close this bookVolume 9: No. 16
View the documentUpdates
View the documentPhilosophy
View the documentJournal calls
View the documentWeb authoring
View the documentBook review

"Philip and Alex's Guide to Web Publishing" is a revised version of Philip Greenspun's "Database Backed Web Sites: The Thinking Person's Guide to Web Publishing" that I reviewed in TCC 7.83. The new version (from a new publisher, Morgan Kaufmann) has over 265 elegant color and BW photos on close to 600 pages of 70 lb. glossy paper. It has new chapters on community, e-commerce, and political action, and I suppose the code examples have been revised -- yet all the text all seems familiar to the repeat reader. The chief differences seem to be much more professional editing and restraint, less chauvinism about Lisp and MIT hacker culture, and deletion of an ill-fitting essay by Noam Chomsky. Philip seems to have solved the "filler problem" of his first book, about which he said "Chapters 1, 2, and 3 are all just random flaming that I stuck in so that I wouldn't run afoul of my contract with Macmillan (calling for a minimum of 350 book pages). If I'd discovered the magic of screen shots sooner, I wouldn't have had to write this filler at all."

The original book was a sales pitch for online collaboration and communities. There's much less selling of the community idea in this edition, perhaps to fit in all the photos. Philip hasn't changed his belief that open-source groups can build better knowledge bases than can individuals, if properly supported. His website says, "I built this site in 1993 to share what I knew. In 1995, I expanded the goal to also share what some other folks know." This book documents some of Philip's contributions; for the rest, see his website.

Pages 494, 539, and 543 offer some nudes, which is certainly unusual for a technical manual. (The beach babe on p. 360 is even more striking.) Other pictures cover everything from sculptures to racing pigs, with lots of shots of Philip's dog Alex. These were selected from his collection, one of the more popular Web destinations. (I believe it's still getting something like 500K hits per day.) Philip didn't include a picture of himself nude with his dog, but you can see that and his other photos on the Web. You might start with , or with the history of Philip's publishing adventures on .

Philip wrote this book in HTML, with pictures embedded in the text, and you can read the whole book for free in this form at . The photos are uncaptioned and usually unrelated to nearby text. One of Philip's chapters is about how to create and display such high-contrast images. Although relevant to his life and work, here the photos seem irrelevant and egotistical. Fun, but still...

This is a self-indulgent book. (I decided this by page xxiv, before the author himself mentioned it -- as Edward Tufte's opinion -- on xxviii.) Philip calls it "the first coffee table Web nerd book." I call it Philip's geek autobiography. It's not about his family and childhood, but about problems he has solved and the code that he and his friends produced. It's also about how to be Philip Greenspun and why you would want to be. (In fact, he wanted to call his first book "How to be a Web Whore Just Like Me.") One could do much worse than to be Philip Greenspun, so it's a legitimate project.

Philip is an MIT professor, but that's nearly irrelevant to the text. What matters is that he and his partners in ArsDigita are top systems hackers who have implemented over 100 (now 200?) database-backed websites. With a few lines of code -- C, Lisp, Tcl, SQL, Perl, or whatever -- they can patch together Unix shells, Oracle databases, AOLserver Web servers, and other modules and components. These custom solutions do in milliseconds what commercial off-the-shelf software may take seconds to do. For sites serving millions of hits, that matters. It's a bottom-up software engineering approach to system construction, as opposed to commercial middleware purchase and integration. ArsDigita can now patch together previous solutions to create almost any kind of interactive Web service. Their hosted sites typically cost $30K-$80K/month. (See for their rates.)

Philip is still preaching that we should abandon the software marketing model based on package sales and instead have flat-rate access to all the world's software, with pooled revenues going to code owners in proportion to the use of their code. He argues that such a payment system would encourage open software development (or at least hooks and interfaces), code documentation, incremental remuneration for incremental work, and easy-to-use systems for all communities of users. Unfortunately, he offers no road map for getting there. Expanding the influence of ArsDigita and its solutions appears to be our best bet -- but ArsDigita is not structured for growth, other than by sharing its code with others.

All of the code described in the book is free, for those who can afford to run it. Or ArsDigita will build and maintain your website on their powerful Unix servers, if your budget is large and your problems pique their interest. This book amounts to a sales pitch, implying that there's nothing you and just a few of the world's most brilliant and experienced hackers can't accomplish (and that there's likely nothing much you *can* do reliably with only ordinary programmers and Windows NT). You should definitely read this book before you start building a million-dollar website.

If you already have a website, you may want Philip's code for Web-based community services: mailing lists, chat groups, discussion forums, Q&A forums, birthday reminders, etc. The "toolkit" code is SQL and Tcl, and can all be demoed and downloaded from or incorporated in your website for free via . "It really isn't any more expensive for this machine to serve hundreds of Web publishers and their many thousands of users than it would be to run a single database-backed Web site with 500 hits/day." The free services are the SPAM system for automatically maintaining mailing lists; BooHoo for automatically maintaining links; Loquacious for adding collaboration to static Web sites; LUSENET for database-backed threaded discussions and Q&A forums; for measuring traffic clicking from Site A to Site B; and Uptime, a service for monitoring your website's availability.

For the small commercial website, Philip's database-backed e-commerce strategies only work if you're willing to run your own Linux or Unix system. He tells me that "Linux is free and Oracle on Linux is free for developers. AOLserver is free and our Shoppe code is free. But the whole thing is a bear to administer compared to Yahoo Store." This isn't the primer you need if you just want an online store hosted by an ISP.

It's hard to rate this book. I've essentially read it twice now, which I've done with almost no other book since high school. There's unquestionably good technical content here, and I like the way it's tied to a rich website instead of just a CD ROM of aging software. Still, I have to wonder about the eye candy and the something-for-everyone approach. It's the antithesis of the functional website design that Philip espouses. (The eye candy is "served" instantly and at almost no extra cost. Is this like using flashy Web graphics, or quite different? Is it justifiable only because of Philip's chapter on photographic publishing technique, or can anyone play this game?) I'd rather see this as several books, each with a different theme and target audience. The current organization is like getting all of the parts for the price of one, but with no focus, limited development, and little ability to skip and choose. Is this a bargain, or a dilution? Having seen the author's strongly worded opinions in his previous book, I'd say it's a dilution. We need more books from Philip Greenspun, not just one bigger book. Perhaps we will get them, now that he and Morgan Kaufmann have pioneered HTML-based book publishing.

The title again is "Philip and Alex's Guide to Web Publishing" (Morgan Kaufmann Publishers, Inc., San Francisco, April 1999, 608 pp., paperback, $44.95), by . It's well worth the money. Order from for just $35.96. (Philip gets a little extra if you use this URL.) Amazon offers over 60 reviews of this book (and 42 of the previous one), and nearly all are five stars. I give it four stars by itself, but five if you include the website. And Philip deserves two thumbs up.

-- Ken