I'm a huge admirer of the hard push we're currently witnessing to make published research freely accessible to all who need it. The move towards universal Open Access (OA) is going more slowly than some would like, but there has been a huge shift in perception among academics, publishers and funders, most of it positive. Many of the issues in this debate have already been aired extensively, but I have a few brief observations (well, mini-rants) based on my own biases and experiences, both as a journal editor (both for commercially and society published journals), someone with a hand in running academic societies, and as an author. This is going to be a stream of consciousness, so apologies in advance ...
1. OA costs authors money. Please don't lambast people for not publishing their papers with Gold OA - they may not be able to afford it. Average fees for Gold OA are high (around $2.5K on average in my field) and at a time when budgets are tighter and tighter finding money to support publications gets trickier. RCUK recently provided the institutions that they fund with block grants to pay for work done via their grants to be published OA. However, these block grants are inadequate: one recently completed NERC grant of mine (which was funded prior to the new rules coming in) generated nine papers - the costs of making each of these Gold OA would have exceeded my institution's entire RCUK annual block grant. In this case, the funding body has imposed a knee-jerk response to a call for more OA, without providing the people that it funds adequate resources to do so. The situation for people in small institutions in developing counties is likely to be even harder.
2. OA costs publishers money. It could be argued that OA fees should be lower (and some places like PLoS and PeerJ manage this, which is admirable), but there is a fixed minimum cost of publishing a paper in any format due to the need to employ people to run journals. Even online journals require IT and editorial staff, as well as the staff needed to look after their pay and pensions, legal issues, maintaining their offices, etc. Costs start to balloon further when printing physical copies and distributing them. Although many large commercial publishers might have slack in their accounting lines to offer lower fees, thanks to economies of scale, many smaller publishers will not. Let's not forget many excellent journals are published by small academic societies, which serve niche communities well, and that are entirely financed by membership dues or small investment portfolios. Many small academic societies could not exist or promote new work without the revenues generated by publications, which are usually fed back to benefit the members of those societies directly. Give these guys a break, at least.
3. Green OA is great. Authors retain copyright over the final versions of their publications in their original form - i.e. the non-formatted final version they submit to the publisher. Posting of this after an embargo period is free, legal and gets the data out. It might not allow access to the beautiful proofed version of the article, but it does the job. Although access to the article isn't instant - it's still free after a relatively short lag. Please remember OA and instant access are not necessarily the same thing. Except in a few very rare instances, does it really matter if you have to wait six months for public access to a paper? (Especially given point 4, below). Admittedly the world might benefit from rapid medical or technological advances based on immediate OA, but let's be honest - it's 99.9% likely that the people with the insight and training to make those advances will have instant access to it through an institutional library or professional network anyhow.
4. Rediscover some scholarly skills. To some extent, I regard the need for instant gratification with some amusement. I did my PhD at time when email was just starting to be used widely (I had to go to a university computer centre to use it!) and PDFs did not exist. Amazingly, I still managed to access all of the material I needed, essentially for free, through libraries, interlibrary loans, requesting reprints, and asking colleagues if they had copies of particularly hard to get publications that I could borrow from them. Getting away from Google and asking a human being for help might actually have some side benefits for your research. All of these are forms of free information exchange that existed for centuries before exchanging PDFs. Ask the author for a PDF of that paywalled paper. You might actually learn something else by accident.
5. A historical perspective is interesting. Originally all scientific publication was funded by private patronage - authors had to pay to publish their work, either directly (often via the historical equivalent of crowd-sourcing) or via high membership dues to a handful of elite societies. Handing over publication to commercial publishers actually removed the need for authors to pay - scientists swapped the initial cash outlay for the fact that commercial publishers would profit from their work. So, the rise of academic publishers allowed people who formerly couldn't pay to get their research published. Free to the author at the point of submission. We've now come full circle, with authors again expected to pay upfront for Gold OA. We're effectively arguing for a nineteenth century financial model, though admittedly a better version of it in which papers are much easier to find for everybody...
In conclusion, it would be great if we could publish everything instantly for everyone and - ulimately - I would be totally in favour of this approach. Where possible, I currently use Gold OA for some of my work but, frankly, pressures to publish in certain venues (in terms of career advancement and visibility) and lack of ready funds often prevents this. However, until we work out where the money to pay for Gold OA will come from we will have an extended period with a mixed model of Green and Gold OA, as well as paywalls for some publishers that hold out for profits either because they want to or simply have to in order to survive. The main ways this might happen are reductions in OA fees associated with a total abandonment of print journals (saving on printing and posting), less reliance on editorial offices (e.g. less proofing - as happens with PLoS ONE - which has pros and cons), some unprecedented drops in expected profit margins for commercial publishers (good luck with the shareholders on that one), and/or governmental intervention to either cap profits on relevant publications or to provide the money to make up the shortfall between the two.