At the beginning of this year I got the idea that it might be a useful service to our community to read and review the HPC books published in 2008 and early 2009. I was surprised to find about a dozen, and even more surprised to find the publishers willing to share copies with me for review. Now the sideboard in the good dining room (where "good" is universally understood to indicate a room we never actually use) is stacked high with books and I'm continually gnawed on by a sense of dread when I rush by in the night to get a drink of water.

But while I was researching what had been published I noticed a pattern. Of the dozen or so books published that I thought would be of interest to HPC, CRC Press (an imprint of Taylor and Francis Group) was the only publisher to have many titles in 2008. Looking back over some of the recent catalogues, it became clear that CRC has an emphasis in HPC and supercomputing that the other publishers don't have.

Regular readers know I'm a curious sort, so I'm interested in learning more about the why's and how's of the supercomputing program at CRC. I hooked up with Randi Cohen, the Computer Science Acquisitions editor there, and peppered her with questions about what she does, and why. I think the answers are very interesting.

West: Do you aim to publish a certain number of books with a certain timeframe (each year, or several years)?

Cohen: We are publishing books in this area on an ongoing basis. We don't have a specific number of books that we plan to publish each year, but right now, on average, we have been publishing around three to five books each year, with many more forthcoming. In 2006, we started the Chapman & Hall/CRC Press Computational Science book series, which is edited by Horst Simon at Lawrence Berkeley Laboratory. The primary goal of this series is to publish books in all areas of computational science, including high performance computing, parallel and distributed computing, grid computing, applications in engineering, physics, climate modeling, and much more. Right now, the series has 14 books, either published or forthcoming.

West: How do you decide what goes in the catalogue? Do authors come to you with ideas, or do you survey the field, identify holes, and try to develop new titles by finding authors?

Cohen: Some authors come to us with book ideas and proposals for new books, and we also seek out authors to write for us. We attend many of the major conferences in the community, including IPDPS, Supercomputing, SIAM CSE, and SciDAC, where we generate ideas by attending sessions on the latest research in the community, and by talking to attendees about possible gaps in the literature. We also do extensive market research, on a regular basis, to identify areas that could be ripe for the development of new textbooks and reference material.

Our decision on whether to publish a book is based on market research and extensive technical reviews at both the proposal stage and during the manuscript development. We only publish manuscripts that have been thoroughly and technically reviewed by other experts, and cover topics that we have determined will fill a market need, according to our market research.

West: Are the folks that look after the HPC and computational science books past practicitioners? What's your background?

Cohen: The editors who handle books in this area, including myself, aren't specialists in HPC. We rely on technical reviewers and our series advisors to help us determine whether the projects we are considering are technically sound. Personally, I have a publishing background mainly, and have worked in scientific publishing for most of my career, so I have a good insight into what makes for a worthwhile project from the publishing and marketing perspective.

West: HPC is a niche within a niche. Why is CRC making such a big investment in books for HPC?

Cohen: Our main goal, as a scientific and technical publisher, is to provide resources for the scientific community, and we are particularly interested in areas of growth and new research. We feel that HPC is an important and growing area in computer science, and judging just from the attendance at the Supercomputing conferences alone, there is a lot of excitement and interest within this community. We decided to focus on HPC a few years ago, as we felt it was good timing to tap into a market where we see a lot of potential for growth. There is also a lot of interest in HPC, beyond the core community, with emerging applications in engineering, computational biology, physics, etc. We see only continued opportunity for new book projects in this field, and we are invested in the continued growth of our program in this area, for the long-term.

West: Do you have a sense for where the books go? Are they bought by students during school, or by professionals for reference? Whichever it is, has it always been that way?

Cohen: Our books are mainly sold as textbooks for upper undergraduate and graduate courses, as well as reference books for professionals. We sell most of our textbooks through adoptions at key schools, while our reference books are sold directly by us to the end-user, through our marketing promotions, our Web site, and key conferences. We also sell many of our books through resellers, such as Amazon and Barnes and Noble. Many of our key reference books are also promoted and sold to key libraries.

West: How do you market the books to an HPC market? We're a pretty small community, and we don't get out much. How do you reach us?

Cohen: We have a variety of marketing avenues to get the word out to the HPC community. Regularly, we send out brochures, catalogs and emails about key titles to the major societies serving the community, including the ACM, IEEE, and SIAM. We attend many of the major conferences in the community, where we promote our books and offer discounts. We also do ads in key publications, both print and online.