One difference between my reviews and the ones outlined by the article is in how to treat fatally flawed manuscripts and majorly flawed ones. The article suggests identifying these errors and being done with the review. In my experience, there is a good chance that the manuscript will still get "accepted subject to major revisions," so this may be the only chance to really convey to the authors what they need to improve. (Otherwise, there's the risk of multiple rounds of reviews, which no one wants.) To that end, I always provide a full review, even if I am recommending rejection of the paper. I always hope that the editor will appreciate the effort, and also that the authors will get more constructive criticism to help improve a new version of the manuscript, or at least have a comprehensive set of comments to address in a revised version that will come back for another review.
On a side note, this is also helpful, at least in principle, in the "dialog" between the climate science community and the climate change denying community. There is a lot of mistrust of the peer-review process among "skeptics" (note they are not skeptical, really). The article in Eos is a simple explanation of how peer-reviews are written, and I think almost everyone will agree that the description is roughly what reviews look like. While sometimes reviews are more or less helpful, the thought behind them is basically the one described by the article. They are meant as constructive criticism, pointing out strengths and weaknesses of work, judging the contribution to the field, and recommending whether publication is reasonable or not. This is how I write reviews, and it is similar to reviews I receive. In climate science, reviews are usually anonymous, though there's always the option of signing a review so that the authors know who you are. The decision about how the submitted manuscript proceeds comes from the editor, not directly from the reviewers, so even if a paper receives an unfair review, the editor can take corrective action. When that doesn't happen, the authors can appeal to the editor, pointing out where the review has gone off track. The usual course of action would be for the editor to find another reviewer, who can then corroborate the criticism or back up the author. There really isn't anything secretive or mysterious about the process. There are obvious measures to avoid conflicts of interest, and procedures for interaction among the participants. So while there are interesting arguments about how to improve peer review (blinded reviews, e.g.), I think the current system is quite transparent and attempts to be fair to all involved. I hope that critics of climate science will read this article (and David Schultz's book, LINK) before they attack the peer review process in general.