First of all you have to recognize that David knows more than all of us. After that everything is easy. In this instance I believe he is correct, and his statement invites questioning and discussion.
The web is organized on a many-to-many (and therefore messy) topology. Libraries are organized on a one (you) to many (the books) and much neater and easier sometimes to deal with, but also limited in what they can offer since they rely on some one or thing to enter all the data (i.e. books) and then index what has been entered (and in a timely way.) Also our wonderful and senitmenal libraries may not index all of the possible references to a given source, and so our searches there may not find what we are looking for.
The web on the other hand has robots that do the indexing for us on a microsecond basis every day all day. That's the good news. The bad news is we do not yet have good enough filters to fine tune our searches and so we get links and references to many things we don't care about - we become a human complier if you will allow such a metaphor. This will change as AI improves and our local computer plus the giant indexers in the sky learn our habits and tastes.
The good news on the poor filters of today is that we also get serendipity and see things we wouldn't normally look for which may lead us to new lines of thought, investigation, or just the joy of discovery.
Jon Peddie