9. Daniel C. Schlenoff, “The Future: A History of Prediction from the Archives of Scientific American,” Scientific American, Jan. 1, 2013, https://perma.cc/UHD4-EVPG.
10. Bernard Knox, Backing into the Future (New York: W. W. Norton, 1994), 11.
11. Thorleif Boman, Hebrew Thought Compared with Greek (New York: W. W. Norton, 1960), 149.
12. John S. Mbiti, African Religions and Philosophy (Oxford: Heinemann Educational, 1969), 17. James Gleick mentions other cultures that do not think about the future as lying in front of them in his book Time Travel (New York: Pantheon, 2016), 137–138.
13. Anthony Sudbery, “The Future’s Not Ours to See,” preprint, submitted May 2, 2016, 2, https://perma.cc/P3J6-CYRM.
14. See the superb, detailed account in Paul N. Edwards, A Vast Machine (Cambridge, MA: MIT Press, 2013), 85.
15. For a fascinating, and readable, exploration of the role of divinatory prognostication in the ancient Greek understanding of cognition, see Peter T. Struck, “A Cognitive History of Divination in Ancient Greece,” University of Pennsylvania Scholarly Commons, Jan. 2016, https://perma.cc/FG36-2YYP.
16. “Laplace’s Demon,” Information Philosopher, accessed Aug. 6, 2018, https://perma.cc/S89N-P4BB. On Laplace, see Martin S. Staum, review of Pierre Simon Laplace, 1749–1827: A Determined Scientist, by Roger Hahn, American Historical Review 111, no. 4 (Oct. 1, 2006): 1254, https://perma.cc/2RYA-9AFP. Pierre-Simone Laplace, A Philosophical Essay on Probabilities, trans. Frederick Wilson Truscott and Frederick Lincoln Emory (New York: John Wiley and Sons, 1902), https://perma.cc/Z6MX-T5J5.
17. Miles Mathis, “On Laplace and the 3-Body Problem,” Miles Mathis’s website, Aug. 6, 2009, https://perma.cc/F32C-9CD8.
18. See Herb Gruning, “Divine Elbow Room,” in Polyphonic Thinking and the Divine, ed. Jim Kanaris (Amsterdam: Rodopi, 2013), 43. Also, the winter 2015 issue of the New Atlantis has excellent articles on Newton’s religiosity: New Atlantis 44 (Winter 2015), https://perma.cc/544V-UN3F. Of particular use here is Stephen D. Snobelen’s “Cosmos and Apocalypse,” New Atlantis 44 (Winter 2015): 76–94, https://perma.cc/X3UC-CJBK; and a 1967 paper that he cites: David Kubrin, “Newton and the Cyclical Cosmos: Providence and the Mechanical Philosophy,” Journal of the History of Ideas 28, no. 3 (July–Sept. 1967): 325–346. Snobelen writes, “Newton’s so-called clockwork universe is hardly timeless, regular, and machine-like.… [I]nstead, it acts more like an organism that is subject to ongoing growth, decay, and renewal.”
19. Laplace, A Philosophical Essay on Probabilities, 3.
20. Jamie L. Vernon, “On the Shoulders of Giants,” American Scientist, July–Aug. 2017, 194, https://perma.cc/DE9Q-PPYJ.
21. Isaac Newton, Newton’s Principia, trans. Andrew Motte (New York: Daniel Adee, 1846), lxvii, https://archive.org/details/100878576/page/n7.
22. From Julie Wakefield’s fascinating biography, Halley’s Quest (Washington, DC: Joseph Henry, 2005), 76.
23. All of this account comes from David Alan Grier’s When Computers Were People (Princeton, NJ: Princeton University Press, 2005), 11–25.
24. Incremental steps: Monique Gros Lalande, “Lepaute, Nicole-Reine,” in Biographical Encyclopedia of Astronomers, ed. Thomas Hockey et al. (New York: Springer Science and Business Media, 2007), 690–691. Ardor: Grier, When Computers Were People, 22. Removed acknowledgment: Lalande, “Lepaute, Nicole-Reine,” 690–691. Unacknowledged for later work: Catherine M. C. Haines, International Women in Science: A Biographical Dictionary to 1950 (Santa Barbara: ABC-CLIO, 2001), 174. Canceled errors: Grier, When Computers Were People, 23.
25. “[N]ot one word of proof or demonstration do you offer. All is probability with you, and yet surely you and Theodorus had better reflect whether you are disposed to admit of probability and figures of speech in matters of such importance. He or any other mathematician who argued from probabilities and likelihoods in geometry, would not be worth an ace.” The Thaetetus, trans. Benjamin Jowett, https://perma.cc/U55R-3ZUE.
26. Leonard Mlodinow, The Drunkard’s Walk: How Randomness Rules Our Lives (New York: Pantheon Books, 2008), 122.
27. I talk about the history of “information overload” in Too Big to Know (New York: Times Books, 2011), 5–6, with more detail in the endnotes.
28. Siobhan Roberts, “John Horton Conway: The World’s Most Charismatic Mathematician,” Guardian, July 23, 2015, https://perma.cc/64WC-9JRG.
29. You can read the rules and try it out at https://playgameoflife.com/ (https://perma.cc/WXB4-KREA).
30. Martin Gardner, “The Fantastic Combinations of John Conway’s New Solitaire Game ‘Life,’ ” Scientific American, Oct. 1970, 120–123, https://perma.cc/ER58-TGV3. While encouraging readers to play the game manually, Gardner notes that some colleagues of Conway’s had programmed a PDP-7 “minicomputer”—“mini” because you could have fit one into a large laundry room—to run the game.
31. Alexy Nigin, “New Spaceship Speed in Conway’s Game of Life,” Nigin’s Blog, Mar. 7, 2016, https://perma.cc/WY2D-5KRF.
32. Daniel C. Dennett, Darwin’s Dangerous Idea: Evolution and the Meaning of Life (New York: Simon and Schuster, 2014), 166ff.
33. Raymond Kurzweil, The Singularity Is Near (New York: Penguin Books, 2006).
34. Stephen Wolfram, A New Kind of Science (Champaign, IL: Wolfram Media, 2002).
Chapter Two
1. Alexandra Gibbs, “Chick Sexer: The $60K a Year Job Nobody Wants,” NBC News, Mar. 4, 2015, https://perma.cc/7FYE-6KZR.
2. Richard Horsey, “The Art of Chicken Sexing,” UCL Working Papers in Linguistics 14 (2002): 107–117, https://perma.cc/98MF-4YZM.
3. “What Does a Chicken Sexer Do?,” Sokanu, accessed Sept. 30, 2018, https://perma.cc/XYP5-FZVN.
4. This idea is usually traced back to Plato’s Theaetetus, although as Richard Polt reminds me in an email, Socrates’s interlocutors fail to come up with an explanation of what constitutes justification, and the dialogue ends without a resolution to the question of what is knowledge.
5. For example, see Robert B. Brandom, “Insights and Blindspots of Reliabilism,” The Monist 1, no. 3 (June 1998): 371–393. Brandom offers a sophisticated analysis of this question.
6. You could argue that the machine learning system’s working model is also a conceptual model, but in fact there is no intelligence that has the concept, so I prefer not to.
7. See, for example, the excellent introduction to machine learning by Adam Geitgey, “Machine Learning Is Fun!,” Medium, May 5, 2014, https://perma.cc/FQ8X-K2KQ.
8. Richard Dunley, “Machines Reading the Archive: Handwritten Text Recognition Software,” National Archives Blog, Mar. 19, 2018, https://perma.cc/NQ9R-NCZR.
9. Sidney Kennedy, “How AI Is Helping to Predict and Prevent Suicide,” The Conversation, Mar. 27, 2018, https://perma.cc/D8K4-ERZV. For an excellent and highly accessible discussion of these issues, see Cathy O’Neil’s Weapons of Math Destruction (New York: Crown, 2016) and the upcoming work Smart Enough Cities, by Ben Green (Cambridge, MA: MIT Press, 2019). (Disclosure: I edit the book series publishing Green’s book.)
10. Nate Silver, The Signal and the Noise (New York: Penguin Books, 2012).
11. Paul N. Edwards, A Vast Machine (Cambridge, MA: MIT Press, 2013), 85.
12. Ibid., 94–96.
13. Ibid., 123.
14. Silver, Signal and the Noise, 386.
15. Jonathan M. Gitlin, “Krakatoa’s Chilling Effect,” Ars Technica, Jan. 9, 2006, https://perma.cc/XET2-GY9K.
16. Richard Mattessich, a professor at University of California, Berkeley, and two grad students created the Budget Computer Program for mainframes in the early 1960s. See Paul Young, “VisiCalc and the Growth of Spreadsheets,” RBV Web Solutions, last updated May 7, 2000, https://perma.cc/2HRF-73RZ.
17. Steven Levy, “A Spreadsheet Way of Knowledge,” Wired, Oct. 24, 2014, https://perma.cc/YQ8H-CCRK.
18. Dan Bricklin, “The Idea,” Dan Bricklin’s Web Site, accessed Sept. 30, 2018, https://perma.cc/5
5SU-NBSM.
19. Dan Bricklin, “Patenting VisiCalc,” Dan Bricklin’s Web Site, accessed Sept. 30, 2018, https://perma.cc/3UF9-UPAW.
20. Levy, “Spreadsheet Way of Knowledge.”
21. George E. P. Box, Robustness in the Strategy of Scientific Model Building (Madison: Wisconsin University Mathematics Research Center, 1979), 2, https://perma.cc/7E54-AGVG.
22. “Armillary Sphere,” Museo Galileo, accessed Sept. 30, 2018, https://perma.cc/9Y3G-V42C.
23. Oxford Museum of the History of Science, “Armillary Sphere,” Epact: Scientific Instruments of Medieval and Renaissance Europe, accessed Sept. 30, 2018, https://perma.cc/6ZPN-LPK6.
24. Martin Kemp, “Moving in Elevated Circles,” Nature, July 2010, 33, https://perma.cc/GW3N-VY6Y.
25. My friend John Frank, a computer scientist, read an early draft of this and commented in an email (Oct. 10, 2018),
As a beautiful example, perhaps the canonical example, of a model foreshadowing its successor in a deeply technical manner, these many circles were approximating the Fourier components of the ellipses of the real orbits. A change of coordinates allows these Fourier components to become their simpler equivalent.
Machine learning models provide a principled procedure for discovering those better coordinates that have not yet been articulated in an explanatory form.
26. C. H. Claudy, “A Great Brass Brain,” Scientific American, Mar. 7, 2014, 197–198, https://perma.cc/7CQL-G5XG.
27. Ibid., 197.
28. Jonathan White, Tides: The Science and Spirit of the Ocean (San Antonio, TX: Trinity University Press, 2017): “hundreds of these eccentricities,” 5–6; never saw an ocean: 115–116, 120.
29. Ibid., 152.
30. “History of Tidal Analysis and Prediction,” NOAA Tides & Currents, last revised Aug. 8, 2018, https://perma.cc/XMZ2-6KD2.
31. Claudy, “Great Brass Brain,” 198.
32. White, Tides, 202.
33. Dylan, “The Mississippi River Basin Model,” Atlas Obscura, accessed Sept. 30, 2018, https://perma.cc/WL52-R2MG. The source of the $65 million figure seems to be J. E. Foster, History and Description of the Mississippi Basin Model, Mississippi Basin Model Report 1-6 (Vicksburg, MS: US Army Engineer Waterways Experiment Station, 1971), 2; this is according to Kristi Dykema Cheramie, “The Scale of Nature: Modeling the Mississippi River,” Places Journal, Mar. 2011, https://perma.cc/DR5X-3Y34. If we assume, based on nothing, that the $65 million figure had already been translated into 1971 dollars, the current equivalent would be $386 million.
34. “America’s Last Top Model,” 99% Invisible, July 19, 2016, podcast, 21:25, http://99percentinvisible.org/episode/americas-last-top-model/.
35. A. G. Gleeman, “The Phillips Curve: A Rushed Job?,” Journal of Economic Perspectives 25, no. 1 (Winter 2011): 223–238, 225.
36. “A Pioneering Economic Computer,” Reserve Bank Museum, https://perma.cc/W7MH-5HLK.
37. Larry Elliot, “The Computer Model that Once Explained the British Economy,” The Guardian, May 8, 2008, https://perma.cc/T88G-RQDB.
38. “America’s Last Top Model.”
39. See the “Scales” subhead on this page from the US Army Corps of Engineers: “The Technical Side of the Bay Model,” US Army Corps of Engineers, accessed Sept. 30, 2018, https://perma.cc/WXZ3-3DSQ.
40. The history of this disruption goes back at least to Peter Brown’s work in the early 1990s at IBM, where he pioneered translating texts through the statistical analysis of existing translations, without providing the computer with linguistic models of the grammar, syntax, or semantics of either language.
41. Dave Gershgorn, “Google Is Using 46 Billion Data Points to Predict the Medical Outcomes of Hospital Patients,” Quartz, Jan. 27, 2018, https://perma.cc/NHS2-HU2G.
42. Riccardo Miotto et al., “Deep Patient: An Unsupervised Representation to Predict the Future of Patients from the Electronic Health Records,” Scientific Reports 6 (2016): article 26094, https://perma.cc/L8YL-6Q69.
43. Julian Mitchell, “This A.I. Search Engine Delivers Tailored Data to Companies in Real-Time,” Forbes, May 11, 2014, https://perma.cc/L3JP-8D38.
44. Jeff Curie and Greg Bolcer, email exchange with the author, July 17, 2017. See also Mitchell, “This A.I. Search Engine Delivers Tailored Data to Companies in Real-Time.”
45. As quoted in David Weinberger, “Our Machines Now Have Knowledge We’ll Never Understand,” Wired, Apr. 18, 2017, https://perma.cc/SP4T-AKZP.
46. There is a great deal of work and discussion about whether all machine learning implementations will always resist explanation. See Cynthia Rudin’s ten-minute video from a webinar on Oct. 2, 2018: “Please Stop Doing Explainable ML,” webcast, Oct. 2, 2018, https://perma.cc/CWG3-4HUK. (It’s the second talk in the video.) Here is a suggestion for how to make machine learning’s “decisions” understandable without having to understand exactly how those decisions were arrived at: Finale Doshi-Velez and Mason Kortz, “Accountability of AI under the Law: The Role of Explanation” (paper presented at the Privacy Law Scholars Conference, George Washington University, Washington, DC, 2018), https://perma.cc/L275-MH4N. Also see Finale Doshi-Velez and Been Kim, “Towards a Rigorous Science of Interpretable Machine Learning,” preprint, submitted Feb. 28. 2017, https://perma.cc/2PSA-NZR4.
47. David Sutcliffe, “Could Counterfactuals Explain Algorithmic Decisions without Opening the Black Box?,” Oxford Internet Institute, Jan. 15, 2018, https://perma.cc/HT5X-CV4L.
48. Deep learning systems are not the first computer programs to create their own models. John Frank, in his comments on a draft of this chapter, observed in an email dated Oct. 10, 2018, “This idea started with Leslie Valiant’s ‘Theory of the Learnable’ which contributed to his winning the Turing Prize [https://perma.cc/H9DR-TP3H]. It was subsequently realized in a variety of fields by different domain-specific experts, such as Peter Brown et al. in 1991 for machine translation.”
49. Jessica Birkett, “What the Dog-Fish and Camel-Bird Can Tell Us about How Our Brains Work,” The Conversation, July 6, 2015, https://perma.cc/W28T-EERD.
50. Julia Angwin et al., “Machine Bias,” ProPublica, May 23, 2016, https://perma.cc/249Q-7XCQ.
51. Dave Gershgorn, “By Sparring with AlphaGo, Researchers Are Learning How an Algorithm Thinks,” Quartz, Feb. 16, 2017, https://perma.cc/V9YY-RTWQ.
52. Mix, “Google Is Teaming Up Its AlphaGo AI with Humans So They Can Learn from It,” TNW, Apr. 10, 2017, https://perma.cc/5MMZ-5FXL.
53. Rory Cellan-Jones, “Google DeepMind: AI Becomes More Alien,” BBC, Oct. 18, 2017, https://perma.cc/EEC4-N859.
54. Dawn Chan, “The AI That Has Nothing to Learn from Humans,” Atlantic, Oct. 20, 2017, https://perma.cc/4EQ8-Z73X.
55. Kuhn actually talks about paradigm in many different ways. See my essay on the fiftieth anniversary of The Structure of Scientific Revolutions: “Shift Happens,” Chronicle of Higher Education: Chronicle Review, Apr. 22, 2012, https://perma.cc/8XPS-84WN.
56. Nicholas Faith, Black Box (Minneapolis: Motorbooks International, 1997), 100–105.
57. Ibid., 103.
58. Jecelyn Yeen, “AI Translate: Bias? Sexist? Or This Is the Way It Should Be?,” Hackernoon, Oct. 6, 2017, https://perma.cc/2A7N-KKSX.
59. Momin M. Malik and Hemank Lamba, “When ‘False’ Models Predict Better Than ‘True’ Ones: Paradoxes of the Bias-Variance Tradeoff” (unpublished manuscript, version 1.6, Dec. 2017), https://www.mominmalik.com/false_models_in_progress.pdf.
60. Parts of this coda are adapted from an article published on the Harvard Berkman Klein Center’s page at Medium under a Creative Commons BY license: “Optimization over Explanation: Maximizing the Benefits of Machine Learning without Sacrificing Its Intelligence,” Medium, Jan. 28, 2018, https://perma.cc/H538-3Q2G. A version of that article by agreement was simultaneously posted by Wired under the title “Don’t Make AI Artificially Stupid in the Name of Transparency,” https://perma.cc/X9N4-8RBT.
61. The McKinsey data comes from Mic
hele Bertoncello and Dominik Wee, “Ten Ways Autonomous Driving Could Redefine the Automotive World,” McKinsey&Company, June 2015, https://perma.cc/G8CT-JTTD. The Tesla information is from the Tesla Team, “An Update on Last Week’s Accident,” Tesla Blog, Mar. 20, 2018, https://perma.cc/Q8CH-7HLP. It’s commonly said that the National Highway Traffic Safety Administration says that 93 percent of traffic accidents are caused by human error, but it’s hard to track down the source of that figure. See Bryant Walker Smith, “Human Error as a Cause of Vehicle Crashes,” Stanford Center for Internet and Society, Dec. 18, 2013, https://perma.cc/9DR3-6EVC.
62. Brett Frischmann and Evan Selinger, Re-engineering Humanity (Cambridge: Cambridge University Press, 2018), 137.
63. Sam Levin and Julia Carrie Wong, “Self-driving Uber Kills Arizona Woman in First Fatal Crash Involving Pedestrian,” The Guardian, Mar. 19, 2018, https://perma.cc/VB4P-27HG.
64. Devin Coldewey, “Uber in Fatal Crash Detected Pedestrian but Had Emergency Braking Disabled,” TechCrunch, Apr. 24, 2018, https://perma.cc/W4L3-SVCM.
Chapter Three
1. The information about Ford’s design of the Model T comes from John Duncan, Any Colour—So Long as It’s Black (Titirangi, New Zealand: Exisle, 2008), Kindle edition.
2. Ibid., chapter 1.
3. Kate Wong, “Oldest Arrowheads Hint at How Modern Humans Overtook Neandertals,” Scientific American, Nov. 7, 2012, https://perma.cc/7H9E-V8G2. Stone spear tips believed to be five hundred thousand years old have been found in Spain. See Kate Wong, “Human Ancestors Made Deadly Stone-Tipped Spears 500,000 Years Ago,” Scientific American, Nov. 15, 2012, https://perma.cc/ET3R-3KRJ.
4. Dana Gunders, “Wasted: How America Is Losing up to 40 Percent of Its Food from Farm to Fork to Landfill,” NRDC Issue Paper 12-06-B, National Resources Defense Council, Aug. 2012, https://perma.cc/DF6M-FECX. Food packages have only had dates attached to them for about the last hundred years, a result of the distribution of food far removed from its sources. See Rosetta Newsome et al., “Applications and Perceptions of Date Labeling of Food,” Comprehensive Reviews in Food Science and Food Safety 13 (2014): 745–769, https://perma.cc/Q3JR-PPJE.
Everyday Chaos Page 22