The Story of an Unlikely Hypothesis (and a Fine Book)
On Sunday 26th October 2008, just nine days before what promised to be an historic US Presidential election, I received an urgent call from Bob, a man close to a Republican Congressman in the American West. He wanted to enlist my services in an effort to prove a scandalous allegation against Barack Obama, which - if justified - would surely impact on his prospects in that election. Namely, that his famous 1995 memoir, Dreams from my Father, on which so much of his reputation was built, was in fact authored largely by Bill Ayers, a Vietnam-era domestic terrorist. It became clear that this allegation had been circulating with increasing speed in the US, on websites, blogs and radio talk shows, and was gathering significant numbers of supporters. Several "stylometric" analyses of the texts had been done, claiming to find significant patterns in the writings of Obama and Ayers to back up the hypothesis. A Washington press conference was already planned for the next week, just in time to sow doubt in the minds of undecided voters before the election ... (For a continuation of this strange story, see The Sunday Times, 2nd November 2008 and also the resulting front page Sunday Times news story.)
After submitting this story, in response to the interest it has generated, I have prepared the following analysis, which will I hope make clear how I see things in regard to the Ayers allegation made by Jack Cashill, an American author. In short, I feel very confident that it is false. In view of what I have found, and the intrinsic unlikelihood of the hypothesis, I would be very surprised indeed if anything came to light to reverse this verdict. I hope that interested visitors to this site, whether Democrat or Republican or indeed entirely independent of American politics, will be pleased to discover that the next leader of the free world did not apparently get his impressive first book written by Bill Ayers!
Is There Any Evidence?
It would be very surprising, to say the least, if this story were true. That Bill Ayers would devote his energies to someone else's life story, six years before he even got round to publishing his own, seems unlikely. And even if parallel passages were to be found between Obama's book and Ayers's Fugitive Days of 2001, the charge of plagiarism would most naturally be directed at Ayers rather than Obama unless those passages appear also in earlier material from Ayers.
In fact, the parallels that Cashill has highlighted between Obama's and Ayers' books seem generally pretty weak. Here (slightly reformatted) is how he starts his discussion of the matter:
The stylistic similarities between the best sections of Obama's "Dreams" and Ayers' 2001 memoir, "Fugitive Days," should be apparent to any serious writer or editor. Consider the two following "nature" passages in Obama's and Ayers' respective memoirs, the first from "Fugitive Days", and the second from "Dreams":
These two sentences are alike in more than their poetic sense, their length and their gracefully layered structure. They tabulate nearly identically on the Flesch Reading Ease Score (FRES), something of a standard in the field.
- "I picture the street coming alive, awakening from the fury of winter, stirred from the chilly spring night by cold glimmers of sunlight angling through the city."
- "Night now fell in midafternoon, especially when the snowstorms rolled in, boundless prairie storms that set the sky close to the ground, the city lights reflected against the clouds."
Well, certainly there are some superficial similarities between the sentences Cashill quotes, but I for one don't find them that striking. They have a couple of significant words - "night" and "city" - in common, and both say something about the fall of light in fairly poetic language. Beyond that, their FRES scores are similar (54 and 54.8 respectively, according to Cashill), but this means very little: the score applying to a single sentence depends only on the number of words and the number of syllables, so it's hardly a sophisticated measure. Besides, we have been given only one sentence from each book to compare, and it's hardly surprising that Cashill is able to find two sentences which match this closely together. If large sections of Obama's book had actually been written by Ayers, one might reasonably expect to find far more striking parallels.
Evidence from Parallel Stories?
According to Cashill, such more striking parallels (pointing in the direction of Ayers' influence on Obama rather than the reverse) are to be found in three specific story themes that seem to be in common between earlier works of Ayers (To Teach of 1993, and A Kind and Just Parent of 1997) and Dreams from My Father. Here is the first of them, quoted from Cashill:
In his 1993 book, "To Teach," Ayers lays out the difference between "education" on the one hand and "training" on the other. "Education is for self-activating explorers of life, for those who would challenge fate, for doers and activists, for citizens," Ayers writes.
"Training," on the other hand, "is for slaves, for loyal subjects, for tractable employees, for willing consumers, for obedient soldiers."
In Obama's "Dreams," these thoughts find colloquial expression in the person of "Frank," the real life poet, pornographer and Stalinist Frank Marshall Davis.
"Understand something, boy," Frank tells the college-bound Obama. "You're not going to college to get educated. You're going there to get trained."
Frank shares Ayers' distaste for training. "They'll train you to forget what it is that you already know," Frank tells Obama. "They'll train you so good, you'll start believing what they tell you about equal opportunity and the American way and all that sh--.".
Frank also tells Obama that "leaving your race at the door" is an essential part of the university's training mission. Ayers makes the same case about training in reference to Indian schools, which insist, according to Ayers, that students be "stripped of everything Indian and taught to be like whites."
Again, one is left with the impression that if this is amongst the best evidence to be found, then the case must be extremely weak! Both Ayers and Obama, apparently, draw a distinction between education and mere training. But this distinction is entirely commonplace - huge numbers of people who've had anything to do with an educational establishment will be familiar with it. The link between training and race is no doubt more distinctive, but the coincidence is still not very impressive. Ayers does not actually mention training in the paragraph where he talks about Indian schools, but in the paragraph before that; and his point is how so much of "education" is really a matter of "certification mills" that ride roughshod over people's distinctiveness. Obama's Frank is making a different point, that education for a black man can mean "Leaving your race at the door ... Leaving your people behind." So Ayers is making a point about the lack of humanity of much education, and Obama a point about the impact of going through a training institution on a black person in a white-dominated world. There is indeed some similarity between the two, but the idea that the passages quoted above give significant evidence of direct influence is just ridiculous. Let's now take a look at Cashill's second example:
In the same 1993 book, "To Teach," Ayers tells the story of an adventurous teacher who takes her students out to the streets of New York to learn interesting life lessons about the culture and history of the city.
As Ayers tells it, the students were fascinated by the Hudson River nearby and asked to see it. When they got to the river's edge, one student says, "Look, the river is flowing up." A second student answers, "No, it has to flow south-down."
Upon further research, the teacher discovers "that the Hudson River is a tidal river, that it flows both north and south, and they had visited the exact spot where the tide stops its northward push."
In his 1995 book, "Dreams From My Father," Obama shares a stunningly similar story from his own brief New York sojourn. As Obama tells it, he takes an unlikely detour to the exact spot on the parallel East River where the north-flowing tide meets the south-flowing river.
There, improbably, a young black boy approaches this strange man and asks, "You know why sometimes the river runs that way and then sometimes it goes this way?" Obama tells the boy it "had to do with the tides."
Cashill again seems to be over-stressing the parallel here: nothing I can find in Obama's text justifies the sentence "As Obama tells it, he takes an unlikely detour to the exact spot on the parallel East River where the north-flowing tide meets the south-flowing river". Obama says nothing of the kind, but only that "I took the long way home, along the East River promenade", at which point the boy approached and asked his question, when Obama "said it probably had to do with the tides". As author he then reports that "I realized I had never noticed which way the river ran." There is no deep parallel here with Ayers' story of a teacher taking a class to see the Hudson River (whose Mohican name famously means "river that flows both ways"), with the intention of showing them "the exact spot" where the tide stops.
Cashill's third example is no better:
In his 1997 book, "A Kind and Just Parent," Ayers tells of a useful reading assignment from the 1992 book, "The Kind of Light That Shines on Texas," by black author Reginald McKnight.
The passage in question deals with the travails of Clint, the first black student in a newly integrated school, who tries to distance himself from Marvin, the only other black boy in the school.
"Can you believe that guy?" Clint tells a white student. "He's like a pig or something. Makes me sick." Upon reflection, Clint thinks, "I was ashamed. Ashamed for not defending Marvin and ashamed that Marvin even existed."
In "Dreams," Obama reflects on his own first days as a 10-year-old at his Hawaiian prep school, a transition complicated by the presence of "Coretta," the only other black student in the class.
When the other students accuse Obama of having a girlfriend, Obama shoves Coretta and insists that she leave him alone.
Although "his act of betrayal" buys him a reprieve from the other students, Obama, like Clint, understands that he "had been tested and found wanting."
First, the Clint example is from McKnight's 1992 book, and Ayers' mention of it in 1997 came later than Obama's book (so if Obama took inspiration from it, that directly indicates McKnight's rather than Ayers' influence). But secondly, again the parallel isn't as close as Cashill suggests. In the McKnight story, there are three black children in the school, two boys and a girl. In Obama's, there are two black children in a class, himself and Coretta. Admittedly there is some similarity in the situation and emotions described, but it is very unclear how big a coincidence this is. It might well be that such experiences were very common for black children where they were a small minority. And if so, then the similarity provides at best very weak evidence of any link between the two books.
If the parallels between the stories in Obama's book and those of Ayers were very close, then we might be forced to conclude that there was some connection involved - maybe that Ayers and Obama had read some of the same works (e.g. by McKnight), or that Obama had read some of Ayers' earlier works and picked up some stories from those. Even in this case, however, the claim of common authorship would be far from the most obvious explanation of the coincidence. But as we have seen, the parallels are nothing like close enough to establish even a modest connection with any confidence. In books written just a few years apart, by authors whose interests include education and race, it is not at all surprising to find parallels of this weak kind.
So much for the three common stories that Cashill claims to find in the Obama and Ayers books. Let us now turn to his stylometric analyses.
Cashill's First Stylometric Analysis
Jack Cashill points to four stylometric analyses (done by others) that claim to support his theory that Ayers wrote Dreams from My Father. Two of these employed my own software Signature, but the first of them used only its most primitive feature, counting words of different lengths. The analyst then applied various statistical measures, leading to conclusions that as quoted by Cashill sound quite striking: "The statistical style analysis performed by our research team suggests that the writing style of 'Dreams From My Father' is significantly more similar to the style observed in 'Fugitive Days' than to the style found in other works by Barack Obama such as 'Audacity of Hope.'" Well, suppose we now look at all this using Signature 2.0:
Here the blue line shows the word length statistics for Audacity, green for Dreams, and red for Ayers's Fugitive. And in respect of word length, the claim made is true! But this does nothing to show that Dreams was really written by Ayers, as can be illustrated if we add Bill Clinton's My Life into the mix:
Notice that the turquoise line (My Life) seems to follow the red one (Fugitive Days) very closely indeed: decisive evidence, perhaps, that Ayers ghost-wrote Clinton's memoir? Or could it perhaps be that Clinton ghost-wrote Ayers's book? Of course both suggestions are ridiculous. All this shows is that analysis of word-length frequencies - at least by itself - is hopeless as an indicator of authorship: there's a fairly high chance that different authors will have the same overall pattern, simply because they're writing in the same language and the same genre. By contrast, a single author could easily exhibit different patterns of word-length frequency when writing, say, a memoir (e.g. Dreams from my Father) as opposed to a popular political work (e.g. Audacity of Hope). So this sort of analysis can do virtually nothing to support Cashill's theory, no matter how much it might be developed with complex statistics. If you want such statistics, however, here is the "Multiple Chi-Square Analysis" table produced by Signature 2.0, comparing the four texts:
|Word Length Chi-Square Analysis|
This clearly confirms what was obvious by eye: according to this test, the closest two texts of the four are those by Ayers and Clinton! So there's nothing at all here to link Obama to Ayers. (For those who understand such things, the relevant Chi-Square 0.1% value on columns 2 to 8 (as used by the cited analysis) is 22.46, and only the Ayers-Clinton pairing gets even close to falling within this.)
Cashill's Second Stylometric Analysis
Another of Cashill's four cited studies did essentially the same sort of test on word-length data (only), comparing Dreams from my Father against Fugitive Days with a single control, the Memoirs of Ulysses S. Grant of 1885. According to this analyst as quoted by Cashill, "The Ayers-Obama matching shows a measurable and substantial effect. It is easily and objectively distinguishable from comparison to a third document. ... the initial data presented is highly suggestive that these two documents share large portions of authorship." Needless to say, this last claim is completely unjustified: the word-length frequency correlations are not remotely close enough to be "highly suggestive" of co-authorship, as can clearly be seen from the graphs and Chi-Square statistics above. Nor does the "easy and objective distinguishability" from Grant's Memoirs count for anything: it isn't the least bit surprising that two memoirs written at the end of the 20th century have more in common than one written over a century before.
Cashill's Third Stylometric Analysis
A third stylometric analysis, which like the first made use of Signature, involved tests of word frequencies, comparing Dreams from my Father, Fugitive Days, and - as a control - "a randomly chosen text" which in fact was Sinclair Lewis's 1919 road novel, Free Air. Tests were done with a large set of words and a small set (of eleven words), but the latter is problematic for the same reason that we saw above with sentence lengths: there is no way that it can possibly be sufficiently discriminating to confirm identity of authorship in any significant way. These analysts found that Dreams was more like Fugitive Days than Free Air in some respects, but that of course isn't surprising at all (given the difference in genre and vintage). If we add more realistic controls, then the apparent similarity - which isn't even impressive to start with - entirely disappears, as shown by the following "Principal Component Analysis" graph using the same eleven words:
Suppose we take a much larger set of words, using Signature 2.0 first to identify the 50 most common words in all of these texts combined, then again doing a Principal Component Analysis:
Again there is nothing to link Obama with Ayers. And all the evidence so far examined if anything points against there being any close link between them.
Cashill's Fourth Stylometric Analysis
Finally, Cashill's fourth analysis (by Chris Yavelow) used far more sophisticated measures, built into FictionFixer, a proprietary software system for helping aspiring writers to develop their style. This analysis is genuinely interesting, but unfortunately it records no "control" measurements at all, so the results produced are impossible to assess. Yavelow describes some of his results as "striking", but to me they seem too weak to draw any really strong conclusions. Besides, without any comparable statistics involving other texts, we have no way of assessing their true significance - the history of stylometry is littered with the remains of defunct hypotheses built on what to the eye looked like "striking" coincidences but which then dissolved away under critical statistical examination. So Yavelow's claim to have made "a strong case for the likelihood that the author of Fugitive Days ghostwrote Dreams from My Father" is so far completely unsubstantiated. However his analysis is the only one of the four that stands any chance of providing any basis for a more substantial case.
Cashill's Literary Observations
For reasons such as those illustrated above - either lack of suitable controls, or unimpressive results - I'm not convinced that there's anything in the analyses cited by Cashill that needs to be considered as at all suspicious. Even if genuine similarities between Obama's and Ayers's books were to be uncovered, they would have to be remarkably strong to overcome the evidence of the many dissimilarities, even to make the ghostwriting hypothesis plausible, let alone to establish it.
Cashill himself came to his theory on the basis of his own literary judgements and observations rather than stylometrics, and so I did a few quick tests on some of these, planning to follow them up were I to undertake a full analysis. For present purposes, just one deeper study will do. Below is a list - produced by Signature's concordance facility - of uses of "memory", a word that Cashill himself singles out as distinctive of both the Obama and Ayers books. In fact the word features proportionately around four times more frequently in Ayers's book than Obama's, so here I include Obama's fifteen uses, and Ayers's first fifteen (all based on texts that I have been given, and whose accuracy I cannot guarantee):
Ayers, Fugitive Days
- 1:83:15 in a flash unhinged and going under. Memory is a motherfucker. I myself remember almost
- 1:115:6 know I'll forget it all. Oblivion trumps memory. I remember the overdrawn story of my
- 1:476:18 of confinement long before I was sixteen. Memory is a house of mirrors, a land
- 1:557:6 idea where we might lake her dictum. Memory is a delicate dance of desire and
- 1:784:3 many white people, Baldwin believes, are without memory, "trapped in a history which they do
- 1:892:2 The past is a foreign country and memory is always in translation—deciphering, paraphrasing, consulting
- 1:892:12 in translation—deciphering, paraphrasing, consulting the thesaurus. Memory rewrites rather than transcribes. Whatever romance might
- 4:34:17 have to reinvent everything, especially the wheels. Memory sails out upon a murky sea-wine-dark
- 4:354:22 arrived to volunteer at our school. The memory I hold still from'Diana's first visit
- 4:590:6 so much about who you would become. Memory is a mortuary, a dead space. I
- 4:592:20 overprivileged—a lie of presidential proportions. Or memory is memorization, habit and routine, the cliches
- 4:593:10 cliches of common sense, ordinary life. Worse, memory can become the black slate monument hiding
- 4:842:17 suspended between worlds, everything possible, nothing guaranteed. Memory is a marvel, quick as a monkey
- 4:844:4 all we've got—we live by distorted memory rather than verifiable truth. Memory is feeling
- 4:844:9 by distorted memory rather than verifiable truth. Memory is feeling, not fact, ghosts and fears
Obama, Dreams from my Father
- 1:103:31 experiences hold for others, selective lapses of memory. Such hazards are only magnified when the
- 1:221:24 months, sometimes years, in my family’s memory. Like the few photographs of my father
- 1:429:9 arrived, it had somehow vanished from collective memory, like morning mist that the sun burned
- 1:852:6 undisguised, indiscriminate, naked, always fresh in the memory. Power had taken Lolo and yanked him
- 1:1228:18 yet when I reach back into my memory for the words of my father, the
- 1:1797:31 my heart, blur the edges of my memory. I had discovered that it didn’t
- 1:2032:4 later, and I still burned with the memory, the anger and resentment I’d felt
- 1:3683:7 little bit of nostalgia, elements of selective memory; but the whole of what they recalled
- 1:4317:17 stopped whenever we threatened to skirt his memory. It was only that night, after dinner
- 1:5172:2 The next moments are blurry in my memory. As I remember it, Linda leaned over
- 1:5509:8 come home.” His face glowed with the memory. “When was the last time you were
- 1:5876:15 the children would no longer retain the memory of that first circle, around a fire
- 1:6432:7 drained away, and I smiled with the memory of the homecoming I had once imagined
- 1:8762:47 rocks roiling the water, the breaks in memory separating the currents, but always the voices
- 1:9686:7 the law is. The law is also memory; the law also records a-long-running conversation
What is really striking here is how fond Ayers is of speaking of memory in the abstract: it is "a motherfucker", is "trumped" by oblivion, is "a house of mirrors", "a delicate dance of desire", something many white people lack, "always in translation", rewriting rather than transcribing, it "sails out upon a murky sea", is "a mortuary", or "habit and routine", it can become a "black slate monument", and is "a marvel", or "feeling". Of all those fifteen Ayers references, that leaves only two that do not clearly take this form:
and of these, only one refers to a definite, concrete, memory event.
- 4:354:22 arrived to volunteer at our school. The memory I hold still from Diana's first visit
- 4:844:4 all we've got—we live by distorted memory rather than verifiable truth. Memory is feeling
The contrast with Obama's book here is massive and obvious. Only his last reference can plausibly fall into the predominant Ayers pattern, but even this does not really do so: it is saying what the law is, not what memory is. So even if Cashill is right to point to "memory" as a distinctive theme within both books (and I am not arguing here with his literary observations, or those of the others he cites), the details of the case here tell clearly against, rather than in favour of his ghostwriting hypothesis. The difference between the two books in what they say about memory is so enormous, that it looks very unlikely that the same person could have written both. Consider in this light what Cashill himself says in responding to this objection:
If not as frequently as Ayers, Obama does write about memory in the abstract. Consider the following passage from Dreams:
“I heard all our voices begin to run together, the sound of three generations tumbling over each other like the currents of a slow-moving stream, my questions like rocks roiling the water, the breaks in memory separating the currents, but always the voices returning to that single course, a single story....”
Now consider the following passage from Fugitive Days:
“The debates swam above and around and through us . . . . The confrontation in the [Student Union] flowed like a swollen river in to the teach-in, carrying me along the cascading waters from room to room, hall to hall, bouncing off boulders.”
I would bet my house against Millican’s mailbox that the gifted writer Ayers wrote both these passages. Note their rhythm, cadence, and layered structure as well as Ayers’ obvious affection for the flow of water and language. I could identify twenty parallel passages just as compelling.
As pointed out earlier, it is not at all surprising that Cashill is able to find some passages in the Obama and Ayers books that have some superficially impressive similarities (he says that he "could identify twenty parallel passages just as compelling": this looks like pure bluster, until he has done so explicitly). But such superficial similarities totally fail to address the point made above about the contrast in the two writers' talk of memory. In 13 passages out of 15, Ayers talks of what memory is (or does) in the abstract. If he was also the author of Obama's book, then one would expect to find some reasonable proportion of similar passages within it, whereas one does not find any at all. The fact that one passage in Dreams which mentions memory is somewhat similar to a passage in Fugitive Days which does not mention memory is entirely irrelevant to answering this point. Cashill, apparently, ignores this simple logic, and so rashly bets his house against my mailbox on the matter. I hereby accept the bet. Let him put up, or shut up. If he is indeed prepared to bet his house on the outcome, then it may well be worth my investing more time on the issue to set out to prove at length that he is mistaken, rather than merely arguing (as I have done so far) that his case as presented is hopelessly unproven.
To sum up, I have found no evidence for Cashill's ghostwriting hypothesis, and rather strong (albeit limited) evidence against. Note, moreover, that the discussion above is all fought on his own chosen ground - analyses and literary patterns that he himself has cited as likely to tell in his favour. It is impossible to know without searching carefully what other evidence against his hypothesis might turn up if one were to look for it. But at least we must judge it as extremely unlikely, and indeed so far unsubstantiated. Moreover given its small initial probability (as a story that was always intrinsically unlikely - it would be very surprising if true), and the evidence already seen, I cannot imagine how the hypothesis could be rendered credible by further examination of the texts. Of course that doesn't mean it's impossible (for example if systematic comparative studies using Yavelow's system were to yield genuinely exceptional results), but I don't currently believe that anything short of documentary proof of Ayers's involvement will take the case further.
Signature used to investigate claims that Obama's book was written by an ex-terrorist.