The Inevitable

by Kevin Kelly

  What to call this very large masterpiece? Is it more alive than machine? At its core 7 billion humans, soon to be 9 billion, are quickly cloaking themselves with an always-on layer of connectivity that comes close to directly linking their brains to each other. A hundred years ago H. G. Wells imagined this large thing as the world brain. Teilhard de Chardin named it the noosphere, the sphere of thought. Some call it a global mind, others liken it to a global superorganism since it includes billions of manufactured silicon neurons. For simple convenience and to keep it short, I’m calling this planetary layer the holos. By holos I include the collective intelligence of all humans combined with the collective behavior of all machines, plus the intelligence of nature, plus whatever behavior emerges from this whole. This whole equals holos.

  The scale of what we are becoming is simply hard to absorb. It is the largest thing we have made. Let’s take just the hardware, for example. Today there are 4 billion mobile phones and 2 billion computers linked together into a seamless cortex around the globe. Add to them all the billions of peripheral chips and affiliated devices from cameras to cars to satellites. Already in 2015 a grand total of 15 billion devices have been wired up into one large circuit. Each of these devices contains 1 billion to 4 billion transistors themselves, so in total the holos operates with a sextillion transistors (10 with 21 zeros). These transistors can be thought of as the neurons in a vast brain. The human brain has roughly 86 billion neurons, or a trillion times fewer than the holos. In terms of magnitude, the holos already significantly exceeds our brains in complexity. And our brains are not doubling in size every few years. The holos mind is.

  Today, the hardware of the holos acts like a very large virtual computer made up of as many computer chips as there are transistors in a computer. This virtual computer’s top-level functions operate at approximately the speed of an early PC. It processes 1 million emails each second, and 1 million messages per second, which essentially means the holos currently runs at 1 megahertz. Its total external storage is about 600 exabytes today. In any one second, 10 terabits course through its backbone nerves. It has a robust immune system, weeding spam from its trunk lines and rerouting around damage as a type of self-healing.

  And who will write the code that makes this global system useful and productive? We will. We think we are merely wasting time when we surf mindlessly or post an item for our friends, but each time we click a link we strengthen a node somewhere in the holos mind, thereby programming it by using it. Think of the 100 billion times per day humans click on a web page as a way of teaching the holos what we think is important. Each time we forge a link between words, we teach this contraption an idea.

  This is the new platform that our lives will run on. International in scope. Always on. At current rates of technological adoption I estimate that by the year 2025 every person alive—that is, 100 percent of the planet’s inhabitants—will have access to this platform via some almost-free device. Everyone will be on it. Or in it. Or, simply, everyone will be it.

  This big global system will not be utopia. Even three decades from now, regional fences will remain in this cloud. Parts will be firewalled, censored, privatized. Corporate monopolies will control aspects of the infrastructure, though these internet monopolies are fragile and ephemeral, subject to sudden displacement by competitors. Although minimal access will be universal, higher bandwidth will be uneven and clumped around urban areas. The rich will get the premium access. In short, the distribution of resources will resemble the rest of life. But this is critical and transformative, and even the least of us will be part of it.

  Right now, in this Beginning, this imperfect mesh spans 51 billion hectares, touches 15 billion machines, engages 4 billion human minds in real time, consumes 5 percent of the planet’s electricity, runs at inhuman speeds, tracks half our daytime hours, and is the conduit for the majority flow of our money. The level of organization is a step above the largest things we have made till now: cities. This jump in levels reminds some physicists of a phase transition, the discontinuous break between a molecule’s state—say, between ice and water, or water and steam. The difference in temperature or pressure separating two phases is almost trivial, but the fundamental reorganization across the threshold makes the material behave in a whole new manner. Water is definitely a different state than ice.

  The large-scale, ubiquitous interconnection of this new platform at first seems like just the natural extension of our traditional society. It seems to just add digital relationships to our existing face-to-face relationships. We add a few more friends. We expand our network of acquaintances. Broaden our sources of news. Digitize our movements. But, in fact, as all these qualities keep steadily increasing, just as temperature and pressure slowly creep higher, we pass an inflection point, a complexity threshold, where the change is discontinuous—a phase transition—and suddenly we are in a new state: a different world with new normals.

  We are in the Beginning of that process, right at the cusp of that discontinuity. In this new regime, old cultural forces, such as centralized authority and uniformity, diminish while new cultural forces, such as the ones I describe in this book—sharing, accessing, tracking—come to dominate our institutions and personal lives. As the new phase congeals, these forces will continue to intensify. Sharing, though excessive to some now, is just beginning. The switch from ownership to access has barely begun. Flows and streams are still trickles. While it seems as if we are tracked too much already, we’ll be tracking a thousand times as much in the coming decades. Each one of these functions will be accelerated by high-quality cognification, just now being born, making the smartest things we do today seem very dumb. None of this is final. These transitions are but the first step in a process, a process of becoming. It is a Beginning.

  * * *

  • • •

  Look at a satellite photograph of the earth at night to get a glimpse of this very large organism. Brilliant clusters of throbbing city lights trace out organic patterns on the dark land. The cities gradually dim at their edges to form thin long lighted highways connecting other distant city clusters. The routes of lights outward are dendritic, treelike patterns. The image is deeply familiar. The cities are ganglions of nerve cells; the lighted highways are the axons of nerves, reaching to a synaptic connection. Cities are the neurons of the holos. We live inside this thing.

  This embryonic very large thing has been running continuously for at least 30 years. I am aware of no other machine—of any type—that has run that long with zero downtime. While portions of it will probably spin down temporarily one day due to power outages or cascading infections, the entire thing is unlikely to go quiet in the coming decades. It has been and will likely remain the most reliable artifact we have.

  This picture of an emerging superorganism reminds some scientists of the concept of “the singularity.” A “singularity” is a term borrowed from physics to describe a frontier beyond which nothing can be known. There are two versions in pop culture: a hard singularity and a soft singularity. The hard version is a future brought about by the triumph of a superintelligence. When we create an AI that is capable of making an intelligence smarter than itself, it can in theory make generations of ever smarter AIs. In effect, AI would bootstrap itself in an infinite accelerating cascade so that each smarter generation is completed faster than the previous generation until AIs very suddenly get so smart that they solve all existing problems in godlike wisdom and leave us humans behind. It is called a singularity because it is beyond what we can perceive. Some call that our “last invention.” For various reasons, I think that scenario is unlikely.

  A soft singularity is more likely. In this future scenario AIs don’t get so smart that they enslave us (like evil versions of smart humans); rather AI and robots and filtering and tracking and all the technologies I outline in this book converge—humans plus machines—and together we move to a complex interdependence. At this level many phenomen
on occur at scales greater than our current lives, and greater than we can perceive—which is the mark of a singularity. It’s a new regime wherein our creations makes us better humans, but also one where we can’t live without what we’ve made. If we have been living in rigid ice, this is liquid—a new phase state.

  This phase change has already begun. We are marching inexorably toward firmly connecting all humans and all machines into a global matrix. This matrix is not an artifact, but a process. Our new supernetwork is a standing wave of change that steadily spills forward new arrangements of our needs and desires. The particular products, brands, and companies that will surround us in 30 years are entirely unpredictable. The specifics at that time hinge on the crosswinds of individual chance and fortune. But the overall direction of this large-scale vibrant process is clear and unmistakable. In the next 30 years the holos will continue to lean in the same direction it has for the last 30 years: toward increased flowing, sharing, tracking, accessing, interacting, screening, remixing, filtering, cognifying, questioning, and becoming. We stand at this moment at the Beginning.

  The Beginning, of course, is just beginning.

  ACKNOWLEDGMENTS

  I am indebted to Paul Slovak, my editor at Viking, who has long supported my efforts to make sense of technology, and to my agent John Brockman, who suggested this book. For editorial guidance on the first draft I relied on Jay Schaefer, master book coach based in San Francisco. Librarian Camille Hartsell did most of the factual research and provided the extensive endnotes. Claudia Lamar assisted in research, fact-checking, and formatting help. Two of my former colleagues at Wired, Russ Mitchell and Gary Wolf, waded through an early rough draft and made important suggestions that I incorporated. Over the span of years that I wrote this material I benefited from the precious time of many interviewees. Among them were John Battelle, Michael Naimark, Jaron Lanier, Gary Wolf, Rodney Brooks, Brewster Kahle, Alan Greene, Hal Varian, George Dyson, and Ethan Zuckerman. Thanks to the editors of Wired and The New York Times Magazine, who were instrumental in shaping initial versions of portions of this book.

  Most important, this book is dedicated to my family—Giamin, Kaileen, Ting, and Tywen—who keep me grounded and pointed forward. Thank you.

  NOTES

  1: BECOMING

  average lifespan of a phone app: Erick Schonfeld, “Pinch Media Data Shows the Average Shelf Life of an iPhone App Is Less Than 30 Days,” TechCrunch, February 19, 2009.

  sea pirates two centuries ago: Peter T. Leeson, The Invisible Hook: The Hidden Economics of Pirates (Princeton, NJ: Princeton University Press, 2011).

  graphic Netscape browser: Jim Clark and Owen Edwards, Netscape Time: The Making of the Billion-Dollar Start-Up That Took on Microsoft (New York: St. Martin’s, 1999).

  not designed for doing commerce: Philip Elmer-Dewitt, “Battle for the Soul of the Internet,” Time, July 25, 1994.

  “The Internet? Bah!”: Clifford Stoll, “Why the Web Won’t Be Nirvana,” Newsweek, February 27, 1995 (original title: “The Internet? Bah!”).

  “CB radio of the ’90s”: William Webb, “The Internet: CB Radio of the 90s?,” Editor & Publisher, July 8, 1995.

  Bush outlined the web’s core idea: Vannevar Bush, “As We May Think,” Atlantic, July 1945.

  Nelson, who envisioned his own scheme: Theodor H. Nelson, “Complex Information Processing: A File Structure for the Complex, the Changing and the Indeterminate,” in ACM ’65: Proceedings of the 1965 20th National Conference (New York: ACM, 1965), 84–100.

  “transclusion”: Theodor H. Nelson, Literary Machines (South Bend, IN: Mindful Press, 1980).

  “intertwingularity”: Theodor H. Nelson, Computer Lib: You Can and Must Understand Computers Now (South Bend, IN: Nelson, 1974).

  total number of web pages: “How Search Works,” Inside Search, Google, 2013, accessed April 26, 2015.

  90 billion searches a month: Steven Levy, “How Google Search Dealt with Mobile,” Medium, Backchannel, January 15, 2015.

  50 million blogs in the early 2000s: David Sifry, “State of the Blogosphere, August 2006,” Sifry’s Alerts, August 7, 2006.

  65,000 per day are posted: “YouTube Serves Up 100 Million Videos a Day Online,” Reuters, July 16, 2006.

  300 video hours every minute, in 2015: “Statistics,” YouTube, April 2015, https://goo.gl/RVb7oz.

  women online first outnumbered men: Deborah Fallows, “How Women and Men Use the Internet: Part 2—Demographics,” Pew Research Center, December 28, 2005.

  51 percent of netizens are female: Calculation based on “Internet User Demographics: Internet Users in 2014,” Pew Research Center, 2014; and “2013 Population Estimates,” U.S. Census Bureau, 2015.

  bone-creaking 44 years old: Weighted average of internet users in 2014 based on “Internet User Demographics,” Pew Research Center, 2014; and “2014 Population Estimates,” U.S. Census Bureau, 2014.

  mcdonalds.com was still unclaimed: Joshua Quittner, “Billions Registered,” Wired 2(10), October 1994.

  2: COGNIFYING

  several hundred “instances” of the AI: Personal visit to IBM Research, June 2014.

  “world’s best diagnostician”: Personal correspondence with Alan Greene.

  $18 billion in investments since 2009: Private analysis by Quid, Inc., 2014.

  in-house AI research teams: Reed Albergotti, “Zuckerberg, Musk Invest in Artificial-Intelligence Company,” Wall Street Journal, March 21, 2014.

  purchased AI companies since 2014: Derrick Harris, “Pinterest, Yahoo, Dropbox and the (Kind of) Quiet Content-as-Data Revolution,” Gigaom, January 6, 2014; Derrick Harris “Twitter Acquires Deep Learning Startup Madbits,” Gigaom, July 29, 2014; Ingrid Lunden, “Intel Has Acquired Natural Language Processing Startup Indisys, Price ‘North’ of $26M, to Build Its AI Muscle,” TechCrunch, September 13, 2013; and Cooper Smith, “Social Networks Are Investing Big in Artificial Intelligence,” Business Insider, March 17, 2014.

  expanding 70 percent a year: Private analysis by Quid, Inc., 2014.

  taught an AI to learn to play: Volodymyr Mnih, Koray Kavukcuoglu, David Silver, et al., “Human-Level Control Through Deep Reinforcement Learning,” Nature 518, no. 7540 (2015): 529–33.

  Betterment or Wealthfront: Rob Berger, “7 Robo Advisors That Make Investing Effortless,” Forbes, February 5, 2015.

  80 percent of its revenue: Rick Summer, “By Providing Products That Consumers Use Across the Internet, Google Can Dominate the Ad Market,” Morningstar, July 17, 2015.

  3 billion queries that Google conducts: Danny Sullivan, “Google Still Doing at Least 1 Trillion Searches Per Year,” Search Engine Land, January 16, 2015.

  Google CEO Sundar Pichai stated: James Niccolai, “Google Reports Strong Profit, Says It’s ‘Rethinking Everything’ Around Machine Learning,” ITworld, October 22, 2015.

  the AI winter: “AI Winter,” Wikipedia, accessed July 24, 2015.

  Billions of neurons in our brain: Frederico A. C. Azevedo, Ludmila R. B. Carvalho, Lea T. Grinberg, et al., “Equal Numbers of Neuronal and Non-Neuronal Cells Make the Human Brain an Isometrically Scaled-up Primate Brain,” Journal of Comparative Neurology 513, no. 5 (2009): 532–41.

  run neural networks in parallel: Rajat Raina, Anand Madhavan, and Andrew Y. Ng, “Large-Scale Deep Unsupervised Learning Using Graphics Processors,” Proceedings of the 26th Annual International Conference on Machine Learning, ICML ’09 (New York: ACM, 2009), 873–80.

  neural nets running on GPUs: Klint Finley, “Netflix Is Building an Artificial Brain Using Amazon’s Cloud,” Wired, February 13, 2014.

  dozen examples as a child before it can distinguish: Personal correspondence with Paul Quinn, Department of Psychological and Brain Sciences, University of Delaware, August 6, 2014.

  thousand games of chess: Personal correspondence with Daylen Yang (author of the Stockfish chess app), Stefan Meyer-Kahlen (deve
loped the multiple award-winning computer chess program Shredder), and Danny Kopec (American chess International Master and cocreator of one of the standard computer chess testing systems), September 2014.

  “akin to building a rocket ship”: Caleb Garling, “Andrew Ng: Why ‘Deep Learning’ Is a Mandate for Humans, Not Just Machines,” Wired, May 5, 2015.

  In 2006, Geoff Hinton: Kate Allen, “How a Toronto Professor’s Research Revolutionized Artificial Intelligence,” Toronto Star, April 17, 2015.

  he dubbed “deep learning”: Yann LeCun, Yoshua Bengio, and Geoffrey Hinton, “Deep Learning,” Nature 521, no. 7553 (2015): 436–44.

  the network effect: Carl Shapiro and Hal R. Varian, Information Rules: A Strategic Guide to the Network Economy (Boston: Harvard Business Review Press, 1998).

  famous man-versus-machine match: “Deep Blue,” IBM 100: Icons of Progress, March 7, 2012.

  rather than competes against them: Owen Williams, “Garry Kasparov—Biography,” KasparovAgent.com, 2010.

  freestyle chess matches: Arno Nickel, Freestyle Chess, 2010.

  centaurs won 53 games: Arno Nickel, “The Freestyle Battle 2014,” Infinity Chess, 2015.

  several different chess programs: Arno Nickel, “‘Intagrand’ Wins the Freestyle Battle 2014,” Infinity Chess, 2015.

  grand master rating of all time: “FIDE Chess Profile (Carlsen, Magnus),” World Chess Federation, 2015.

  AI that can view a photo portrait of any person: Personal interview at Facebook, September 2014.

  70 percent of American workers: U.S. Census Bureau, “Current Population Reports: Farm Population,” Persons in Farm Occupations: 1820 to 1987 (Washington, D.C.: U.S. Government Printing Office, 1988), 4.