Thursday, February 26, 2009

Book Review: What Would Google Do? by Jeff Jarvis

In which I defend serious business thought under the guise of a book review.

Not to bury the lede: Jeff Jarvis's What Would Google Do? is entertaining enough to be worth a quick read from a library copy. This sounds ungenerous, but it puts the book squarely on the top percentile of the "Business and Management" literature.

Management and business books fall into four broad categories:
A vast majority has no content or bad content: wrong, unsound and dangerous, value-destroying, nonsensical buffoonery. These are books that cater to the prejudices of the masses, promise riches without work, or offer magic formulas for productivity and creativity and leadership. Many ghostwritten "autobiographies" of executives or entrepreneurs fall in this category as do most books about companies, industries, or economies written by journalists.

Most of the rest fall into the category I call "one-pagers," meaning their useful content can be summarized in approximately one page. Alas, to garner the full price of a book, two or three hundred pages are added to that one, generally over-taxing the credibility of the main point.

There is a small fraction of popular management, business, and economics books with good, rich content, usually tracking some specific industry, business, management practice, or research stream. Some recent examples are Dan Ariely's Predictably Irrational, Charles Morris's The Tycoons, Thaler and Sunstein's Nudge, or Ian Ayres's Supercrunchers -- all books popularizing a research field. These are the Carl Sagans of business and economics, and generally can be identified by the hooks to real research: references and footnotes, pointers to peer-reviewed studies.

Finally, there are textbooks and textbook-quality popular books. Many are expensive and all require a lot of thinking to really comprehend. Light reading yields almost no benefit other than learning a vocabulary; deep reading is like taking a good business course at a top-tier school, and equally hard. Examples are Varian and Shapiro's Information Rules and Davenport and Harris's Competing on Analytics. Reading both would build a solid foundation for understanding success in data-intensive network industries, but would take significant time, as absorbing these books requires many pauses to mull over.
What Would Google Do? is a one-pager and has the same problem as all one-pagers: to sell a book you need pages, you need broad applicability, you need a theory. Unfortunately, most popular business writers are what Paul Krugman calls "accidental theorists." (I'm not appealing to authority, just giving credit for an idea I wish was mine. Krugman is a master of pith.)

Accidental theorists generally lack the perspective necessary to analyze their theories. For instance, one of the rules in the book is "small is the new big," an idea with an interesting history. That history, apparently unknown to the author, includes decade-long oscillations in firm size and scope since the Korean War.

Rules such as "small is the new big," and its dual "bigger is better" (popular in alternate decades) are too simplistic to drive business decisions. But drive they did, for six decades; so we had concentration followed by unbundling followed by vertical integration followed by focus on core business followed by conglomeratization followed by strategic divestment followed by... you get the point. Here's a better rule:
Broaden the scope as long as the gains from positive spillovers make up for lost economies of scale. Increase scale as long as it is economical to do so, including the effect of networks: on the customer side (value of network), on our operations side (learning effects), and on the competitive side (creation of barriers to entry).
Not as pithy as "small is the new big," but useful. What is good for Google and what is good for Wal*Mart and what is good for Rick 'n Ann's are different things, depending on their environment, resources, and business objectives. Again, not as pithy as "small is the new big." But true.

Just to be clear, accidental theorist-ism is not a criticism of WWGD alone. It applies to most one-pagers, and WWGD is not a particularly egregious offender. But it happens to be the one at hand.

On to some problems with the book. Keep in mind that, unlike 99% of management books, this one is worth reading.

WWGD misses two important differences between Google and other New Web companies: first, the two-tier equity structure that allows a minority of equity holders to control a majority of votes; second, Brin and Page's smart decision to hire professional managers and advisors. The first allows core stockholders to control the firm and to implement their long-term vision with limited interference from short-term players. The second explains why Google's business decisions show a maturity much beyond the age of the company. These are the kind of business details that don't make for a good yarn unless the company implodes. So we never read them in one-pagers about success.

(They are also the kind of business details that determine success.)

Given Google's equity structure, it is rather ironic that WWGD posits as its first rule "give the people control and we'll use it." In the sense of the book, that collaborative filtering works well with large data sets, this rule is correct. But the pomposity of "give the people control" deserves a little deflating.

When I teach marketing I try to give the students a clear hierarchy: value proposition first, revenue model second. Mostly I'm unsuccessful, as students fall for "let's hold the value proposition constant and try to squeeze the customer more" CRM promises. But the general idea is that marketing is first and foremost about delivering value to customers. So, WWGD is right in that empowering customers is generally good. But not all customers are the same, and not all empowering is reasonable. There are three aspect to consider: cost/benefit in the transaction, cost/benefit in the relationship, and cost/benefit in the market ecology.


Yes, business thought is more elaborate than the Dilbertesque caricatures populating the press and the blogosphere. Marketers have known about transactions, relationships, and ecologies (not just networks of customers but broader ecologies of market agents) for some time. Good marketers consider a decision's impact at these three levels: sometimes, when catering to a customer is unprofitable at the transaction- or relationship-level, good marketers will still do it for the gains at the ecology level. To create positive word-of-mouth, for example, or to deny a competitor the opportunity to respond to that customer's needs and learn something about the market.

On the other hand, there are many situations in which unreasonable requests are to be politely refused. With the utmost diplomacy, of course. Identifying these situations requires both analytics and strategic thought. Not eight-word rules.

Moving past the "small is the new big" rule, since it's its decade to wreak havoc on well-scaled and well-scoped organizations, we find that "atoms are a drag." Again, true for content, though I'm looking at a flat panel made of atoms and typing on a keyboard made of atoms, writing about a pile of atoms called a book. The value here is in the information, not the atoms.

The problem is again the extending of a theme -- the value of information is not its physical medium -- to a broad domain in which the theme does not hold. As a result, WWGD falls for the trap of "post-scarcity economics." This raises a red flag. Granted, technological change moves production frontiers and some activities have quasi-zero marginal costs. On the other hand, most of what we do still takes time and requires effort or money; a sizable portion of the economy is atoms, not information (though information can lead to better use of the atoms); and most resources used to produce information are physically scarce.

Post-scarcity illusions tie in nicely with WWGD's fixation on gift economies and free stuff. The most over-rated concept since the mid-nineties is "free as a business model." It works for some companies and markets, not all. But most proponents of free don't understand how revenue models, in general, and price segmentation, in particular, work. Everyone knows that advertising-supported companies existed before Google; advertising is the revenue model of broadcasting. Yet, many one-pagers laud free-to-user as the new business model.

Free, that is, as long as it doesn't apply to the author's work. WWGD acknowledges this contradiction, but my point is broader. For example, Cory Doctorow, an advocate of free, does give his books away. But he makes money as a public speaker, which is his personal revenue model. Saying that authors can make money by giving books away, while having a different revenue model for yourself, is misleading.

What about the open-source movement, benefiting me, a user of Unix, LaTeX, Python, and R? Why are all those people offering software and technical advice for free? They too have a revenue model: reputation is worth money. Imagine a kid who is a great programmer attending a middle-of-the-pack college. How can she signal parity with MIT-trained engineers? Before, she had little alternative to credentials; now, she can present her portfolio of real software being used in real applications.

(Am I not writing this entirely for free, as there is no advertising on this site? Yes, but as a recovering former academic I still have some lingering histrionic personality disorder. 2000 words are cheap therapy.)

My point is not that Google's revenue model is just a very good implementation of advertising-supported content. That's obvious. My point is that different businesses need different revenue models.

(Annoyingly, chapters are not numbered, apparently having a high-level quick-reference system was not WWGD.) After page 102, the problems with overextension accumulate. WWGD suffers a "End of History"-level disturbance, in which what is hot in the author's intellectual circle now is touted as the future.

(I believe this is partly a marketing ploy. People who think seriously about business don't buy many one-pagers; the people who do buy these books are attracted by the companies and celebrities in it, so pandering to their prejudices is good product design. On the other hand, that pandering makes one-pagers free fire zones for the thinking-inclined.)

More to my point, trends in business thought and management science that started in the eighties are touted as the next big thing, which they appear to be because the author suffers from accidental theorist-itis. Borrowing from Krugman again, the problem with accidental theorists is that they'd rather read many tomes with impressive concepts (but no technical details) than take an afternoon to study the basics of the field.

As light reading, the examples are fun. But the author doesn't seem to know how data mining and collaborative filtering actually work, nor understand strategic thinking, which creates irritating, easily-avoided mistakes throughout the book. In many cases the examples include simplistic stories that a little business thought would reverse.

These stories come mostly from the author's desire for change (someone should write What Would Burke Do?), from company statements, or from web celebrities. There's nothing wrong with interviewing people or reading company statements, but to understand the workings of a firm, an industry, or an economy one must be careful and skeptic of those who have an interest in a particular explanation. The second half of the book thus misses the opportunity for proposing contingent rules: when would this work, when wouldn't it? What data would one need? How would we track progress?

To be clear, my problem is not with the ideas themselves, though some are bad, rather with the haphazard way in which WWGD forces collaborative filtering on every industry without understanding its environment, resources, objectives, value proposition, and revenue model. Basing decisions on good information and using technology to improve business is smart, but it must be tempered with consideration of other parts of the business environment and, yes, of those pesky atoms.

After page 210, where apparently the author buys the idea that watching MIT classes online is a good way to get an education (MIT learning rule of thumb: 1% is the class, 9% is studying, 90% is solving the problem sets), there are some "what if Google, run the world?" flights of fancy and a Gladwellian "but there are exceptions" for Apple Computer.

Of course WWGD wouldn't need the Gladwell exception ("this rule always applies except when it doesn't") if it circumscribed its domain to what Google does:
Google creates value by using collaborative filtering to maximize the probability that a user finds what she searches for; then captures part of that value by selling advertising. It does both exceptionally well, having invested in several other businesses that feed data into the filters. Its corporate structure protects its vision; its professional management is first-rate; and the corporate culture it fostered helps Google acquire the scarcest resource of the so-called post-scarcity economy: talent.
That, its implications, and a few examples would take about one page. Still, I recommend a quick read. The book is entertaining, it may spark some ideas, and there are some good stories in it; Mark Zuckerberg's art class story is bound to become a classic. Any deep reading will stumble upon the failings of its accidental theory, though.

That explains the quick read. Why read the library copy? It's not the money; it's the clutter. As overcrowding spread over my bookshelves, I made a rule: buy only books I'll reread. One-pagers don't qualify, I just keep my notes. (In the Great Purge of 2007, I donated or recycled over a thousand books. The library spares me such trauma.)

Sunday, February 22, 2009

Tempora mutantur, nos et mutamur in illis

The old blog is dead; long live the new blog.

As with the previous one, this is more of a tool to test and divulge ideas than a personal log. Topics will stay within the previous framework: Science, Technology, Business, Economics, Books, Music, Science Fiction, Logic, Statistics, Mathematics, Bon Mots, mostly.

Instead of a long bio, here are some revealing things about myself. First, I'm a member of five groups, and these were my joining statements:
Bay Area Classical Music Society Life is too short to listen to bad music.

Bay Area R Users I'm a recovering Matlab and Stata user.

Make S.F. - Bay Area Makers Sure I can fix that. Where's the duct tape?

The San Francisco Semantic Web Meetup I do research on various forms of machine learning and their implications.

S.F. New Tech Technologists build civilization, other people just live in it.

Second, some favorites. My favorite book of 2008 was Anathem, by Neal Stephenson; like Fraa Orolo, I too suffer from Attention Surplus Disorder. Neal Stephenson is my favorite living author, P.G. Wodehouse is my favorite author, and Hercule Poirot is my favorite fictional character. Johann Sebastian Bach is my favorite composer and Mozart's Requiem and Rachmaninoff's third piano concerto are my favorite pieces of music.