rblog

The Big Switch: Rewiring the World, From Edison to Google
Author: Nicholas Carr
Amazon info

I enjoyed the famous Harvard Business Review article by Nicholas Carr "Does IT Matter" and generally agreed with the premise that developing ones own systems in-house is generally a bad idea. Leave that to the software vendors and you will have fewer headaches in the long run. So, I was favorably predisposed to Carr when reading The Big Switch. However, as I read, I became increasingly disillusioned with Carr's negative perspective buttressed by relatively few experts and supported with little or no original research. Editor's note: Richard thought that Carr was a professor; this book makes much more sense given Carr's actual background as journalist - as one that is more "provocative" designed to sell and influence, rather than one that is reasoned and analytical. While there is a logical thread to Carr's arguments, it is a very thin line with some rather broad jumps.

The high points of the book are the historical pieces - both the actual history of electrification and the attempt to apply the lessons of history to today's internet environment. While I think the attempt fails, it certainly is worth writing a book about.

Since this is a business book, I have provided a chapter by chapter summary - along with some additional notes.

Part 1: One Machine
Chapter 1: Burden's Wheel - The basic analogy is created - electricity and computing. Tracing the history of power starting when factories produced their own power from natural sources (waterwheels, windmills) then to factories producing their own power using on premises electrical equipment (purchased from Thomas Edison's company) and finally to today - factories using electricity produced by a utility. The author then applies this concept to computing. Originally we used natural computers and calculators (human beings) then companies bought their own computing equipment (from IBM, Dec, Sun, others) and now/soon/perhaps companies will use "utility" computing - a combination of Software As A Service (SAAS), outsourced computing, utility computing (e.g. from Amazon and others) and a number of other buzzwords (Web 2.0, Cloud Computing, SOA). By the way, Henry Burden was an engineer/factory owner who built an enormous water wheel to run his company in 1851.

Chapter 2: The Inventory and His Clerk - The different approaches of Edison and Samuel Insull. Edison is making electrical gear, while Insull goes out and invents the electric utility. The key enabling factor is the economy of scale. When generating energy using large dynamos, the cost of each kilowatt of energy decreases the larger your dynamos are. So, by centralizing the production of energy, you get the lowest cost per kilowatt. To give you a sense of the speed of takeover of utilities, Insull takes over Chicago Edison in 1892 - it is one of 20 small utilities in the city of Chicago. Renamed Commonwealth Edison after the turn of the century, it increased output by 45x from 1899 (10kWh) to 1015 (450kWh). And, in general, the utility share of the market went from 40% in 1907 to 80% in 1930. Utilities won the game.

Chapter 3: Digital Millwork - Herman Hollerith invents the punch card to help with the 1890 census enabling it to be completed years earlier than the 1880 census, at a savings of $5M(!) - and that was with the country growing by 25%. Hollerith merges with another company and IBM is born. Carr traces this history of the computing and points out that in the 50s/60s/70s having your own software could be just as big an advantage as Burden's huge water wheel (which enabled him to produce ironworks more cheaply than competitors). Carr has another book "Does IT Matter" which argues that proprietary software is no longer worth it - don't write your own accounting software - buy it from Oracle and adjust to their ways of doing things (I agree with him for the most part on that). The PC and client/server computing enter in the 1980s and the computer gets more personal. Carr argues that computing also became far less efficient - he appears to be talking about computing power being wasted on servers that are only running one application and having CPU utilization at 10%. He then connects this inefficiency to low bandwidth between the client and server. Then comes the fiber optic boom during the Internet bubble. Now the bandwidth problem is solved, setting the stage for utility computing.

Chapter 4: Goodbye Mr. Gates - Documenting the rise of "utility computing". Bill Gates sends a memo on October 30, 2005 highlighting the rise of a new force that threatens Microsoft's most lucrative products (e.g. Office). Meanwhile, Google is building truly enormous server farms to handle its impressive search capabilities. Carr makes the argument that Google's cost of computing is 1/10 of that for a typical company. But the most compelling example of utility computing is SalesForce.com - one of the first software companies to build what is now termed "Software As A Service". You don't buy the software and install it on your servers - you run the application from a browser where it accesses your data stored on their servers. Upgrades are invisible and automatic, customization is provided, backup and maintenance are handled for you, there is no client software to manage. You pay a monthly fee for the service. All in all, a great idea - if the performance and reliability are there - and if you can live within the basic parameters of capabilities that the service offers. But of course salesforce.com doesn't help you with anything that is not sales force automation. Amazon is moving toward a more generic utility computing - starting with Amazon Web Services (for Amazon based processing) to S3 - which is generic file storage services and then to EC2 (Elastic Compute Cloud) - which provides a computer for you to use as a processing engine for 10 cents/computer/hour. Virtualization and thin client technologies are also discussed - all in line with the argument that the underlying technologies have matured to a point where utility computing can be realized.
This is where I start to part ways with the author. Electricity is generic, but computing is not. I can't use Google Search to run my financial analysis and I think the difference matters. I use Google's gmail and even some of Google Applications (replacement for Office), and I use storage from Mozy - but this is still different than doing all my business computing in the cloud. I do believe in SAAS and think that will be a big win. But I don't see generic utility computing any time soon.

Chapter 5 - The White City - The 1893 Columbian Exposition (celebrating the 400th anniversary of Columbus) - so brightly lit it was called "The White City". People were full of hope and optimism about the power of electricity - it would cure all of society's ills. And the results of electrification are so widespread that they seem "built-in" to our culture. But Carr begins to point out some of the negatives - certain jobs and industries were eliminated (in the name of progress), other jobs were deskilled (think of the assembly line) and even the labor saving appliances didn't make women's work any easier. Higher wages for those employed did create a burgeoning middle class - but with the car (enabled by low electricity costs) the middle class fled the cities. Carr contrasts the initial optimism with what he believes are more mixed results.

Part 2: Living in the Cloud
Chapter 6: World Wide Computer - Carr traces the beginnings of the web to its Arpanet days and highlights the idealism that people felt for what the net was bringing to them. He frames the next chapters as a discussion of the two opposing forces of the web - personal liberation vs. bureaucratic control. But this chapter focuses on how the web can be seen as a giant computer. In fact, he quotes Sun's old slogan - "The Network is the Computer" as an apt description (sadly he gets the decade wrong, it was a slogan in the 1980s when I was still there, not in the 1990s as he claims). For the web to be a big computer it has to be, well, big and a computer. Clearly "big" is not an issue. So, the question becomes to what extent is the web a programmable computer. Carr provides several examples of the "programmability" of the web - ability for users to generate their own avatars in Second Life, companies creating geographic mash-ups using Google Maps (he used SalesForce.com as an example, but Zillow is probably better known), the CERN Hadron Collider using millions of volunteer PCs to crunch data (a la SETI@home or Folding@home). Carr also claims that the web is programmable by individuals who can use freely available hosted tools like Blogger to create blogs that reference YouTube videos, Flickr photos, provide an RSS feed, etc - at no cost. This chapter also raises the question - yes, the web is pervasive and efficient - but it is a force for good.
Do we really have a programmable World Wide Computer? I don't think so. I have used nearly all of the tools mentioned in this chapter and I don't feel like I have a WWC at my disposal. Yes, I certainly have the capabilities to start a small business using free software (as described in the next chapter) - but it is a bit of a stretch to say that just anyone could figure out how to put it all together (maybe just anyone under the age of 25). But perhaps I am being too literal in terms of "programmability".

Chapter 7: From the Many to the Few - At this point the narrative takes a decided turn to the negative. The chapter starts with a story about Google's acquisition of YouTube - marked by a YouTube video of the YouTube founders thanking the people that uploaded videos (and they get characteristically goofy). But the point is that there have been several companies where the founders have made significant money due to the contributions of the crowd (YouTube, MySpace, FaceBook, Wikipedia). In today's world, you can use the "World Wide Computer" to start a business that is fundamentally digital - the only physical asset being the software you create. As such, there are increasing returns to scale as there are basically no costs associated with your product (not even CDs to burn or documentation to publish). Although this might seem like utopia, Carr points out the negative side. The newspaper business is dying, assembly lines still exist - they just don't have any people any more. Computers and software are replacing workers (along with outsourcing) and the new jobs that are created are jobs that can be done by computers. Craigslist manages 10M visitors per month looking at 5 billion pages per month - with 22 employees. In addition to the job loss, the "gift economy" of crowdsourcing, where people contribute their efforts for another person's profits enlarge the divide between rich and poor and the elimination of the middle class - all dire results.
Remembering that Carr is a journalist is (in hindsight) helpful to understanding the negativity of this chapter. Yes, there are some individuals who got extremely wealthy due to contributions from others - but there are very few (and I think Carr found them all - fewer than a dozen). And I believe that all of them started out simply to build something cool - and some have bypassed the riches (e.g. the Wikipedia folks, and I don't think Craig Newmark has sold yet). As one of the people who is considering working in the gift economy, I believe that economy only works if you don't think you are a sucker and that others will rake your profits. I also contend (without proof, but Carr doesn't offer much proof either) that perhaps the growth in the economic divide can be attributed to 20 years of Reagan/Bush administration policies that reward the wealthier segment of society.

Chapter 8: The Great Unbundling - As mass media moves from analog (e.g. newspapers) to digital (e.g. websites) the product becomes unbundled. Users easily find the articles they want without having to deal with the "whole". While this seems convenient, Carr points out that it eliminates some hidden subsidies. For example, classified ads subsidized in-depth investigative reporting. But as classifieds lose out to EBay and Craigslist, newspapers may not be able to provide high quality news content - at least not for free. Additionally, more and more outlets are available for your news and information. While this might seem good at first, Carr brings in the work of Nobel prize winning economist Thomas Schelling to show that small preferences lead, over time, to dramatic differences. In a simulated housing experiment, a preference to have have at least 50% of your neighbors be of your race will quickly turn an integrated neighborhood into two segregated neighborhoods. Applied to the world of information on the web, Carr sites studies that show that liberal blogs/sources tend to link to other liberal sources (and the same for conservative blogs/sources). Furthermore, other studies have shown that being around others who share your views will reinforce and strengthen those views. Putting this all together, Carr concludes that the web is a force for intensifying and polarizing people - especially as search engines like Google show you results that are correlated with your previous searches and choices. While the designers and creators of the web envisioned that it would foster greater sharing and community, Carr argues that belief may be naive.
Another chapter that seems to be a somewhat personal reaction to the demise of the print media. While I don't have a good solution, I do have the sense that high quality content will attract readers who may be willing to pay (a la the Wall St. Journal) and will attract advertisers. Of course, the content is online - newspaper companies will eventually stop printing paper versions - but that doesn't mean that high quality content won't exist online. Now as far as Google's personalized searches leading to increased polarization, I just don't buy it. I think that the search personalization presents me with software results rather than car parts when I search for "Struts" - and I think that is good.

Chapter 9: Fighting the Net - Carr's purpose in this chapter seems to be pointing out threats of/to the internet. The internet links and facilitates world wide terrorism (as do, say, cell phones). And the net generates its own types of threats - spam, bots, and zombie machines (which are running software you don't know about). Furthermore, while the internet is well designed to be incredibly resilient, the energy sources to run it are becoming an issue, not to mention disruptions due to natural disasters. Given how pervasive and critical the net is, the risks are far higher than they ever have been. While governments seem to be recognizing these threats, Carr claims they are missing the political issues. In particular, governments have different views on how the net should be handled (and American domination of existing rules and standards shouldn't be assumed to continue). Will the web remain basically open (US style) or centrally controlled (Chinese style). As utility computing moves more and more content outside the US, a countries "net politics" matter. Carr leaves us to ponder whether "Cold War 2.0" (over technology standards and issues) is somewhere in our future.
Cold War 2.0?? - No comment.

Chapter 10: A Spider's Web - We might think that we can travel through the web without leaving a trace, but Carr shows how clever and hardworking people are able to combine seeming anonymous pieces of data into something that can identify you. Scott McNealy, when he was CEO of Sun, famously remarked "You have zero privacy. Get over it." While this nicely sums up my attitude, Carr is less sanguine. He takes us back to the industrial revolution which sped up the processing of matter and energy - but the processing of information is only now becoming harnessed. Computers, to Carr, are fundamentally about controlling information and that would be especially true for the "world wide computer". While the introduction of the PC provided users with a sense of empowerment, Carr argues that the client/server architectures once again limited the user's freedom. And the emergence of the web is the next threat to control - but Carr believes that it will be temporary. Not only will the web (and constant connectedness) force employees to be "on call" 24/7, but it will also turn us into optimized consumers with targeted, highly effective ads - some based on neuromarketing techniques. Carr's perspective comes from the observations that the power created by new technologies always ends up in the hands of governments and business - in the end, we are flies caught in the spider web of the military-industrial complex. Worse, we are happy there because we have sold our souls for the convenience that the web brings - and we don't even realize what has happened.
I can't imagine anyone spending any amount of time/effort to uniquely identify me or my browsing habits - but that might be just because I am not paranoid. My experience of working on the web (I work from home) and shopping on the web (which I have been doing far before it was popular) is all positive. The 24/7 work world is also the world where I can pick up the kids from school and where I can purchase birthday presents from my computer. I have read estimates that people spend 2 hours of their work day online (not working) - but Carr doesn't mention this part of the story. And claiming that neuromarketing techniques (based on fMRI imaging) have any reality is disingenuous at best. This chapter struck me as the most one-sided in the book.

Chapter 11: iGod - Google founders (and other tech wizards) have stated that they would love to see a merging of human mind and computer - whether it is computers that acquire mind-like capabilities as an emergent property or whether it is humans with computer assisted minds. Carr states that this idea "to most of us ... is troubling". Carr believes that the time has already begun when humans are working for the computer and humans are becoming subservient. One example of this is Google's Page Rank. Whenever you add a link to some other site to your own web site, you are telling Google that that page is important (and thus its Page Rank goes up and it shows up higher in the search results). Humans are doing Google's work for it - without pay. Although Carr doesn't believe that humans will become "computer enhanced" he sees (in Google Search) the beginnings of the emergence of Computer Intelligence. The World Wide Computer now follows our instructions, but eventually it will write its own. Further, Carr sees that humans are already beginning to outsource some of their brain functions to the web - why remember what the computer will store? Why think when I can Google? In fact, he sees people as clicking more (and giving more information and therefore power to the world wide computer) and thinking less deeply. Eventually this culminates in humans being just a part of the great computer.
It is not clear to me why Carr is so anti-progress. Perhaps it is the difference between a literature major and a computer science major. I personally look forward to seeing my kids computer augmented. I think it is the human/machine hybrids that will lead Earth into the 22nd century - eclipsing both the "dumb computers" and "homo sapiens". Perhaps it is my rose colored glasses? :)