A distinction central to one strand of analytic philosophy in the early twentieth century, drawn most influentially by Bertrand Russell, holds that there are two ways of knowing what an object is — by acquaintance and by description. To know something by acquaintance is to be directly presented with it in experience, in a way that does not require any further inferential machinery; to know something by description is to know that there is a unique thing answering to a certain set of properties, without necessarily being acquainted with the thing itself. The distinction was meant to do considerable epistemological work, separating the kinds of foundation our knowledge can rest on.
Russell took as paradigmatic of acquaintance the relation between a thinker and a present sense-datum — a patch of colour as it appears to vision, a sound as it appears to hearing — and held that propositions about external objects, persons distant in space, and historical figures must be analysed in terms of descriptions ultimately grounded in acquaintance with such sense-data. The proposal had a clarity that recommended it. It also generated worries that have proved more durable than the proposal itself. If the only items of acquaintance are present sense-data, the external world and the past require a path back through descriptions whose terms have to be guaranteed by what one is now experiencing; the path turned out to be difficult to construct.
The difficulty has, over the twentieth century, produced several lines of response. Some philosophers have rejected the foundational role of sense-data, treating the distinction itself as resting on a problematic phenomenology; others have weakened the requirement, so that acquaintance need not be with sense-data and may extend to physical objects under suitable conditions; others have abandoned the project of grounding knowledge in any class of privileged direct relations and have shifted to inferentialist or holistic alternatives. Each line of response has its own difficulties. None has produced the consensus Russell's original proposal also failed to produce.
What survives, less an answer than a clarified question, is the recognition that any account of knowledge faces a choice between privileging a class of direct relations whose terms have to be defended and abandoning the project of grounding altogether in favour of accounts whose own commitments have to be defended in turn. The choice is not avoidable. Russell's particular implementation of one side has not survived intact; what has survived is the structural difficulty that his implementation, more clearly than what came before, brought into focus.
The primary purpose of the passage is to: