Computer communication

Most protocols for exchanging data between computers have in the end traditionally had rather simple structures—with different pieces of information usually being placed at fixed positions, or at least being arranged in predefined sequences—or sometimes being given in name-value pairs. A more general approach, however, is to use tree-structured symbolic expressions of the kind that form the basis for Mathematica—and now in essence appear in XML. In the most general case one can imagine directly exchanging a representation of a program, that is run on the computer that receives it, and induces whatever effect one wants. A simple example from 1984 is PostScript, which can specify a picture by giving a program for constructing it; a more sophisticated example from the late 1990s is client-side Java. (Advanced forms of data compression can also be thought of as working by sending simple programs.) But a practical problem in exchanging arbitrary programs is the difficulty of guarding against malicious elements like viruses. And although at some level most communications between present-day computers are very regular and structured, this is often obscured by compression or encryption.

When a program is sent between computers it is usually encoded in a syntactically very straightforward way. But computer languages intended for direct use by humans almost always have more elaborate syntax that is a simple idealization of ordinary human language (see page 1103). There are in practice many similarities between different computer languages. Yet except in very low-level languages few of these are necessary consequences of features or limitations of actual computers. And instead most of them must be viewed just as the results of shared history—and the need to fit in with human cognitive abilities.