On Open Access to the literature
Obviously, the devil is in the details, and I don't know that the specific legislation is the best possible approach. That said, I'm hugely in favor of open access and a requirement that government funded research be released under some sort of Creative Commons license, to allow that research to be repackaged and reused for non-profit purposes with attribution.
I also know that journal editors invest colossal amounts of time and effort on very challenging work and, except for those at journals published by professional publishers, get nothing but a lot of whining and perhaps a small stipend. Printing journals costs money and time, and selling subscriptions serves a vital role in the system. I'm not too familiar with the details of how arxiv.org has worked for the physics community and how it fits into the traditional journal system, so I don't know if it offers a different path.
I will say that the basic outline of the Federal Research Public Access Act of 2006 looks reasonable. Research would have to be available as open access within 6 months of publication. Specialists in the field (or their institutions) would still subscribe to the journals they needed immediate access to, but the public could readily access the historical record of the literature. Six months is on the short end, 2 years would be at the long end, but some timeline between those would balance the public's needs with the publishers' and societies' real needs.
PP doesn't seem to think that the 6 month window is adequate to recoup costs and he may be right. Paper subscriptions for libraries who want them will probably go up in price, especially since online publishing makes expensive to print graphics more popular. An online publication can include color, movies, sound recordings and a volume of data that would be prohibitive for a print publication. It's hard to include that material in a printed format, so the shelves full of paper journals will thin out, but lower cost online publishing will allow smaller societies to produce and distribute niche journals. Whether they're funded by subscriptions, or if ad space and author fees have to be added to make up the cost, science and society both benefit from the outcome.