A Deep Thinking

Errors in Darwin's Dangerous Idea

Daniel C. Dennett

As of January 25, 2006, readers have identified the following errors in Darwin's Dangerous Idea. (I have considered other criticisms offered by readers, but decided that they were in error. Further criticisms are, of course, invited.):

P24. I describe Aristotle’s Prime Mover as a final cause, but this is incorrect; Aristotle’s Prime Mover is the efficient cause of all things. Pointed out by Chris Hammel, 1/23/06.

p57. "Simulated annealing" may be a misnomer, since the process I describe--and describe blacksmiths as employing--is signicantly different from standard annealing practices. source: John Verhoeven.

p97. My argument about how to identify Mitochondrial Eve (ME) is flawed. This has been pointed out by Ian Gillies, Bill Margolis, Chris Viger, and Gilbert Scott Markle. (See also, the useful discussion in Jack Cohen and Ian Stewart's new book Figments of Reality). There are two complications, which I will call multiple local MEs and grandmother stacks.

Multiple local MEs: There could in principle be long periods of "stasis" in which the sets of mothers of mothers of . . . . didn't shrink in size for hundreds or thousands of years on end. No set can be larger than its successor, but it needn't be any smaller. In such a circumstance, going backwards in time, the "remaining" lineages each funnel down to a "local ME" (with two daughters) and that local ME's mother, (maternal) grandmother, greatgrandmother and so forth, single-file lineages marching back into the past. Whenever two such local ME strands converge, they discover a "new," less local ME, discharging their two more local double-daughtered MEs. In principle such multiple local ME strands could go back to multiple independent origins of mitochondrial life (of eukaryotes)--but there is very good reason to suppose that global ME (the True Mitochondrial Eve) is a hominid of some ilk, if not a member in clear standing of H. sapiens. I was wrong to suggest (by saying the sets "must" contract") that such a twin-lineage (to take the simplest case) couldn't "go on forever"; in principle it could, but with negligible--but not Vanishing--probability.

A grandmother stack is a subset of women related by maternity within any of the sets A, B, C, of mothers of mothers . . . . Thus in set B, the mothers of people alive today, there is the following grandmother stack: Andrea (mother of my grandson), Susan (my wife), Ruth (my mother-in-law), and her mother, the late Sylvia. Set C will drop Andrea and add Sylvia's mother, and so forth. Subsequent sets will always have a four-member grandmother stack generated by today's contemporaneous generations in my family. There are probably some seven- or eight-deep grandmother stacks somewhere in the world today, but I would think that the biological limit is about nine (you can get a ten-deep stack by assuming a stack of 13-year-old mothers with a still living 104-year-old ancestor). When local MEs evaporate by coalescence in the manner described above, the grandmother stacks involved do not shrink in the sets in which they appear. But the number of distinct grandmother stacks diminishes. When we reach Mitochondrial Eve, we will go right by her, since when she first appears she will be a member of a set of at least four women (Brandon's maternal ancestors) and probably seven or eight or nine. But this will quickly be apparent, since soon enough all members of the set will be a single grandmother stack, and we can locate ME as the only one with two daughters in the previous set.

p122-23. The standard account of the QWERTY phenomenon, as presented here (and by Papert and Gould and others), has been challenged by S. J. Liebowitz and Stephen E. Margolis, in "The Fable of the Keys," Journal of Law and Economics, 33, April 1990, pp1-25. source: [I have misfiled the letter that drew this article to my attention. My apologies, whoever you are.] But see also Jared Diamond's article, "The Curse of QWERTY," in Discover, April 1997, pp. 34-42, and the further bibliography he lists on the last page, including the web site http://ww.ccsi.com/~mbrooks/dvorak/dvorak.html. Diamond, in a letter to me, finds grounds for disagreeing with Liebowitz and Margolis, so the issue remains unresolved.

p126. H. Allen Orr says: "Population genetic theory, for instance, does not prove that evolution by random change is faster than evolution by natural selection." ("Dennett's Strange Idea," Boston Review, 21, Summer 1996, p28.) I misspoke, but the result was ambiguity, not error. If we compare regions of DNA that are under selection pressure with regions that are not, we discover that ON AVERAGE, the regions that are under selection pressure show a lower rate of change--because of stabilizing selection. That is, when a new selection pressure is introduced, this typically leads to a relatively short, rapid burst of change followed by a long period of near stasis. If you stop the race at fixation time, then of course evolution by natural selection looks faster than random drift, but that is an artifact of the time window chosen. In his reply to my original clarification of this point in the Boston Review (October/November, 1996, p37), Orr claimed not to know what I was talking about, and rudely suggested-- "I can find no polite way of putting this"--that I didn't either, but when I gave him this elaboration of my intended meaning, he acknowledged that it was not mistaken, but that in the context he had not considered it.

p171. line 9: "figure 7.6" should be "figure 7.7" source: W. Luff

p207. fn, line 2: "their" should be "its" source: W. Luff

p271. Robert Mark's fine article, "Architecture and Evolution," in the July-August 1996 issue of American Scientist (pp383-389), shows that I underestimated the structural demands of support for a large dome over arches, so that my bracket diagram (the middle diagram in Figure 10.3) would not be sound, lacking the necessary surcharge. Squinches would also not be sufficient, he claims, for such large, heavy domes. His final conclusion is that my "treatment of crucial structural elements as a kind of surface decoration that can be altered at will--'You have to put something there to hold up the dome--some shape or other, you decide which"--ignores the years, or in some cases even centuries, of construction experience that led to their incorporation in historic buildings." In other words, according to Mark, the pendentives of San Marco are very definitely structural adaptations, not Gouldian "spandrels". This point is elaborated upon by Alasdair I. Houston, in "Are the spandrels of San Marco really panglossian pendentives?" in Trends in Ecology and Evolution, March 1997, 12.

p.274. Robert Mark (see above) also criticizes my claim that the bosses in the fan vaults of King's College Chapel could have been replaced by "neat round holes," but see the letter from Stephen Grover, and Robert Mark's reply, in American Scientist Nov/Dec 1996, p518. As Grover points out, Walter C. Leedy's 1980 book Fan Vaulting does support me on this, and Patrick Bateson, Provost of King's College, has also written to me supporting my account.

p300. Cambrian explosion occurred around 530 million years ago, not "around 600 million years ago" source: Stephen Jay Gould.

p301. line two. Walcott himself did not "literally dissect" the fossils. source: Stephen Jay Gould.

p304. up 4-6 lines: the spelling should be Naraoia, Sanctacaris, and Leanchoilia. source: Stephen Jay Gould. As Gould points out (Wonderful Life, p68-9), Walcott's terms are "a strange-sounding lot. Decidedly not Latin in their roots, they are . . . sometimes nearly unpronounceable. . . . Walcott, who loved the Canadian Rockies, . . . labeled his fossils the the names of local peaks and lakes, themselves derived from Indian words."

p320. Teilhard de Chardin did his work in China before, not after, his difficulties with the authorities of the Roman Catholic Church. "His 'exile' was to the Wenner-Gren Foundation for Anthropological Research in New York City." source: Christopher Peebles.

p360. Richard Feynman is wrong! As Irwin Tessman, Department of Biological Sciences, Purdue University, points out in a letter to Nature ("Feynman faux pas," 381, 30 May, 1996, p361), "The example he gives would not prove replacement; in fact, it suggests quite the opposite. In an endearing lapse, Feynman appers to have blanked out on the half-life for the decay of 32P to sulphur, which just happens to be two weeks; if the radioactivity decreases to one-half in two weeks it means (within experimental error) that there was no turnover of phosphorus in the cerebrum."

p370. Goethe did indeed say it, in Faust. Mephistopheles says, in Part I:

Denn eben wo Begriffe fehlen,

Da stellt ein Wort zur rechten Zeit sich ein.

Ernst Mayr was the first to draw this to my attention, and his citation from memory [!] was almost exactly correct. Simon van der Meer also sent me the German text, and Wolfgang Heinemann provided me with the Wayre translation:

For if your meaning's threatened with stagnation,

Then words come in, to save the situation:

They'll fight your battles well if you enlist 'em,

Or furnish you a universal system.

Thus words will serve you grandly for a creed,

Where every syllable is guaranteed."

p488. Many readers have urged that I am wrong about which cards need to be turned over in the Wason test, but it is they who are mistaken--which just goes to show how powerful this cognitive illusion is. However, Simon van der Meer finds fault with my expression of the task: I should have said "Your task is to find all exceptions to the rule" (since otherwise, one could sometimes find an exception by just turning over the first card).

p499. Three Mile Island was in 1979, not 1980. source: Joseph P. Calendriello.

p.544. Quine's Word and Object was published by MIT Press, not Harvard University Press. source: Lynwood Bryant.

p545. Schrödinger's first name is Erwin, not Ernst. (Elsewhere in the book I get it right.) source: Andrew P. Cassidy.