Subbiah Arunachalam forwarded a link to an interview of Professor P. Balaram regarding Open Archives. (Balaram is currently the director of the Indian Institute of Science, Bangalore so he is in a position to influence national policy in such matters).
I broadly agreed with his views and I posted a comment which I am reproducing here. Primarily, there are differences between different disciplines. The quotes below are from the interview.
If I publish a paper in Nature, for example, it can be made freely available in my institute's repository after six months. Some journals let archived papers be made available immediately.
... and again later ...
But many publishers have no problem with the open archives approach, even if some still require a delay.
In our discussions with Elsevier we were assured that the policy of that publisher is to allow authors to use institutional repositories at any time (but see the caveats below).
One issue that is yet to be resolved, however, is copyright. I argue that we should be permitted to put in the repository the full text article as it appears in a journal.
There is a difference between "full text" and "as it appears in the journal". If the journal has put in any effort at all into re-formatting your text/images in order to get it to conform with their journal standard, then the copyright for that re-formatting work (usually a combination of manual labour and software) belongs to the journal.
For example, Elsevier insists that you do not put the PDF (or other) image file of the exact form of the journal publication in your institute repository.
I personally find this quite acceptable since papers in the Mathematical Sciences are perfectly readable in the form in which the authors submit them to the journal. However, re-creating a distributable PDF or other online form of one's paper may be far more difficult in other disciplines --- I have no idea.
For this, countries such as India should have a law specifying that the copyright for articles published with publicly-funded research always vests with the authors and their institutions.
One need not wait for copyright law to change. The directors of each of the research institutes can insist that every paper sent for publication should be made available in the form of a research report with the institute library. By extension, the electronic form of this report would be in the institutional repository. One way to ensure compliance is to only allow reports that are available in the institute library to be used for evaluation (for promotions and the like).
The scientists would then refrain from giving up all copyright to their publishers. In my (limited) experience, publishers will not object to authors retaining the right to make copies as long as such copies are not independently sold.
Unfortunately, scientists as "the silent compliant mob" does not fit with my experience. In all matters which relate to science administration (and publications are at the centre of that), scientists are by-and-large "status quo"-ists. It will need a strong push from those at the top of the ladder to change the status quo. Professor P. Balaram is certainly one of those at "the top of the ladder" so I would like to see more action to match his words!
Update: Fixed the URL thanks to Abhishek Dasgupta. Also clarified the last sentence.