This is a great question that has some complex answers. All source material is not equal. All material is not in a library that allows free access, so you migh be talking about paying for access as an inhibitor. In some cases, you're talking about material that cannot be digitized for fear of damaging the source material i.e. flashes or scanners.
Another issue becomes should everything be digitized just for the sake of access? In many ways, what is accessed is decided largely due to relevance. If someone needs to access source material that requires a trip, they usually apply for grants or find ways to fund accessing the material. The lack of digitization never really hinders research just due to that fact alone, but digitizing materials definitely makes things more accessible without question so it is kind of an odd duality. In the same way it does not make sense to record every e-mail ever written just because they were written, academic materials do not necessarily need to be digitized to "find the hidden cure for cancer". The academic process just does not rely so heavily on digitization to "find answers" so to speak.
As well, the materials that are not accessible online are usually very specialized source materials that are very rare or valuable. That is not going to be something that even "undergrads" are necessarily dealing with, let alone someone who is transitioning from poverty to online education. They are just very different demographics and MOOCs have made the idea of not having "access" one more of "Internet access" for disenfranchised, and not "information access".

Fair points Larry. I agree with some parts but I was most concerned with researchers in the field of research and diagnostics. Those very specialized source materials that are rare and valuable is exactly the point I am referring to. Indeed, undergrads and others accessing educational content are not going to touch on such materials. But, those conducting researching and looking for hidden relationship discoveries for future development will be interested in finding solutions from our digital world.
Take knowledge automation of curation for example, these platforms are designed to access information that is already available in the digital world and only in the digital world. These correlations then become substantial for most recent research relevant to their study and taking it further from there. The scientific publications not available in the digital space get left untouched/unread/unheard of. These are not any source materials we are talking about...they are scientific published papers and articles with study information to drive innovation, diagnostics, and cure forward. We are now moving into a space of automated curation that strengthens our ability to search and find meaninful extraction in a faction of a minute. Its also interesting to see what we may be leaving behind.