Knowledge is a relationship between two terms, at least one of which contains information about the other. We have not yet been able to find an answer to the two great questions we have asked—where does the information get stored, and how does it get from one term to the other? In a search for an answer, we have come along several different roads only to find that, contrary to proverbial expectations, none has led to Rome. When we tried to come from within and started with consciousness, we found that consciousness does not really store information and that such information as it does store cannot be represented. What is more, we found that what consciousness does succeed in representing is not information about the world. We fared differently but no better when we tried to come from the front. The idea that information is transferred when or because there is a causal link between the two terms turned out to be a pipe-dream because such causally induced information cannot stand on its own two feet but is in need of linguistic expression. Thus the information gets bushed, and by the time it finally comes out, it is no longer causally related to the first term of the relationship.
Next, we tried to find out what we get when we take it that the information comes from behind. The possibility that the information is actually generated by the rules of sociability, including the rules of language, seemed more promising. We were able to show that the principles of the invariance of experience and of the need for consistency of explanations can be derived from the conditions of sociability. It was only in the very last instance, when we were looking for an explanation of the fact that all actual sociabilities are related to one another, that the recourse to sociability failed. The view that knowledge is a social convention must enshrine relativism. It is quite viable, until it fails to explain how one relative system is related to another relative system. Such an explanation is indispensable because we know that all systems are part and parcel of one and the same universe and must therefore be compatible with one another, or—to put it negatively—cannot be left to agree to differ from each other. This kind of explanation of the ultimate compatibility and consistency of all conceivable systems, it seemed, is only