I have an on-going interest (as well as a stake) in how academia will advance in the digital world, where new capabilities afforded by technology force us to question the way we currently produce and distribute research works. Particular problems that have already been highlighted involve journals. One of these is the idea of the "impact factor", an idea particularly maligned by active researchers; another is the availability of articles to the public. These problems are not short of proposed solutions. (David Colquhoun is a prodigious and no-nonsense writer in this respect who I recommend; over here, I have already written about article-level metrics as a remedy against the first problem).
A stopgap solution to the second problem of the availability of research papers is to be found at the newly-launched ScienceLeaks (leaking seems to the all the rage right now). Put simply, the problem is that access is unfair, because papers are usually only available in journals that block access to them based on ability to pay. Realistically, this means you have to be affiliated with some kind of research institute that will pay the access fees (even an individual paper can cost upwards of £10 - £20, and when writing an article of your own 10 - 20 references minimum is typical). And this doesn't always solve the problem; I studied at the University of Lincoln, a fine institute, but it could not at the time afford access to every publisher we needed in the computer science department. This is to say nothing of the immorality when you consider that much research is paid for by public money, only to be then made practically unavailable to the public. ScienceLeaks's mission is to provide a place where researchers can request papers they cannot access and people can post copies for them.
Time will tell if the project manages to fulfil that need.