Science And Poetry Both Depend On Metaphors
Science's signature moves share something with good poetry. Good metaphor-making can make geniuses of both kinds. But bad metaphors can mislead whole fields.
12 August, 2016
<p dir="ltr"><span>1. Science, like poetry, depends on </span><a href="http://t.co/ZYUG9C0OAx" target="_blank"><span>metaphors</span></a><span>*</span><span>. They can hide, sometimes causing mischief.</span></p> <p dir="ltr"><span>2. Science’s signature moves deploy two or more metaphors. Pythagoras’s “All things are </span><a href="http://io9.gizmodo.com/what-did-pythagoras-mean-by-all-things-are-number-1717748417" target="_blank"><span>number</span></a><span>.” Plus at least one other, framing what the numbers mean (often tacitly, through tools or models). </span></p> <p dir="ltr"><span>3. Scientists should “think like poets and work like </span><a href="http://blogs.scientificamerican.com/guest-blog/2013/04/09/sciences-cult-of-calculation/" target="_blank"><span>accountants</span></a><span>,” E. O. Wilson advises. Useful number crunching builds on the poet's rarer skill of making good metaphors. </span></p> <p dir="ltr"><span>4. Good metaphor-making can make geniuses: Energy conservation is like balancing </span><a href="https://www.google.com/search?tbm=bks&hl=en&q=more+heat+than+light+Joule+wanted+to+cost+out+the+amount+of+heat+held+within+vats+of+liquid%2C+in+order+to+compare+it+to+the+cost+of+the+power+required+to+control+it.+With+the+methodical+attitudes+and+predispositions+of+the+accountant" target="_blank"><span>account-books</span></a><span> (Joule). Evolution’s “struggle for existence” is like humanity’s economic </span><a href="http://pllqt.it/PQxkHo" target="_blank"><span>struggles</span></a><span> (Darwin).</span></p> <p dir="ltr"><span>5. But bad metaphors can mislead entire fields: People ≠ biological </span><a href="http://bit.ly/1zmwOY4" target="_blank"><span>billiard balls</span></a><span>. Economies ≠ </span><a href="http://blogs.scientificamerican.com/guest-blog/2013/07/19/sciences-mobile-army-of-metaphors/" target="_blank"><span>gases</span></a><span>. (Alarmingly, “economic </span><a href="https://www.google.com/search?tbm=bks&hl=en&q=mccloskey+rhetoric+Becker+%22is+an+economic+poet%22" target="_blank"><span>poet</span></a><span>” Gary Beck metaphorized—>family = “little firm,” kids = “durable </span><a href="https://www.google.com/search?tbm=bks&hl=en&q=mccloskey+rhetoric+%22Becker+is+an+economic+poet%22#hl=en&q=McCloskey+rhetoric+Becker%27s+metaphors%2C+from+criminals+as+small+businessmen+to+the+family+as+a+little+firm.+Becker+is+an+economic+poet%2C+which+is+what+we+expect+from+our+theorists.&tbm=bks" target="_blank"><span>goods</span></a><span>,” heroin = </span><a href="http://blogs.scientificamerican.com/mind-guest-blog/2013/05/17/what-rational-really-means/" target="_blank"><span>bowling</span></a><span>.)</span></p> <p dir="ltr"><span>6. Let the “data do the talking” (preach Freakonomics </span><a href="http://pllqt.it/UuCFax" target="_blank"><span>folk</span></a><span>)? Alfred Marshall noted that can be “</span><a href="https://books.google.com/books?id=YW6m0tzFhHYC&pg=PA197&lpg=PA197&dq=reckless%E2%80%A6of+all+theorists%E2%80%A6let%E2%80%A6+figures+speak+for+themselves&source=bl&ots=c5lHLGNDx6&sig=POA0W6r3wFcaPBLUcHwuitQm4Bo&hl=en&sa=X&ved=0ahUKEwikscWqmrfOAhUEIMAKHadAAxoQ6AEIIzAB#v=snippet&q=%22facts%20by%20themselves%20are%20silent%22%20%22The%20most%20reckless%20and%20treacherous%20of%20all%20theorists%22%20%22professes%20to%20let%22%20%22figures%20speak%20for%20themselves%22&f=false" target="_blank"><span>treacherous</span></a><span>.” </span></p> <p dir="ltr"><span>7. You can’t always count on Pythagoras’s number-world move. For example many concepts in biology, </span><a href="http://bit.ly/1QFmG24" target="_blank"><span>economics</span></a><span>, and social science (e.g., fitness, utility, </span><a href="http://bit.ly/1NeIHH9" target="_blank"><span>happiness</span></a><span>) don’t have the mathematical properties of mass or length. They don’t fit a </span><a href="http://bit.ly/2aMZDDW" target="_blank"><span>ratio scale</span></a><span> and aren’t as measurable (=weakens the utility of math).</span></p> <p dir="ltr"><span>8. Many are confused about how quantitative and </span><a href="http://time.com/12551/nate-silver-fivethirtyeight-hiring/" target="_blank"><span>qualitative</span></a><span> relate; e.g., Nate Silver says that those not "quantitatively </span><a href="http://pllqt.it/nqI2Zg" target="_blank"><span>inclined</span></a><span>” risk creating "a lot of </span><a href="http://pllqt.it/2aE3iA" target="_blank"><span>bullshit</span></a><span>." </span></p> <p dir="ltr"><span>9. But fruitful quantification requires sound qualitative distinctions, otherwise it risks increasing bullshit—>e.g., the average human has ~1 ovary + ~1 testicle. Mixed-types math can be fruitless (≠ apples-to-apples comparison).</span></p> <p dir="ltr"><span>10. Statistical methods are especially slippery number-world tools. They require that underlying phenomena have sufficiently stable representative patterns—valid for physical traits like height variation, but often not for behaviors (different kinds of variability).</span></p> <p dir="ltr"><span>11. Stats harbor new versions of old logic woes, like the fallacy of composition—projecting properties of parts </span><a href="https://en.wikipedia.org/wiki/Fallacy_of_composition" target="_blank"><span>onto wholes</span></a><span> (e.g., apples are made of atoms, all atoms are invisible, therefore apples are invisible). And its opposite the fallacy of division (ascribes properties of wholes onto parts). </span></p> <p dir="ltr"><span>12. Consider data on shootings by police. Sendhil Mullainathan blunders in claiming that police racial bias has “little </span><a href="http://pllqt.it/GDxZkt" target="_blank"><span>effect</span></a><span>.” That’s a fallacy of division, assuming national data represent </span><a href="http://bigthink.com/errors-we-live-by/is-all-the-truth-we-need-in-the-data" target="_blank"><span>localities</span></a><span> well. Conversely, Rajiv Sethi notes a statistical “fallacy of </span><a href="http://rajivsethi.blogspot.com/2016/07/a-fallacy-of-composition.html" target="_self"><span>composition</span></a><span>,” can stats from one city be of any use in any differently composed city?</span></p> <p dir="ltr"><span>13. Top researchers often mishandle stats, e.g. </span><a href="http://andrewgelman.com/2016/07/26/29552/" target="_blank"><span>p-value</span></a><span> cherry-picking (in medicine, economics, psychology), multiple </span><a href="http://bit.ly/1Sy6sd7" target="_blank"><span>regression</span></a><span> (social </span><a href="http://bit.ly/1Sy6sd7" target="_blank"><span>sciences</span></a><span>, randomized </span><a href="http://pllqt.it/F0qBf1" target="_blank"><span>trials</span></a><span>). And standard stats moves can’t always help. Randomization still drops the ball on average testicle counts, and more data doesn’t automatically overcome lumpiness (Mullainathan’s misstep).</span></p> <p dir="ltr"><span>14. Three distinct pattern types exist with intrinsically increasing levels of variability: see Newton vs. Darwin vs. </span><a href="http://pllqt.it/GtcPMe" target="_blank"><span>Berlin</span></a><span> patterns. And tools like statistics and algebraic equations yield more in physics than in social sciences.</span></p> <p dir="ltr"><span>15. For example, Diane Coyle calls the seemingly objective GDP a </span><a href="http://pllqt.it/gOkrz7" target="_blank"><span>tarnished</span></a><span> measure. It’s a badly built number, it doesn’t distinguish “</span><a href="http://pllqt.it/DJDpdV" target="_blank"><span>bads</span></a><span>” from genuine goods, and it omits all that isn’t sold (marry your housekeeper—>GDP </span><a href="https://books.google.com/books?id=WnU9DAAAQBAJ&printsec=frontcover&dq=who+cooked+adam+smith%27s+dinner&hl=en&sa=X&ved=0ahUKEwjb2PLA7arOAhVCGx4KHe0ZDXkQ6wEIHzAA#v=snippet&q=%22if%20a%20man%20marries%20his%20housekeeper%2C%20the%20GDP%20of%20the%20country%20declines%22&f=false" target="_blank"><span>declines</span></a><span>).</span></p> <p dir="ltr"><span>16. The cult of calculation and data is seductive. And I’m no </span><a href="http://blogs.scientificamerican.com/moral-universe/2014/03/25/quantiphobia-and-turning-morals-into-facts/" target="_blank"><span>quantiphobe</span></a><span>. But number crunching has no monopoly on precision or </span><a href="http://blogs.scientificamerican.com/guest-blog/2013/08/30/maxims-are-fitter-than-maximization/" target="_blank"><span>truth</span></a><span>. Words, metaphors, non-numerical logic, images, and patterns can be exact and can exceed what numbers can do. </span></p> <p dir="ltr"><span>17. A desire to jump to “the numbers” isn’t always wise. We often shouldn’t ignore unquantifiable factors, or the metaphoric or qualitative weaknesses hidden in the number-world mindset.</span></p> <p dir="ltr"><span>*—Deep conceptual metaphors structure most of our thinking (George </span><a href="https://en.wikipedia.org/wiki/George_Lakoff" target="_blank"><span>Lakoff</span></a><span>). </span></p> <p><span> </span></p> <p><span>Illustration by Julia Suits, author of </span><em>The Extraordinary Catalog of Peculiar Inventions</em><span>, and </span><em>The New Yorker</em><span> cartoonist.</span></p>
Keep reading
Show less
