We examine research issues that arise when most information items in an enterprise can be linked to each other via short paths, implicit or explicit. In such high-recall settings, the treatment of metadata management, indexing and ranking needs new attention. Additional issues arise as to the best way to handle updates to the connections, whether on or off the transaction path. Even traditional techniques, such as classification and clustering of documents, which stand to benefit from the extra information provided by the so-called network of meaning, need to be reexamined for how best to exploit the extra information. The talk ends with an examination of some promising avenues for using high recall as a driver for the next wave of business process automation