To take Ryan’s idea further, I’d also distinguish IR from Data Retrieval, using Keith van Rijsbergen’s classic introduction.

Additionally, I think one of the fascinating academic aspects of IR in the web era is that of large-scale evaluation.
Similar to talking about Wikipedia, which demonstrates the, say, “explicit” wisdom of the crowds, you could talk about web search evaluation (and also website analytics in general) as a task that spawned a whole set of sub-fields that glean info from huge amounts of query logs, such as result quality, query rewrites (synonyms, spellcheck…) and advertising bids.
It also boosted the concept of A/B testing as a viable and successful scientific methodology, and you probably remember Ben Schneiderman’s take on that