Schmidt & Rosenberg's How Google Works

[Non-fiction/business] [**++]

How Google Works is a book about how Google works! Since it is a work of non-fiction, it has a perfect title. The book tells you what to expect from it and delivers on its promise. Most equally accurate titles are far less specific, and the few titles that are as accurate and specific don't tend to be as pithy. How Google Works discusses some of the history of Google, some of the philosophy of Google, and some of Google's products, but it always does these things in the context of describing how Google works. It covers the basics of how Google hires people, how communication happens at Google, how workers are expected to behave, how people are managed, and how strategy is (and was) determined.

How Google Works is very well laid out. Apart from the introduction and conclusions, the book has six major sections called Culture, Strategy, Talent, Decisions, Communications, and Innovation. Each of them is about approximately what you would expect given the title. (These sections have sub-titles are pithy to a fault, sometimes slightly misleading and other times highly uninformative.) These sections also have sub-sections with titles that are meaningful later when you think about the book whether or not they are meaningful in advance. For example, "How was London?" is a sub-section that talks about Google's policy of breaking at the start of meetings with an informal question, such as a question about vacation. It goes on to tell the story of one person returning from vacation and giving a detailed report on what was happening in stores selling Android phones in London because the employee decided to visit those stores and talk to the sales people inside while he was on his vacation. Once you've read the book, the table of contents does a good job of reminding you what you read with considerable detail, even if many of them don't necessarily give you a good idea of what to expect. The book also has a good index.

Another great thing about How Google Works is that it is pretty fun to read. It has great illustrations. It has a lot of anecdotes, and many points of humor, most of which are on point, though a few are diversions. It also has the best brain-teaser I've ever read: one that gives you enough information to solve it without needing to know anything else (many brain teasers require you to know something like the fact that muscle strength scales sublinearly with volume). I solved it in twenty four hours, but I did have to sleep on it. I found it more challenging than any of the puzzles or exercises in Gödel, Escher, Bach which is one of the few other books with any comparable thought puzzles in it. It's also very satisfying since the solution doesn't depend on any tricks of semantics or anything like that.

Overall, How Google Works deserves several points for style. It gets pretty much everything related to style exactly right. The content, on the other hand, is a mixed bag.

I'll start by talking about innovation. The basic premise How Google Works is that Google knows what it's doing in all of the topics discussed, so you should follow Google's example. This is not an assumption I am making, the book actually switches back and forth between the descriptive and the imperative voice. The authors don't just tell you how Google works, they tell you while very clearly stating that you should go and do likewise. Schmidt and Rosenburg can get away with this because everyone knows that Google is a great company that really understands innovation, hiring, management, and all that. However, the fact that this is common knowledge doesn't make it correct.

Google is an incredibly successful company that has a many exceptionally successful services.

Before its IPO, Google created its search engine, Gmail, AdSense, and AdWords. These properties account for most of Google's profits. Though YouTube is an exceptionally profitable property that Google owns that I did not mention, and Android makes all of Google's other properties more valuable

Since its IPO, Google has created quite a few products and made many acquisitions.

YouTube was an acquisition. Android was an acquisition. Drive was an acquisition. Nest was an acquisition. Pretty much all of the things that Google has launched since its IPO that have done really well financially have been acquisitions. Google has launched a lot of cool products internally including Project Loon, Google Books, Google Fiber, Google Glass, and self-driving cars. But the evidence that these projects contribute much to Google's profitability is pretty lacking. Even more of Google's internal projects from this time period have been complete flops. Even Google Code which seemed like it was going to be a big deal when it launched is now defunct having been completely overtaken by GitHub (which is well on its way to becoming a wildly successful company).

Google Chrome (taken together with ChromeOS) is the only extremely successful Google project that I can think of that was created internally by Google since its IPO, and even that doesn't seem to have much profitability apart from adding value to Google's other products, most notably search, AdWords, and AdSense. Another internal project that is almost certainly profitable with good margins that Google has launched internally since its IPO is GAE, but that that is getting trounced by both Microsoft and Amazon's cloud offerings.

So the question of whether Google is a company other companies should be emulating (in terms of policy) if they seek to excel at innovation deserves to be answered before the suggestion is made, but Rosenberg and Schmidt don't acknowledge the question. The 800 employees that worked for Google in the six years before its IPO seem to have done a much better job of figuring out how to create extremely profitable transformative new technology than its 53,000 employees have in the past ten years. (That number is growing. 53,000 is how many Google had in 2004; not the average number of employees Google has had over the past 10 years, just like 800 is the number of employees Google had in 2004 rather than the average from the time it was founded in 1998 until its IPO in 2004.)

How Google Works does share a fact that seems to have a pretty strong explanatory power for why the employees hired post-IPO have been so much less innovative for Google than its pre-IPO employees. Google has a policy against rewarding its employees for the success of the projects that they start in their 20% time, and Google advocates this policy. All of the projects I've mentioned as having been extremely successful (including ChromeOS) were initiated by people who stood to gain a lot from the success of the project. The people who created projects that were acquired were compensated handsomely. Google's pre-IPO employees owned significant amounts of Google that stood to gain considerably. AdSense as a standalone product is not very valuable, but it adds a ton of value to search. Google's pre-IPO employees stood to gain a lot more by building these technologies for Google than they did in attempting to create their own start-up to run it. The same is emphatically not true of employees working in their 20% time, as a matter of Google policy.

This is the fundamental problem with How Google Works as a book. Google's original products are among the most profitable products in human history, and they give some fantastic self-perpetuating advantages to whoever dominates the market. Google's upper management also has a very good eye for the future of technology that has served them well in their acquisitions.

Google was well on it's way to becoming extremely successful in 2004, and it's hard tell whether its unconventional management practices have increased its successfulness or whether Google would have been more successful if it had a more conventional approach to management. How Google Works does go part of the way to answering that question when it talks about Google's culture and specifically the philosophy of Don't Be Evil. It tells us that Larry and Sergey knew they could afford to do things with Google that most managers at other companies couldn't afford to do. Google is Larry and Sergey's utopia, not an embodiment of what they think leads to financial success. That's what Don't Be Evil means. They know that most companies are forced to make decisions based on economics that Google can make based on principal.

So when Google tells you that it hires "smart creatives" (which is Google short-hand for highly intelligent extroverts who are not conscientious), they are not telling you that this kind of person makes for a good employee. They are telling you that these people are who Larry and Sergey consider ideal. As the linked Wikipedia article notes, extroversion does not correlate with success but conscientiousness does strongly, and it does so even more strongly after adjusting for intelligence. When Google tells you to tolerate divas (Google short-hand for people who have earned the right to be obnoxious), they aren't telling you that divas have a tendency to add more value than they destroy. They are telling you, that Larry and Sergey have a soft spot for divas. (Accounts of the two of them in their Stanford years suggest that both of them could be described by that term.)

This is not really a criticism of Google. Larry and Sergey have earned the right to build their own utopia. As I mentioned when discussing Google's acquisitions, they also them extremely well-suited to lead a tech company even if they don't seek to manage it optimally. An organization as profitable as Google needs to be run by extremely talented investors much more so than it needs to be run by profit-driven managers if it is going to maintain its profitability. Google's founders have were not only brilliant founders, they have continued to be visionary leaders of their company. Their management philosophy would drive most companies into the ground and they know it, but Google just simply doesn't have to worry about profitability the way that most companies do.

This brings me back to the high points of the book. How Google Works gives you a lot of vision into how Google made decisions its pre-IPO days, it tells you a lot about how Google's upper management still works (after all it was written by a couple guys who are much more familiar with how Google works at the top than they are familiar with how Google works at the ground level -- this is a necessary truth about upper management at large companies, not a criticism of Google, Schmidt, or Rosenberg), and it gives you a lot of insight into the general shape of the future that Google's leadership is looking for as they guide their company forward.

As is the case with anything a book does well, it covers the subjects much better than I can hope to summarize them in my review, but I'll quickly reference some of the highlights. Great leaders have a knack for soliciting criticism and responding to it well. This is especially true of people whose leadership relies extensively on their ability to make accurate judgments and predictions. How Google Works tells you how this is done at Google. Visionaries also tend to have very specific things they are looking for in ...

The first is simply a quibble with terminology. Pseudo-mathematical terminology appears to be in vogue, even when it is misleading. At Google (and more generally in the Silicon Valley -- Peter Thiel uses the same term), people refer to thinking about transformative changes as 10X thinking. Contrary to what the name implies, 10X changes refer to changes that are differences in kind rather than differences in degree. They refer to optimizing along  a different dimension than the competition (as Christensen explains so well in The Innovator's Dilemma -- without using the silly pseduo-mathematical misnomer). Back on the plus side of the same phenomenon, How Google Works explains the concept of using technical insights to inform innovation, and in the process gives the best articulation of how to think about these sorts of transformations that I have yet encountered. (Technical insights -- also a Google term -- is a much better term than 10X thinking, they are somewhat related to each other, but not identical.)

The second criticism is also a complaint about terminology, but it's more than just a nitpick. Apparently, at Google, they regard the word data as the plural of anecdote. (They point out that datum is the singular of data themselves. That is not my gripe.) The problem is, they completely overlook the part where you correlate data with the rest of reality to see if your predictions make sense!

They talk about data as objectively measurable qualities, but never once in the whole book do they mention ever having bother to check whether the objectively measurable qualities that they look for ever have any impact on the results that they hope to achieve.

So ultimately, as you come to know How Google Works, you learn a lesson that is not very applicable to places outside of Google. Google works because it has brilliant founders that happen to have a lot of accurate opinions, and it has processes in place to ensure that its founders' beliefs guide most of what happens inside of Google. People looking to learn how to make better judgments will find many interesting ideas in How Google Works but would ultimately be better served by reading Feynman, Judea Pearl, or Less Wrong.