For the longest time, computer professionals have been accustomed to purchasing, hauling, and loaning (as well as trying to get back!) innumerable technical books essential for learning the knowledge of their fields, and for referencing specific details. These dog-eared and coffee-stained tomes fill our closets, weigh down our bookshelves, and litter our desks. Admittedly, these volumes can be fun to show off to colleagues, and can prove more effective than academic courses for learning new computer languages and other technologies. But storing and moving those books (especially to a new residence) can become a real burden. When they are packed away in boxes, they are inconvenient to reference. When they are placed on bookshelves for easier access, they are bulky and heavy. In addition, they have a tendency to collect dust, both physically, and figuratively, in the sense that they inexorably go out of date — victims of the rapid obsolescence of their topics, and the publication of newer and corrected editions.
That may all change, as more technical publishers offer their wares on the Internet. This can take many forms: A growing number of writers are making available entire books at no cost, typically downloadable from their websites, in PDF, HTML, or Word formats. If these pioneering authors' comments are any indication, this approach has allowed them to effectively spread their work and names throughout the technical communities, as well as garnering corrections and suggestions from countless readers worldwide. At a more corporate level, some major publishing firms now give away portions of a book, such as a few chapters, in an effort to entice readers to purchase the full volume. But even then, after a purchase, the consumer ends up with one more "dead tree edition" to add to their stockpile. For many techies attempting to locate that key command or programming solution, it can be a case of overkill: Having more pages to search through can actually slow down the process, and compel the frustrated reader to instead search through Usenet newsgroups or simply reinvent the wheel by coding from scratch.
A more creative and possibly more cost-effective approach is to offer books in their entirety, and as many as possible. Embracing this strategy, O'Reilly & Associates, the well-known technical publishing firm, has teamed up with The Pearson Technology Group, to offer Safari Tech Books Online, a subscription-based virtual library. Safari currently boasts over 1200 books, all accessible and searchable via a Web browser. That number is growing at a healthy pace, as they are currently adding more than 10 books per month, on average, to the lineup. Naturally, a large portion of these books is from O'Reilly itself. In fact, approximately three-quarters of their catalog is now part of Safari. Yet the titles are not limited to those published by O'Reilly. Rather, users of the system are able to search and read books from Addison-Wesley Professional, Adobe Press, Alpha, Cisco Press, Macromedia Press, Microsoft Press, New Riders, Prentice-Hall, QUE, and SAMS.
Subscriptions to Safari can be obtained at various levels of price and corresponding number of slots in an online bookshelf. They can be purchased at the level of 5, 10, 20, and 30 slots; and the monthly fees are just under $10, $15, $25, and $30, respectively. Most of the titles consume one slot, while a few small "pocket" books fill one-half of a slot. Hence, a 10-slot subscription for example would usually translate into a maximum of 10 books that a user could have checked out during a 30-day period. At the end of the 30 days, the user could elect to remove any or all of those titles from their bookshelf, and then later fill those emptied slots with new selections.
As can be imagined, there are many advantages to using a Web-based technical library such as Safari (which may be the only one presently in existence).
One clear benefit is the ability to read up-to-date books at a fraction of the cost it would take to purchase their non-virtual brethren. With more and more information being made available online, particularly technical information, this is clearly a trend with no sign of diminishing. For software developers and system administrators, being able to read books within a Web browser, provides the extra benefit of allowing the users to copy source code and other examples straight from the text, to be utilized in their own programs and scripts. This is the more advanced equivalent of the diskettes and now CDs found in the backs of a large proportion of technical books (which of course increases their production and retail costs). However, selecting and copying text from the Web is faster and more convenient than tearing open a plastic or paper sleeve to retrieve and load the CD. Moreover, the marketing blurbs on some books are vague in describing the CD's contents, which turn out to contain not the sample code in the text, but instead additional chapters that would have made the dead-tree incarnation even more bulky.
Some users of virtual libraries may argue that the primary benefit is not so much the saving of space and the copying of source code, but rather the ability to search through the library in a matter of seconds. This is far better than the tedious and error-prone process of using often-inadequate indexes to search by individual keywords, and then resorting to flipping through hundreds of pages to find the relevant material and sample code that one vaguely remembers seeing in a particular volume. Even better, the Safari user is not limited to searching through the titles on their bookshelf, but rather can opt to search all of the books within the Safari catalog. This probably helps Safari as well, because then the user becomes aware of — and is encouraged to choose for their bookshelf — publications that they may not have even heard of before. The system's home page has a single field that allows the user to do a simple search by keyword(s), and for indicating whether the entire catalog or just their bookshelf is to be the target of the search. A nearby link brings up the Advanced Search page, where the user can search in book titles and author names, or limit the search by category, ISBN, publication date, or publisher. In addition, the user can specify that all keywords be required, or exactly those words (presumably in the order given, i.e., as a phrase) or at least one of them. Both search modes have an option to restrict the search to source code fragments only.
Another handy feature is the ability to make notes and bookmarks in the selections on a bookshelf. Specifically, a Safari user can create — and have stored on the Web — private and public notes on any section of a book; and Safari will keep track of the note's date and title, as well as the book's title and section. The virtual bookmarks, too, are dated and given a user-chosen title.
Greater access to the contents of the books is definitely a plus. As long as the user has a connection to the Internet (and these days, who in the technical professions does not?), then their chosen books are literally at their fingertips. The books are available 24 hours a day, seven days a week — unlike the bookshelves of even the largest bookstore. While it may be true that the same 24x7 accessibility can be said of paper-based publications once purchased, it is important to bear in mind that when the typical knowledge worker is seeking information to solve a technical problem, they are already on the computer and looking at its screen. Thus, having to turn away and search for the needed information in one or more books (assuming they can locate those volumes) consumes much more time and carries a much greater risk of sidetracking the user and breaking their train of thought.
One advantage that Safari has over newsgroups is the higher quality of source code. Developers can better rely upon sample code from technical books tested by professional reviewers and countless readers, and supplemented with more extensive explanations and errata — as opposed to code culled from thrice-borrowed scripts, and usually presented with insufficient context. This is not to say that Usenet newsgroups cannot serve as a valuable resource for posting questions, (sometimes) receiving answers, and examining other people's source code. In fact, there are some extremely bright programmers participating in newsgroup discussions, and much can be learned from reading their postings, testing their given code, and separating the wheat from the chaff. Nonetheless, the thoughtful answers and solid code are easily lost amidst the hurried replies, untested code, blind alleys, misleading responses, and unfortunate arrogance and flaming.
Unlike many Web-based businesses, Safari has customer support staff who are readily available to help the user. They provide a toll-free telephone number for users within North America, and a regular phone number for people outside the region. The hours of operations are 6 AM to 5 PM, Pacific Time, thus making it easy for subscribers to contact them at any point during the business day, regardless of which time zone they are calling from. In addition, Safari offers an email address devoted solely to customer service. My experience with using that email address has been quite positive, in the form of several timely and positive responses. Furthermore, when my credit card billing hit a snag, the telephone representative provided flexible service to overcome the obstacle. This can certainly help assuage any frustration a user might feel if and when they have any difficulties using a service over the Internet.
Last but not least — especially when you have encountered a technical problem that is preventing you from accessing the critical information you need to do your job — the technical support staff of Safari seem to be willing and able to fix errors. For example, when I reported that the index of PostgreSQL Developer's Handbook was missing, the technical service personnel replied quickly, and eventually made the index available.
Like any product or service (and perhaps especially a fledgling one such as this), Safari is not without its areas in need of improvement. Perhaps the most alarming of all is the practice of including account passwords in plaintext in email messages. As a result, the password could be read by anyone who has access to the user's email account, or happens to read the message over their shoulder, or is actively sniffing Internet email traffic for the string "Password". These passwords are sent in the welcome message received after registration, and in the reminder message sent halfway through a trial period. The problem may stem from Safari not requesting that the new user select a custom password, when they first sign up. Instead, an easily guessable one is assigned to the user and of course must then be communicated to them. Of course the user can always change their password; but that does not entirely solve the problem if a subsequent email message includes their new password.
Most Internet-based services requiring the use of a password, are wise enough never to email people's passwords unencrypted. Those few who did make that mistake in the early days of such services, quickly received negative feedback from their customers and unhappy ex-customers. It might be argued that a Web-based technical book account, such as Safari, is not as critical to secure as one's online brokerage account, for example. Nonetheless, a thief could then easily login to your account, change the password to their own, fill your limited empty slots with their own choices, and effectively take over your account for their own use. No doubt Safari customer service would be eager to rectify the situation, but it would certainly be a confusing and time-consuming hassle.
Another annoying flaw is how some of the books' pages contain only a few sentences — sometimes only one. This appears to be a result of the formatting of the material by section (and naturally not by the page numbers of the paper books), and the apparent intent to maintain the same depths of the sections across chapters. (In this case, the term "section" refers to the lowest level division of the material seen when the user opens a part of a book, and then a chapter in that part, and then that chapter's sections, by clicking on the "+" symbols in the "Table of Contents" area on the left portion of a book's Web page.) These low-content sections would of course be easier and faster to read if all of the small individual paragraphs, fragmented among many pages, were combined into one or two pages. Fortunately, few instances of this problem have been found.
Without a doubt, the most inexplicable problem I have encountered is the inability of Safari to locate a certain book (from the Advanced Search page) when using its exact title. The following search entry turned up no results: Managing & Using MySQL. Yet this book is certainly within the Safari catalog, and can be located by scanning the book's appropriate area in the "Browse by Category" section found on the left side of the home page. Initially I suspected that the "&" character was throwing off the search (coded in SQL), but that does not seem to be the case, because another title containing that character, Web Database Applications with PHP & MySQL, was located easily. At this time, I do not know how many other books within Safari cannot be found using their titles. That is certainly one thing they could test for, once the cause of the bug has been resolved.
The Safari service welcomes users' recommendations for titles to be added to the lineup, from those that the participating publishers have not made available. There is, however, no guarantee that a particular title will be added upon request. That is certainly understandable. Yet it is not clear as to why some of O'Reilly's own best titles are missing from the catalog, and may never be added — regardless of whether the book in question is a well-regarded classic, or an especially useful newcomer. An example of the former is Mastering Regular Expressions; an example of the latter is Oracle in a Nutshell. I suggested to Safari that these two books be added to their catalog, but was informed that the publisher has no plans for doing so in the near future.
One minor problem, mitigated by the use of bookmarks, is that the apparent home page of the service does not appear to have a login link for users. Instead, users need to bookmark the address sent to them in the introductory email message, or use one of the two portals. It's not clear why there are multiple portals ("InformIT.com" and "Oreilly.com"). Some clients may be wondering if there is a difference in content (I was told there is not).
A visual fault that should be noted is how the Safari home page and other non-portal pages linked from it, presently use CSS to underline and change the color of entire paragraphs when a mouse pointer is positioned over each one. This is a good illustration of the Web design principle that just because something can be done on a Web page, does not imply that it should be done. Web designers learning visual effects for the first time tend to overdo them.
One would think that the "My Account" page would list the expiration date of the user's Safari account, but it does not. Consequently, if a subscriber failed to record the date that they signed up, then they could not be sure as to exactly when their subscription or trial period will end. Perhaps Safari is relying upon the reminder email messages — at least for trial subscriptions — sent halfway through the trial period, and then 48 hours before its termination. That approach would work fine for myself and other users who stay on top of their email messages. But it probably would not work well for those people who get behind in checking their pending messages in the particular email account that they used when registering with Safari. In addition, it was not easy finding the details of the alternate subscription plans (their number of slots, monthly cost, etc.). Too much information is deferred with the instruction to the website visitor to "contact us". That approach wastes the time of the visitor and the Safari support staff.
When testing Safari for the first time, I encountered several problems. For instance, the introductory email message gave the address "http://safaribooks.webex.com" for attending live tutorials over the Web. But when I initially went to that address on that same day, to attend a tutorial, an error message appeared announcing that the page had moved to a PHP page at "http://safaribooks.webex.com/safaribooks/site/". Fortunately, that address now forwards without generating an error message.
Another minor technical issue is that all of the links to top-level directories (e.g., "http://safaribooks.webex.com") are missing a trailing "/". This generates an error from the HTTP server each time, and forces it to resolve to the correct address (in the case of our example, "http://safaribooks.webex.com/"). This error is repeated throughout the introductory email message as well as most of the site's pages. In defense of Safari, it should be noted that they are not alone on the Internet in neglecting the trailing "/" in published addresses; though one would not expect it from a tech-savvy organization such as O'Reilly.
Lastly, I was dismayed by the absolutely abysmal underlying HTML code of the Safari web pages. It was ironic to be reading a Safari book on improving one's HTML — such as the excellent Refactoring HTML — and then to see that it was presented by horrid Safari page code. Doctor, heal thyself.
Admittedly, none of these are critical problems. But they give the user a sense that the service has not fully matured from a technical standpoint, and could benefit from some redesign for greater usability. Fortunately, none of these issues are significant enough to justify not recommending the service, or to pass up their offer for new users to try out the service at no charge for 14 days. Nonetheless, the weaknesses needed to be mentioned. In particular, let us hope that in the future, Safari will discontinue the practice of sending passwords unencrypted in email messages to customers. That would be especially upsetting to customers who employ the same passwords for multiple uses.
On balance, Safari Tech Books Online is an innovative and extremely convenient service that may prove to be a central force in transforming the way publishing firms distribute their technical works. In the meantime, computer professionals are urged to try Safari, explore its potential, and fully utilize its features to quickly locate technical keywords and sample source code. One of the many benefits may be that, in the future, the only time you will need to move those dusty old programming books is when you package them for shipment to used-book buyers, who have not yet discovered Safari.