It seems like O'Reilly has greatly expanded its list of offerings over the last 18 months. At first I welcomed that development, as I had always found O'Reilly books to be by far the best in the field. But I find myself having to re-evaluate that.

The O'Reilly books that I have bought recently (Running Linux; Apache: The Definitive Guide) don't measure up to your old standards (though the Apache book obviously tries). It's a hard thing to pin down--something about the thoroughness and clarity found in Chapman & Zwicky that defined your product but that's missing in recent offerings. Not to mention simple error control, as on p.63 of the Apache guide, where the discussion/example about IP-based virtual hosts leaves me scratching my head (Is it 'sales.butterthlies-IP.com', as in the config file example, or 'sales-IP.butterthlies.com', as in the text explanation below? And how could www.sales.com possibly work at all?)

I still think your books are good. I just no longer regard the O'Reilly label as a guarantee of quality. Bummer.

Dear Steve:

It is certainly true that we have expanded our offerings of the last couple of years. We've also expanded the areas of interest we cover. I have no internal data, though, that indicates a decrease in quality in our offerings.

We monitor our customer complaints and the errors in our books. We have many more errors than before, to be sure, but not when you consider the volume of books we're publishing. If anything, our data indicates that we have fewer errors per book than we did in the past.

We know that the requirement to get information out to the marketplace sometimes means we rush a book. We don't like to do that, but I believe that timeliness is also a part of quality. If a book isn't on the shelves when you need it, it's cold comfort that, when it comes out six months later, it's extremely accurate. As it is, we publish later than most publishers, usually waiting until our authors have some significant experience with a technology before writing about it.

We try not to make errors, but we do on every book. Typesetting and proofreading code is especially difficult. Even when the code has been tested by the author, mistakes can be introduced in the production and printing process. Doing Python books, for example, is very difficult, because Python programs use indentation for meaning. Our tools and production editors labor over tab stops throughout all the Python books.

We have a process by which we fix errors in books before every printing, about twice to three times a year for most books. We collect errors found by the authors, readers, the editor, and collect them for the next reprint. Before the reprint, the author reviews the suggested fixes and approves or disapproves them.

Overall questions of quality, those areas which you sensed in Apache: The Definitive Guide and Running Linux but couldn't describe, are clearly more difficult to measure. To some extent, such judgments are a matter of taste or expectation and vary from user to user. Furthermore, they may represent differences between one book and another, not related to an overall change in quality. You mention the Chapman and Zwicky book. It is the product of two careful writers and one of our most demanding editors. I think its quality is so high that it would shine next to other books from any period of O'Reilly publishing. As for Running Linux, it seems to get very enthusiastic reader responses. It's been around since 1996, so I think we'd know of any major quality problems.

O'Reilly readers are not a docile lot. Apache: The Definitive Guide had some quality problems in the first edition; we weren't happy with it, even though it was the best Apache book available. I feel that the second edition fixed these problems. We're pretty happy with the current version. Both are well thought of by the readers we're familiar with, and both have been selling very well for years with few returns.

(Let me mention that new versions of Building Internet Firewalls and Apache: The Definitive Guide are underway. If you have comments about improving either of these titles, we'd be happy to hear from you.)

I'd be pleased to hear any more detail about Running Linux and Apache: The Definitive Guide that would indicate to us how they can be improved. I hope you'll feel better about O'Reilly quality in the future. We think our quality is higher than that of our competitors, and we want our readers to find that to be true as well.

Sincerely,

Frank A. Willison
Editor-in-Chief, Technical Publishing
O'Reilly & Associates

Follow-up article regarding this letter

Return to: Frankly Speaking