experiments before they ran them. They didn't publically say in advance how they'd spell the names of the rabbis, for example. And their definition of "near" wasn't very good -- it is possible to tune the results by adjusting just a few spellings. After the experiment, they did publish a list of "rules" for these spellings. But the Hebrew experts tell us that these rules are very ad hoc. The rules could have been adjusted after the experiment to produce the "right" names -- the names that worked.
Brendan McKay and his collaborators have shown that by adjusting a few spellings for the rabbis' names, you can get "amazing results" in the Hebrew translation of War and Peace, and terrible results in Genesis. That's the essential flaw in the Great Rabbis Experiment.
That's the reason I began my investigation. People seemed to be locked in a stalemate, arguing about what was right or wrong with the Great Rabbis Experiment, which was clearly flawed, clearly open to charges of bias. The Great Rabbis Experiment failed to use scientific method rigorously enough. That's why people are arguing about it.
I asked myself how we could change the question so as to eliminate all bias. The solution turned out to be fairly simple. The idea is to look for meaningful information, not for words.
As I discussed at great length in my book, when you encode information using human language, you always introduce redundancy. That's the amazing thing about language -- it carries a bit of redundancy, that is, a bit of order -- and you can measure this. You can measure it objectively.
I reasoned that if God encoded meaningful information using human language, then He couldn't help but increase the orderliness of the text. Equivalently, He would have to reduce the so-called entropy. (Entropy is a measure of disorder, in language just as in physics. It can be computed using precisely defined methods. Even if I'm biased as heck, I can't change the definition of entropy. It's been around since before I was born.)
So I measured the entropy of the "skip-text" that you get by skipping letters sequentially through the Bible. I predicted that if God had encoded a number of words at many different skips, then that should reduce the entropy of the skip-texts. In fact, I made this precise. I encoded some meaningful information into a random text, and showed that you can detect the presence of a small number of meaningful words in a large random text by measuring the entropy. This is true, even if you don't know which words were encoded, even if you don't know what language the words were written in, and even if the number of occurrences of each word appears to be what you'd expect by random chance.
However, when I measured the actual entropy in the skip-texts of the Bible, I found no measurable effect.
The conclusion is that very little (if any) meaningful information is encoded in the Bible at constant skips. The beauty of this method is that there is no room for bias. Entropy is well-defined. Skip-texts are well-defined. I didn't look for words, I looked for information, using commonly accepted tools.
Now please note one thing. I don't claim to have shown that absolutely zero information is encoded in the Bible. Nobody can prove that. Nobody can prove that God didn't encode even a single word.
But I have shown that the amount of information actually encoded is relatively tiny. There are a vast number of words encoded as ELSs within the Hebrew Bible. If even 1 in 2000 of these were intentionally encoded, then my tests ought to be showing a statistically significant signal. My tests don't see it.
This, of course, is exactly; the opposite of what we've been told by enthusiasts of the Bible code for some years. There is supposed to be a vast amount of information -- about every man, woman, child, animal, plant, etc., in the world. At least, this was the claim made in the Discovery Seminar book and in Jeffrey Satinover's book and Michael Drosnin's book. This provided the reason to go look for the codes. With so many codes, people reasoned, you could find real codes wherever you looked.
Nobody seems to believe this anymore. Real codes, if they exist, are rare. And that's the unbiased truth.
