The notion of data and information being different in kind is based on the Foundationalist thesis that there exist raw, brute, facts which constitute the data that form the basis for information, and, ultimately, of knowledge. Foun dationalism fails, however, because if these data are really different in kind from information and knowledge, then no comparisons are possible between the former and the latter, and the notion of data (in the sense of raw, brute, fact) becomes useless. If, on the other hand, comparisons can be made between data and information or knowledge, the data would have to be of the same kind as the latter. Given that an Information Processing System (IPS) cannot process data except in terms of whatever representational language is inherent to it, data could not even be apprehended by an IPS without becoming representational in nature, and thus losing their status of being raw, brute, facts. The representational language of the IPS provides the categories in terms of which the IPS 'views' reality, and thus this language will define what constitutes reality for the IPS in question. Consequently, this language will define what constitutes signals, signs, and information for the IPS, as well. Any definition of information must therefore be relative to a given IPS, and, in the case of a human IPS, what is regarded in common sense terms as linguistically encoded information cannot be independently characterized in purely physical terms.
A Dilemma for the Scientific Realist
