Prepublication of scientific papers on preprint servers such as arXiv.org allows prompt scrutiny of the research by the scientific community (see, for example, go.nature.com/nwjmbt). Because the results are freely accessible, the arXiv approach is infinitely superior to the 'author pays' model of open-access journal publishing that is being pushed as a way to penetrate the paywall.
The number of mathematics and physics papers being posted to arXiv is rapidly increasing; subsequent publication in a journal serves mainly to validate the results.
Making authors pay to publish their research endangers the open and egalitarian nature of the scientific enterprise. Researchers in developing countries, unaffiliated researchers, graduate students and faculty members without large federal grants could all be priced out of publishing their work. The arXiv model offers a sensible and affordable alternative.