I have published four books with O'Reilly, and I have had brief encounters with other publishers. As a book consumer interested in cloud computing, Java, scripting languages, I look to Manning, Prag, and O'Reilly most of the time, and I'm also impressed with books from AW. Here are my experiences, maybe they might help:
When I wrote Jakarta Commons Cookbook, I had no agent, I worked directly with a great editor, Brett McLaughlin. Brett has since moved on from animal books and is now focused on books in the Heads First series. I would have never finished my book was it not for Brett's attention and guidance. If you are new to the book writing process, you will want to find an editor who knows your technology and who believes in the idea behind your book. I had maybe 5 reviewers, the book sold something in the range of 6-7k. While the publisher didn't view this as a success, I was satisfied with the sales and exposure, and I enjoyed the writing process. This book was written in Word, to properly insert XREF (cross references between sections and chapters) I had to load the entire book into Word and then run some hefty macros. I was constantly freezing the machine and Word was much more a nuisance that a helpful tool.
The second book I wrote as "Maven: A Developer's Notebook". To say it didn't go so well would be a dramatic understatement. I was less that satisfied with the book writing process, there were too many reviewers. Part of the problem with this book was that the book covered Maven 1 which was already on its way out. Maven 2 was released the same week that this book on Maven 1 was released. Sales were not very good; in fact, less than six months after printing, a number of people (myself included) were recommending that people avoid purchasing this book. I didn't actively seek out this second book, that should have been the first warning sign, I was recruited by my editor to help smooth out the writing. This book was written in Word using the O'Reilly macros, we had endless problems with Word. I had an agent for this book from Studio B, and the only time I spoke with my agent was during the contract negotiations. Studio B is "ok", but I don't think you need an agent to write a book, maybe someone can convince me otherwise?
The third book I helped to write was also something that my editor suggested. Jim Elliott was updating Harnessing Hibernate, and they wanted someone to write some chapters on Spring integration. This was my first exposure to using XMLMind and editing DocBook directly. Jim is one of the brightest authors I have ever worked with, and his colleague Ryan Fowler and I quickly started to use the DocBook XML from the first edition to update the book for a Second Edition printing. The writing process was a bit prolonged because two of us had some big distracting "life events" during the writing process. I'm proud of the end-product, and writing the book in DocBook was an eye opening experience. I would suggest that you get into the "craft" of book writing, code the book in DocBook using something like XMLMind, learn how to create and index, learn what it takes to create a book with all of the necessary markup. Writing a book is an entirely different beast from throwing some words into a Word Processor, and there's something to be said for understanding the entire process from start to finish.
The fourth book is an almost entirely different beast. Because of my experience with Jakarta Commons Cookbook and Maven: A Developer's Notebook I didn't want to be involved with a book that wasn't open source from the very beginning. The fourth book I'm involved with is Maven: The Definitive Guide, it is a comprehensive reference for Maven. This book is available for free from http://books.sonatype.com/ and you can also purchase a printed copy from O'Reilly. The book is an open source project on GitHub, the book is covered under a creative commons license, and we've attracted a global team of people who are translating the book. There are still some conflicts between the idea of giving away a book for free online and selling a printed version, and I think there is still work to be done at speeding up the pre-production process. The biggest difference between the fourth book and the first three is that it is an open source book which has a life beyond print, it is much more satisfying to be involved in a book that can be constantly updated and which can attract a community.
- Find a publisher that has technical editors. Two of my editors at O'Reilly (Brett McLaughlin and Mike Loukides) are capable technologists in addition to being great editors. I also know that there are others on staff who make a constant investment in knowing technology first-hand. I heard good things about Prag and Manning as far as editors go. I'm constantly impressed by the quality of AW books wrt tech.
- What are your interests and motivations? Really work out why you are interested in writing. If you are interested in writing because of the "reputation" or because you think it will make you a boat load of money, you are going to be disappointed. You should write the book because you have passion for the subject matter. I write because I see an absence of well-written open source documentation, I would write even if it wasn't something that I was paid to do. I often cringe when I see someone pimping the fact that they are "a published author". Writing is a humbling experience, and if you are not in love with the idea, it will break you. I can't tell you how many people I've met that tell me that the first book they wrote almost drove them insane. Most publishers are going to know this, and honest publishers are not going to let a first time writer commit to an impossible deadline.
- If you are writing a book about a fast-paced open source project, propose an "open book" to your publisher. Some publishers are more amenable to this idea than others. It is a paradox, no doubt, but I'll bet you that there are more than a few copies of the Subversion book where you work. I think open source books are important for the communities they support. CouchDB will have more end users and more developers because they have an open book published by a reputable publisher.
- Don't be discouraged when your publisher wants to know about things like audience size and potential market. There are publishers that consistently publish books that sell hundreds of copies or even tens of copies. The reality of publishing is that the publisher has to know that they are going to sell a certain number of units for something to be worth the money it costs to proofread, format, and print. Be prepared to work with your publisher on electronic delivery formats, the days of printing 10k copies of a book and shipping those books to retail book sellers like Borders and Barnes and Noble are about to end.
- A mind is a terrible thing to waste, on Microsoft Word. Find a publisher that will not force you to use Word. I did it for two books, and it was a soul crushing, time wasting hassle. Word (or even tools like Pages or OpenOffice) were not developed to support writing a 500+ page book full of cross references. My advice is to dive into DocBook, it is not "easy", but after you write a 500 page book you will realize that WYSIWYG tools are the least of your problems. The toughest part of writing is staring at a blank page and wondering what comes next.
- Find a publisher that can attract good technical reviewers... and make good use of them.
- Agent? I just don't think you need an agent, they will take 15% off the top of an already meager royalty. If you expect to make "real money" off of your book, get an agent so you can make sure that you won't get screwed, but understand that you probably won't be making "real money". People who make real money off of books are people like Stephen King writing a horror novel or Elliot Spitzer writing a memoir of public failure. Unless you happen to write that one in a million Perl book that sells 500k copies, your book will not be a primary source of income.
Disclaimer: I'm partial to O'Reilly books, I grew up reading my father's programming books and I'm also an infrequent contributor to the website.
"The more that came out, this was before he'd even given up the passwords and so forth, the stories I was reading and the information I was able to glean from, the court records, etc., was sending up red flags all over the place. They were claiming the now famous 1100 modems that they initially said Terry Childs may have planted around this city and a whole host of other things that just didn't jive."
Venezia writes in "Could the Childs case put all network admins in danger?":
...if Childs is convicted on the modem charges, then just about every network administrator in the world could be charged with the same "crime." You might as well start arresting carpenters for carrying hammers and saws because they could be used as weapons.
Link to Original Source
Link to Original Source
Link to Original Source