In this article, we push the cognitive and information-science consequences of the new sociology of science and technology as far as they will go. Since there is no coherence stronger than a circumstantial network condensed or summarized into forms, structures and microtheories; since the computer, once it has shed its anthropomorphic and epistemological projections, is already capable of dealing with significant association networks; and since there are no computer programs on the market for researchers in the social sciences who deal with a great number of heterogeneous textual sources, we propose to construct a machine capable of revitalizing Hume and Condillac's unfashionable philosophical program of associationism, by giving it the material base it hitherto lacked.