The following is an excerpt of a letter authored by the Lubaviture Rebbe:
[…] Basically the problem has its roots in a misconception of the scientific method or, simply, of what science is. We must distinguish between empirical or experimental science dealing with, and confined to, describing and classifying observable phenomena, and speculative science, dealing with unknown phenomena, sometimes phenomena that cannot be duplicated in a laboratory. Scientific speculation is actually a terminological incongruity; for science, strictly speaking, means knowledge, while no speculation can be called knowledge in the strict sense of the word.
At best, science can only speak in terms of theories inferred from certain known facts and applied in the realm of the unknown.
Here science has two general methods of inference:
The method of interpolation (as distinguished from extrapolation), whereby, knowing the reaction under two extremes, we attempt to infer what the reaction might be at any point between the two.
The method of extrapolation, whereby inferences are made beyond a known range, on the basis of certain variables within the known range.
For example, suppose we know the variables of a certain element within a temperature range of 0-100, and on the basis of this we estimate what the reaction might be at 101, 200, or 2000.
Of the two methods, the second (extrapolation) is clearly the more uncertain.
Moreover, the uncertainty increases with the distance away from the known range and with the decrease of this range. Thus, if the known range is between 0 and 100, our inference at 101 has a greater probability than at 1001.
Let us note at once, that all speculation regarding the origin of the universe comes within the second and weaker method, that of extrapolation. The weakness becomes more apparent if we bear in mind that a generalization inferred from a known consequent to an unknown antecedent is more speculative than an inference from an antecedent to a consequent.
That an inference from consequent to antecedent is more speculative than an inference from antecedent to consequent can be demonstrated very simply:
Four divided by two equals two. Here the antecedent is represented by the divided and the divisor, and the consequent – by the quotient. Knowing the antecedent in this case, gives us one possible result — the quotient.
However, if we know only the end result, namely, the number two, and we ask ourselves, how can we arrive at the number two, The answer permits several possibilities, arrived at by means of different methods:
(a) 1 plus 1 equals 2;
(b) 4 – 2 equals 2;
(c) 1 * 2 equals 2;
(d) 4 divided by 2 equals 2.
Note that if other numbers are two come into play, the number of possibilities giving us the same result is infinite (since 5 – 3 also equals 2; 6 divided by 3 equals 2 etc, ad infinitum). Add to this another difficulty, which is prevalent in all methods of induction.