A good start would be increased quality of interfaces and abilities to find and connect similar sources; that's all I'm really asking for. There is a huge amount of research published within the Gale-Wilson-ElSevier-JSTOR-Ebsco-Taylor&Francis vaults, properties of immense academic and economic value, yet the tools they charge to access them aren't much more advanced than Yahoo circa '99. Don't even get me started on the transition from physical-copy journal publishing where you always 'own' the copy to a model of journals-as-a-service (the ultimate DRM; they host it) that cost well into the five figure range per year where you lose all access once you let the subscriptions expire.
Meanwhile, 'search' as we know it rules the world and is the driving force enabling the entire tech industry. It wouldn't be hard to throw a little effort in building out their platforms, but there's no competition to force further development in their products.
Not sure you realize, but these companies already are throwing more than a little effort into advancing their platforms.
Elsevier in particular is spending ridiculous amounts on internal research & acquisitions for its platform (called SciVerse, http://www.hub.sciverse.com/), following the latest advancements in machine learning and NLP for data mining, etc etc.
They also host competitions for start-ups and developers to produce new apps for their platform.
These businesses may be unethical and not to everyone's liking, but they are not stupid.
For most scientific fields, the last decade's research is much less important than the next decade's research. New research is much less locked in than stuff that was written before the internet became ubiquitous, so I'm not as concerned about that issue.
Meanwhile, 'search' as we know it rules the world and is the driving force enabling the entire tech industry. It wouldn't be hard to throw a little effort in building out their platforms, but there's no competition to force further development in their products.