In this week’s eSkeptic:
Last Week’s Mystery Photo is of Alfred Russel Wallace, the co-discoverer of natural selection, playing chess with his sister Frances, shortly after his return from his collecting expedition to the Amazon rain forest in 1852, six years before he discovered natural selection in his second big expedition to the Malay Archipelago (in 1858). Michael Shermer documents his discoveries and evolutionary theorizing in his full-length biography, In Darwin’s Shadow: The Life and Science of Alfred Russel Wallace, originally his Ph.D. thesis and published by Oxford University Press. It is sub-sub titled A Biographical Study on the Psychology of History, and presents a rather different perspective on how biography can be written (although my narrative is written in a fairly straight forward biographical style).
We will reveal the answer to this week’s Mystery Photo in next week’s eSkeptic.
A Blind Taste Test for Literature?
(a note from the Editor about this week’s feature article)
In this week’s eSkeptic, we are publishing an article entitled “Scientific Evaluation of Charles Dickens” by Mikhail Simkin, who received a Ph.D. in physics from Brown University and currently works as a research engineer at UCLA. Mikhail submitted this paper to me many months ago for publication consideration in Skeptic. I liked the article, edited it, and told him that we would publish it. I liked his approach of trying to quantify the assessment of literature blind — that is, without knowing the author. In this case, he compares paragraphs taken from the works of Charles Dickens to paragraphs taken from the works of Edward Bulwer-Lytton, who is apparently considered a very bad writer. I actually rather enjoyed the puncturing of a literary tradition of ranking writers (it initially reminded me of the skeptical debunking of postmodernism and deconstructionism of the 1990s), but then I don’t know anything about literature and have been frequently accused of practicing scientism — the excessive over-application of science to areas where it probably doesn’t belong. After my editors and several board of advisor members read the article, however, I was strongly urged not to publish this piece, not because it might be considered controversial, but because there is no controversy at all. As one explained it to me, a modern reader could no more be expected to tell the difference between a paragraph of Dickens from a paragraph of Bulwer-Lytton and assess the author’s overall value than a modern listener should be expected to hear the difference between a brief snippet of Mozart and a brief snippet of Salieri and assess the composer’s overall value (the film Amadeus notwithstanding). I asked a professor of literature who specializes in 19th century literature to read Mikhail’s article. He responded thusly:
Dickens is considered a powerful novelist of social protest (consider the implications of his novel, Hard Times), and a creator of brilliant characters. But his prose is fairly common to the period. That is, it is still suffering from the 18th century propensity to equate good prose with Latinate prose. Latinate prose is as simple as the use of multi-syllabic words in English, which are generally of Latinate origin. Using Latinate prose tends to make English more abstract and wordy. For instance: One could say in Latinate prose, “I resume my journey to my domicile,” or in the more Anglo-Saxon, “I’m going home.” (From Wikipedia: Latinate prose in English is the use of words derived from Latin rather than those originating in Old English, e.g. suspend rather than hang. A Latinate style may also be marked by prominent syntactic inversion, especially the delaying of the main verb: while the normal English word-order is subject-verb-object, Milton frequently uses the Latin order object-subject-verb in his poem Paradise Lost (1667), as in the line, “His far more pleasant garden God ordained.”
It would not surprise me that sampling Dickens’ prose as this fellow has done would produce no visible difference from other prose of the same period. Also, I don’t think anyone in my field would set out to prove or disprove Dickens was a genius because the term is recognized as subjective and therefore unprovable.
The study also assumes that if Dickens is a “genius,” every line of his prose should be “genius.” It’s a false analogy to presume that genius is like a pie; i.e., if a pie is a pumpkin pie, every bite will taste like pumpkin pie. Genius is nothing like pie. If you’ve ever read the first half of A Tale of Two Cities, which is confusing and desperately in need of revision, you’ll know better.
I’m afraid I think his premise is rather silly, and the study is irrelevant.
Interestingly, on the day I penned this editorial note, I listened to a lecture in a Teaching Company course on “Thinking Like an Economist” in which Professor Randall Bartlett of Smith College made the point that Dickens differed from Hemingway in one very basic economic sense: Dickens was paid by the word and serialized most of his works into multi-part stories that appeared in multiple magazines and therefore tended to drag out his prose, whereas Hemingway was paid by the story or manuscript and was therefore motivated to get to the point and be more economical with his prose.
I ultimately decided not to publish this article in Skeptic magazine, and Mikhail was understandably upset about my decision. So I thought it might be instructive to publish it in eSkeptic with a feedback system in place whereby you, the reader, can judge for yourself whether or not you think that Mikhail Simkin’s scientific approach to literature in evaluating its quality and worth is a valid one. So, in addition to the usual comment forum at the end of (the online version) of eSkeptic, we have also set up a quick polling system whereby you can register your vote on whether or not you agree with Mikhail’s scientific approach to assessing the value of literature.
After reading this week’s feature article below, be sure and PARTICIPATE IN THE POLL, and feel free to share this week’s eSkeptic with your literary friends and colleagues using the share buttons below!
Michael Shermer, Editor in Chief
of Charles Dickens
by Mikhail Simkin
Abstract: I report the results of the test, where the takers had to tell the prose of Charles Dickens from that of Edward Bulwer-Lytton, who is considered by many to be the worst writer in history of letters. The average score is about 50%, which is on the level of random guessing. This suggests that the quality of Dickens’s prose is the same as of that of Bulwer-Lytton.
F. Scott Fitzgerald wrote, “Let me tell you about the very rich. They are different from you and me.” On this Ernest Hemingway commented, “Yes, they have more money.”
Are the very famous writers different from the obscure ones?
The question may seem shocking to some people, but recent scientific research makes it quite reasonable. The study of misprints in scientific citations had lead to the conclusion that about 80% of citations are copied from the lists of references used in other papers1. Thus, in a majority of cases, a citation is not a result of an independent evaluation of the qualities of the cited paper but merely an imitation of another citer’s behavior. This way a paper that already was cited is likely to be cited again, and after it is cited again it is even more likely to be cited in the future. Thus some papers can become much more cited than others even when identical in merit. Mathematical modeling of the process of citation copying demonstrated that major features of the citation distribution could be explained even under assumption that all papers are created equal2. One can suspect that, similar to highly cited scientists, highly popular writers can become such as a result of the ordinary law of chances. One way to check that is to see if people can appreciate the prose of a famous writer when his name is detached from it.
Edward Bulwer-Lytton is the worst writer in history of letters. An annual wretched writing contest3 was established in his honor. In contrast, Charles Dickens is one of the best writers ever. Can one tell the difference between their prose? To check this I wrote the “Great prose or not?” quiz4. It consists of a dozen of representative literary passages, written either by Bulwer-Lytton or by Dickens. The takers are to choose the author of each quote. They face a formidable task and are often surprised to learn the correct answers. “What mindless boob would write such tripe? Dickens, one would know now.” – wrote me one of respondents. The distribution of the scores received by over nine thousands quiz-takersA is shown in Figure 1. The average score is 5.78 or 48.2% correct.
There are two possible answers to each test question. If one is completely clueless and resorts to random guessing, he will on average get 50% of the questions right. With the average score of 48% our quiz-takers lost to a monkey. On average, a quote from Bulwer-Lytton was selected as Dickens (or great prose) by 52% of quiz-takers, while a quote from Dickens was selected as Dickens by only 48%. Does this mean that Bulwer-Lytton is a better writer than Dickens? Probably not. Table 1 shows for every quote the fraction of people who attributed it to Dickens. This fraction varies between the quotes with the lowest being 36% (No. 9) and the highest 74% (No. 12). This suggests that a different selection of quotes could lead to a different average score. For example, if we remove the most Dickensian Bulwer (No. 12) and the most Bulwerian Dickens (No. 10), and recalculate the scores based on 10 remaining questions, the average score becomes 51%.
|Question number||The real author, and the book the excerpt is taken from||Selected as Dickens||Selected as Bulwer-Lytton|
An interesting thing is that out of 9,461 people, 38 got every question wrong and 37 got everything right. The approximate equality of these numbers is consistent with random guessing, but their magnitude is not. It is more than fifteen times bigger than what random guessing would give. The explanation is that some of quiz takers can sniff stylistic similarities between certain literary passages and attribute them to the same writer. This would help them to get a higher score, if they can determine which writer is good and which is bad, otherwise they are equally likely to get a very high or a very low score.
The performance of our quiz-takers is bad. But could this be because they don’t know English? The feedback demonstrates, however, that even educated people can’t tell Dickens from Bulwer. One of quiz-takers wrote me “I got a 50%. My cat could do that well. The wine experts say a peek at the label is worth a thousand sips, and that seems to hold here. As a classicist I’m frequently called on to teach stuff I think is wretched, just because it’s old.” Some experts do not even dare to take the quiz. Prof. Scott Rice, the founder of the Bulwer-Lytton fiction contest, wrote to me “I haven’t really taken it yet myself. Perhaps I am afraid to.” We can address the education issue in a scientific way. Fortunately, the quizzing script records taker’s IP address. From it, one can infer where their computers were located. I selected a subset of scores, which were received by people coming from English-speaking (American, British, Australian, and New Zealandian) universitiesB. The histogram of the scores received by 602 such people is shown in Figure 2. The average score is 5.76 or 48.0% correct. The standard error of this average is 0.095 or 0.8%. Educated English-speaking folks lost to the general public, whose average score is 48.2%. The difference between the scores is, however, statistically insignificant, because it is less than the standard error.
|Elite School||number of respondents||minimum score||maximum score||average score|
|University of Cambridge||16||2||9||6.13|
|University of Oxford||10||3||10||6.3|
|University of Pennsylvania||7||2||11||7.71|
But, perhaps, just knowing English is not enough? May be the beauty of Dickens’ prose is so far beyond the apprehensions of the vulgar that only the most cultured people can appreciate it? To check this I selected a subset of scores, earned by people coming from elite universities (Ivy League and Oxbridge). Table 2 contains the statistics of scores received by 76 of the chosen. The average score is 6 or 50% correct. The elite won over crowd by the whole 2%. The difference between the elite and general scores is, however, statistically insignificant. Due to the small size of the elite sample the standard error of the average elite score is 2.6%.
I began this paper with the question: Are famous writers different from their obscure colleagues? The answer is: Yes, they have more readers.
Recently I reported similar results for the case of Modern Art 5.
- M.V. Simkin, V.P. Roychowdhury “Read before you cite!” Complex Systems. 14 (2003) 269–274. Also available at http://arxiv.org/abs/cond-mat/0212043
- M.V. Simkin, V.P. Roychowdhury “A mathematical theory of citing” Journal of the American Society for Information Science and Technology, 58 (2007) 1661–1673. Also available at http://arxiv.org/abs/physics/0504094
- Bulwer-Lytton fiction contest, http://www.bulwer-lytton.com/
- M.V. Simkin, “Great prose, or not?” http://reverent.org/bulwer-dickens.html (Published on September 16, 2004)
- M.V. Simkin, “My statistician could have painted that! A statistical inquiry into modern art” Significance, l 4 (2007) 93–96. Available online at: http://www.significancemagazine.org/details/magazine/868985/My-statistician-could-have-painted-that-A-statistical-inquiry-into-modern-art-.html. Also available at arXiv.org: http://arxiv.org/abs/physics/0703091
- When I looked at the quiz results I noticed hundreds of cases when two or more scores came from the same IP address within few minutes. In many of such cases the later score was 100%. This suggests that many people took several shots at the quiz. To eliminate this cheating I cleaned the data by selecting only the first score from each IP address. Afterwards I cleaned the data from the results, where one or more questions were skipped.
- I identified American, British, Australian, and New Zealandian universities by Internet domains: edu, ac.uk, edu.au, and ac.nz.