Scholarly publishing is fundamentally broken at essentially all levels starting with basic functionality and ranging to access, incentives, attribution, archiving, share/re-use and cost/benefit. What could be the feasible alternatives which would move scholarly publishing closer to a modern, IT-based system? A system which assists scientists in sorting, filtering and discovering relevant scientific findings? Which provides open access to tax-payer-funded research at a reasonable price? Which provides an incentive structure with an attribution system that benefits science and scientists rather than publishers and bureaucrats? I will argue that a natural candidate for developing such a system is the one institution on this planet which has centuries of experience in archiving and making accessible the literature and data of scientists: university libraries.
In addition to centuries of experience, many libraries in our digital age lack a sense of purpose or direction because of it. Creating a world-wide, peer-reviewed, open access, federated digital library of science is exactly the kind of task any modern university library should dream of taking part in. This digital utopia is exactly what scientists today are in desperate need of and libraries would be able to deliver.