Like many scientists considering alternatives to research careers, I looked into a scientific editing position because I enjoy writing and science. After graduate school and a postdoctoral fellowship, I worked as a full-time editor at Neuron  , which is published by Cell Press. I recognized very quickly that the senior editor position was primarily about understanding what constitutes outstanding, cutting-edge research and making sure that that was the research published in the journal.
Academic journals are believed to be in the business of providing service to their segment of the public. However, in the end, they do have to make money. Most journals are supported by a combination of subscription fees and advertising revenue (and sometimes page charges). Advertisers buy space based on the number of readers (read subscribers), so ultimately journals make money by selling subscriptions. At Cell Press, as at many high-impact journals, the philosophy was to make sure the journals publish the most important papers that everyone wants to read, that everyone is talking about. Thus the journal becomes a "must read" for scientists. At the same time, the journals keep their subscription rates low enough that a high proportion of scientists will simply buy their own subscription, rather than rely on a shared copy.
In my experience, editing has much more to do with understanding science and scientists, than with constructing beautiful prose. The primary role of an editor is to act as a liaison between the author and the audience. At a scientific journal, the scientists are primarily content editors. Their job is to decide what papers are appropriate to publish in the journal.
I had to learn quickly on the job. It was helpful that I had spent too much of my time as a graduate student going to seminars and reading papers in areas outside my own research. Reading recent issues of several journals also gave me a broad introduction very quickly. One of the most important things I did to broaden my understanding of a variety of research topics was travel to several meetings each year. I focused on small, somewhat exclusive meetings on specific topics, where I would meet and hear talks from a cross-section of researchers in a specific field. This gave me an appreciation for what the field thought was interesting (i.e., who was invited to speak on which topics), who is competing to answer the "hot" questions in a given area, and who asked incisive questions and might therefore be appropriate reviewers for that area.
But most important was reading the manuscripts submitted to the journal, and the reviewer comments, which make clear what researchers find interesting (or not) and why. At most journals, manuscript reviewers are asked some version of the following two questions about the manuscripts they review: 1) Is it true? That is, are the data in the paper sufficient to support the conclusions as the most likely explanation for reality? 2) Is it interesting? That is, will the readership of this journal find the conclusions exciting enough to convince them that the journal is worth reading? It is the answer to the second question that relies on more intangible factors and is the subject of most of the disagreements that arise between journal editors and authors.
For some journals, particularly those at which the submission rate is high and the acceptance rate is low, the editors have to make a decision on whether a manuscript merits the time and effort spent on the peer-review process. Based on experience with other manuscripts in a particular area, they ask themselves, assuming that this work is true, is it also interesting? It is at this point that a well-written cover letter accompanying the submission can make all the difference. An explanation of why the conclusions are interesting and who (outside the obvious) will find the paper worth reading can make a compelling case to the editor to send the manuscript out for peer review.
A good cover letter also suggests appropriate reviewers. The broader the content of the journal (and its likely readership), the broader the author should think about researchers appropriate to review their manuscript. The editor chooses reviewers who will have the expertise to answer both of the questions above. Again, it is often the answer to the second question that will convince the editor whether a manuscript should be published. The editor relies on reviewers who hold a similar standard for "interest" as the editor. At the same time, however, they are always on the watch for new reviewers to add to their list of experts.
After the reviewers' comments are in, the editor has to decide what to do with the paper. Typically, the answer is not a simple yes, publish, or no, reject. The more common response is some version of the following: "This paper is in an interesting area and describes some intriguing findings. However, they haven't done the key experiment that would allow them to make the great leap to a compelling conclusion. It is that conclusion that would make this a very interesting paper." The editor must then decide whether they agree with the reviewer and then try to find a way to convince the author that the key experiment is really worth doing to support this manuscript. It is the authors' response to this argument that typically determines whether the manuscript is eventually accepted for publication. If the key experiment is impossible (perhaps the necessary reagents or technology are not available), the author may simply submit the manuscript elsewhere and try their luck with a different set of reviewers or a journal with a different philosophy of what is "interesting." However, typically the authors know what the key experiment is, are working on completing it, and may or may not want to include those data in this particular manuscript.
Having been on both sides of the manuscript review process, one of the more interesting phenomena I observed was the assumption by authors that they can figure out who reviewed their paper and who must have been responsible for its rejection. As an editor, it was amazing to me how often their assumptions were wrong. In one instance an author told me that he knew Dr. X had trashed their paper because Dr. X had bragged to them about having kept it out of the journal. I checked and found that Dr. X was not a reviewer on the paper at all. I've also seen people veto reviewers based on apparently false impressions that the reviewer would not be impartial. Close competitors are not great choices for reviewers sometimes, but often they are in a great position to comment on the interest in solving a particular problem. Attending scientific conferences also allowed me to meet informally with authors and potential authors and explain journal policies and decisions we had made. I quickly learned the value of talking face-to-face with authors whose papers had been rejected.
I left the editing position to join the National Institutes of Health a few years ago, and I currently manage peer review for the intramural research program at the National Institute of Mental Health. I have found that many of the skills I learned at Neuron have been critical in my current position. Most important is understanding who your constituency is. Writing skills are part of this, but a smaller part than I might have assumed. However, communication with scientists to assure them that you are providing the service they need (in this case, fair and impartial peer review with rigorous standards) is essential. In addition, I brought knowledge of the issues and concerns that frequently arise during the scientific peer-review process. And finally, I figured out how to learn enough about a new research area quickly, so that I could identify the appropriate individuals to provide expert advice.