A proper Torah Code experiment is an experiment by whose results one can determine whether or not a Torah code table is likely to occur by chance in a population of monkey texts. Formally, a proper experiment is one which tests the hypothesis of No-Torah Code effect against an alternative of Torah code Effect. If the experiment does not reject the hypothesis of No-Torah code effect, then there is nothing statistically interesting about the involved Torah code tables.
In order for results of the experiment to enable this determination, the experiment must have an experimental protocol that is in place before the experiment begins. The experimental protocol gives all the details of the experiment that would enable another experimenter to replicate the experiment and expect the same results within statistical sampling error.
The probability estimated by the experiment must have the following meaning: It is the probability that a text from the monkey text control population would have as good compactness as the table that was produced by the Torah text. The compactness measure must have the meaning of not highly compact for tables produced under the NULL hypothesis of No-Torah Code effect, that is, from texts in the monkey text population. It must have the meaning of highly compact for tables produced under the alternative hypothesis of Torah Code effect, meaning for tables produced from the Torah text.
Those so-called Torah code tables that are shown, as is done in many websites, without any associated statement of experimental protocol and without giving the probability that a result as good as what was observed under the Null hypothesis of No-Torah Code effect are statistically and scientifically worthless.
Sometimes there are tables that are shown with a probability computed analytically, meaning by formula. There is no known experiment which would produce the values that those formulas produce. It is easy to find examples where the probability produced by such formulas are two or more orders of magnitude too small with respect to what one would obtain by some reasonable experimental protocol.
The experimental protocol specifies in an a priori way how the experiment is done in sufficient detail that the experiment with its results can be replicated by other people. A priori means that the protocol is specified before doing any experiments and before looking at the text in any way for ELSs of the key words.
In the case of Torah code experiments involving one set of key words, the experimental protocol must have: the given set of key words, an ELS skip specification, a cylinder resonance specification, the compactness measure to be used, the control population from which the p-level will be estimated. Torah code experiments involving more than one of set of key words have more complicated protocols.
Key Word Set
The set of key words describing an event or partially describing an event, or making up a sentence about the event, must be specified before looking for any ELSs of these key words in the Torah text and before doing any kind of interactive snooping for ELSs of these key words. The event associated with the key words must have occurred prior to the specification of the key word set.
Torah code critics claim that all Torah code experiments producing small p-values can be explained by hypothesizing that the researcher doing the experiments was not honest and in fact the key words used were not a priori. In that case, the probabilities produced by the Monte Carlo experiments are meaningless.
The ELS skip specification tells for each key word the minimum and maximum skip permitted for the key word's ELSs. The search process then finds all ELSs from the minimum skip specified for the ELSs of the key word to the maximum skip specified for the ELSs of the key word.
Cylinder Resonance Specification
The cylinder resonance specification specifies the relationship between the cylinder size and the skip of the ELS which qualifies for resonance. For example the maximum row skip and maximum column skip an ELS can have on a cylinder is one way of specifying a resonance criteria.
The compactness measure specification indicates the choice of compactness measure, such as area of best table containing one ELS of each of the given key words, or the furthest distance between the pairs of the letters of the ELSs on the the best cylinder size. Compactness measures that attempt to evaluate redundant encoding involve more than just the best table.
Control Population Specification
The control population determines the meaning of the probability that a table of compactness measure better than a given size would occur by chance. The most meaningful group of control populations is a control population of texts. In this case the probability that as good a table would happen by chance means the fraction of texts in the control population whose best table is more compact than the compactness of the best table of the Torah text.
Text populations that are monkey text populations can be created by various random shuffling techniques. These include letter shuffling, word shuffling, verse shuffling, and chapter shuffling. Letter and word shuffling can be done within verse, within chapter, within book, or globally. Verse shuffling can be done within chapter, within book, or globally. Chapter shuffling can be done within book or globally. All these shuffling techniques guarantee that the letter frequency of the text is preserved in all the monkey texts. However, the ELS statistics will change. To keep the ELS statistics the same we favor the ELS random placement monkey text population discussed.
In the ELS random placement monkey text population, the ELSs found in the Torah text are randomly placed. That is their skips are kept and their beginning positions are chosen at random. In this population, the number of ELSs and their skips are kept the same and the only thing that changes from text to text is the placement of the ELSs in the texts.