As I investigated the question of whether biological information might have arisen by chance, it became abundantly clear to me that the probability of the necessary events is exceedingly small. Nevertheless I realized, based on my previous conversations with Bill Dembski, that the probability of an event by itself does not alone determine whether the event could be reasonably explained by chance. The probabilities, as small as they were, were not by themselves conclusive. I remembered that I also had to consider the number of opportunities that the event in question might have had to occur. I had to take into account what Dembski called the probabilistic resources.
But what were those resources—how many opportunities did the necessary proteins or genes have to arise by chance? The advocates of the chance hypothesis envisioned amino acids or nucleotide bases, phosphates, and sugars knocking into each other in an ocean-sized soup until the correct arrangements of these building blocks arose by chance somewhere. Surely, such an environment would have generated many opportunities for the assembly of functional proteins and DNA molecules. But how many? And were there enough such opportunities to render these otherwise exceedingly improbable events probable?
Here again Bill Dembski’s work gave me a way to answer this question. Dembski had calculated the maximum number of events that could actually have taken place during the history of the observable universe.1 He did this to establish an upper boundary on the probabilistic resources that might be available to produce any event by chance.2
Dembski’s calculation was elegantly simple and yet made a powerful point. He noted that there were about 1080 elementary particles3 in the observable universe. (Because there is an upper limit on the speed of light, only those parts of the universe that are observable to us can affect events on earth. Thus, the observable universe is the only part of the universe with probabilistic resources relevant to explaining events on earth.) Dembski also noted that there had been roughly 1016 seconds since the big bang. (A few more have transpired since he made the calculation, but not enough to make a difference!)
He then introduced another parameter that enabled him to calculate the maximum number of opportunities that any particular event would have to take place since the origin of the universe. Due to the properties of gravity, matter, and electromagnetic radiation, physicists have determined that there is a limit to the number of physical transitions that can occur from one state to another within a given unit of time. According to physicists, a physical transition from one state to another cannot take place faster than light can traverse the smallest physically significant unit of distance (an indivisible “quantum” of space). That unit of distance is the so-called Planck length of 10–33 centimeters. Therefore, the time it takes light to traverse this smallest distance determines the shortest time in which any physical effect can occur. This unit of time is the Planck time of 10–43 seconds.
Knowing this, Dembski was able to calculate the largest number of opportunities that any material event had to occur in the observable universe since the big bang. Physically speaking, an event occurs when an elementary particle does something or interacts with other elementary particles. But since elementary particles can interact with each other only so many times per second (at most 1043 times), since there are a limited number (1080) of elementary particles, and since there has been a limited amount of time since the big bang (1016 seconds), there are a limited number of opportunities for any given event to occur in the entire history of the universe.
Dembski was able to calculate this number by simply multiplying the three relevant factors together: the number of elementary particles (1080) times the number of seconds since the big bang (1016) times the number of possible interactions per second (1043). His calculation fixed the total number of events that could have taken place in the observable universe since the origin of the universe at 10139.4 This then provided a measure of the probabilistic resources of the entire observable universe.
Other mathematicians and scientists have made similar calculations.5 During the 1930s, the French mathematician Emile Borel made a much less conservative estimate of the probabilistic resources of the universe, which he set at 1050.6 More recently, University of Pittsburgh physicist Bret Van de Sande has calculated the probabilistic resources of the universe at a more restrictive 2.6 × 1092.7 MIT computer scientist Seth Lloyd has calculated that the most bit operations the universe could have performed in its history (assuming the entire universe were given over to this single-minded task) is 10120, meaning that a specific bit operation with an improbability significantly greater than 1 chance in 10120 will likely never occur by chance.8 None of these probabilistic resources is sufficient to render the chance hypothesis plausible. Dembski’s calculation is the most conservative and gives chance its “best chance” to succeed. But even his calculation confirms the implausibility of the chance hypothesis, whether chance is invoked to explain the information necessary to build a single protein or the information necessary to build the suite of proteins needed to service a minimally complex cell.
Recall that the probability of producing a single 150-amino-acid functional protein by chance stands at about 1 in 10164. Thus, for each functional sequence of 150 amino acids, there are at least 10164 other possible nonfunctional sequences of the same length. Therefore, to have a good (i.e., better than 50–50) chance of producing a single functional protein of this length by chance, a random process would have to generate (or sample) more than one-half of the 10164 nonfunctional sequences corresponding to each functional sequence of that length. Unfortunately, that number vastly exceeds the most optimistic estimate of the probabilistic resources of the entire universe—that is, the number of events that could have occurred since the beginning of its existence.
To see this, notice again that to have a good (better than 50–50) chance of generating a functional protein by chance, more than half of the 10164 sequences would have to be produced. Now compare that number (call it .5 × 10164) to the maximum number of opportunities—10139—for that event to occur in the history of the universe. Notice that the first number (.5 × 10164) exceeds the second (10139) by more than twenty-four orders of magnitude, by more than a trillion trillion times.
What does this mean? It means that if every event in the universe over its entire history were devoted to producing combinations of amino acids of the correct length in a prebiotic soup (an extravagantly generous and even absurd assumption), the number of combinations thus produced would still represent a tiny fraction—less than 1 out of a trillion trillion—of the total number of events needed to have a 50 percent chance of generating a functional protein—any functional protein of modest length by chance alone.
In other words, even if the theoretically maximum number (10139) of amino-acid sequences possible were generated, the number of candidate sequences would still represent a minuscule portion of the total possible number of sequences (of a given length). For this reason, it would be vastly more probable than not that a functional protein of modest length would not have arisen by chance—simply too few of the possible sequences would have been sampled to provide a realistic opportunity for this to occur. Even taking the probabilistic resources of the whole universe into account, it is extremely unlikely that even a single protein of that length would have arisen by chance on the early earth. (And, as explained in the accompanying endnote, proteins capable of performing many necessary features of a minimally complex cell often have to be at least 150 amino acids in length. Moreover, there are good reasons to think that these large necessary proteins could not evolve from simpler proteins or peptide chains.)9
To see this probabilistic reasoning in everyday terms, imagine that Slick performs a blind search for a single red marble in a huge gunnysack filled with 10,000 marbles, the remainder of which are blue. To have a better than 50 percent chance of finding the one red marble, Slick must select more than 5,000 of the marbles. But Slick has only ten seconds in which to sample the marbles. Further, it takes one second to find and put each marble aside in another jar. Thus, he can hope to sample only 10 out of the 10,000 marbles in the time available. Is it likely that Slick would find the red marble? Clearly not. Given his probabilistic resources, he has just 1 chance in 1,000 of finding the red marble, which is much less than 1 in 2, or 50 percent.10 Thus, it is much more likely than not that he will not find the red marble by chance in the time available.
In the same way, it is much more likely than not that a random process would not produce (or find) even one functional protein (of modest length) in the whole history of the universe. Given the number of possible sequences that need to be sampled and the number of opportunities available to do so, the odds of success are much smaller than 1/2—the point at which the chance hypothesis becomes reasonable (see below). Indeed, the odds of producing a single functional protein by chance in the whole history of the universe are actually much smaller than the odds of Slick finding the one red marble in my illustration. And beyond that, of course, the odds of producing the suite of proteins necessary to service a minimally complex cell by chance alone are almost unimaginably smaller. Indeed, the improbability of that event—calculated conservatively (see Chapter 9) at 1 chance in 1041,000—completely dwarfs the probabilistic resources of the whole universe. Taking all those resources—10139 possible events—into account only increases the probability of producing a minimally complex cell by chance alone to, at best, 1 chance in 1040,861, again, an unimaginably small probability.
And that is the third reason that the origin-of-life researchers have rejected the chance hypothesis. The complexity of the events that origin-of-life researchers need to explain exceeds the probabilistic resources of the entire universe. In other words, the universe itself does not possess the probabilistic resources necessary to render probable the origin of biological information by chance alone.
The “Chance” of Chance
I knew from my conversations with Bill Dembski and my study of statistical hypothesis testing that the occurrence of an improbable event alone does not justify rejecting a chance hypothesis. To justify eliminating chance, I knew that it was also necessary to recognize a pattern in an event and to consider the available probabilistic resources. The calculations presented in the previous section meet both of these conditions. DNA and proteins manifest functionally significant patterns. And, given all the probabilistic resources of the universe, the odds of producing a functional protein of modest length stands at less than one chance in trillion trillion.
In making this calculation, I have computed what statisticians call a “conditional probability.” A conditional probability measures the probability of one thing being true on the assumption that another thing is true. In this case, I have calculated the probability of a functional protein occurring by chance given, or “conditioned on,” a best-case estimate of the relevant probabilistic resources.
But this calculation has also been “conditioned” on something else. Recall that all along I had been attempting to determine the odds of a functional protein occurring by chance. That itself was another kind of conditional or “given.” I had not just been asking: What are the odds that a functional protein would arise? I had been asking: What are the odds that a functional protein or a minimally complex cell would arise by chance, given the available probabilistic resources? In other words, I had been asking: What are the odds that a functional protein or a cell would arise given the chance hypothesis (i.e., given the truth of the chance hypothesis)? Recall that the chance hypothesis in this case asserts that functional proteins or information-rich DNA molecules arose from the random interactions of molecular building blocks in a prebiotic soup. Framing the question this way—as a question about the probability of the origin of proteins given the chance hypothesis—provided grounds for evaluating whether it was more reasonable or not to accept the chance hypothesis.
This was important, because I often encounter people who think that it makes sense to cling to the chance hypothesis as long there was some chance—any chance, however small—that life might have arisen by some specified or even unspecified random process. They often say things like, “Sure, the origin of life is overwhelmingly improbable, but as long as there is at least some chance of life arising by chance, then we shouldn’t reject the possibility that it did.”
This way of reasoning turns out to be fallacious, however, because it fails to recognize what probabilistic resources can tell us about whether it is more reasonable to accept or reject chance. Consider a case in which all probabilistic resources have been considered and the conditional probability of an event occurring given the chance hypothesis is greater than 1/2. That means that it is more likely than not that the event in question would have occurred by chance (if every opportunity for it to occur had been realized). If Slick is given not one minute, but, say, twelve hours (720 minutes) to sample the bag of marbles, it is more likely than not that he would find the red marble by sampling randomly. In twelve hours Slick can sample 7,200 marbles, which gives him a better than 50 percent chance of finding the red marble.
Conversely, if after all probabilistic resources have been considered and the conditional probability of an event occurring by chance is less than 1/2, then it is less likely than not that the event will occur by chance. In the case that such an event has already occurred and we have no direct knowledge of how it occurred, it makes more sense to reject the chance hypothesis than to accept it.
My earlier illustrations made this point. When our hypothetical criminal claimed to have flipped 100 heads in a row on his second day of jail, suspicions were rightly raised. Given the improbability of the required outcome (producing 100 heads in a row by flipping a fair coin) and the limited probabilistic resources available to the prisoner, the court assumed—reasonably—that the prisoner had cheated, that something other than chance had been at work.
Similarly, imagine that after starting my demonstration with Scrabble letters, I left the classroom for a few minutes and instructed my students to continue picking letters at random and writing the results on the board in my absence. Now imagine that upon my return they showed me a detailed message on the blackboard such as Einstein’s famous dictum: “God does not play dice with the universe.” Would it be more reasonable for me to suppose that they had cheated (perhaps, as a gag) or that they had gotten lucky? Clearly, I should suspect (strongly) that they had cheated. I should reject the chance hypothesis. Why?
I should reject chance as the best explanation not only because of the improbability of the sequence my students had generated, but also because of what I knew about the probabilistic resources available to them. If I had made a prediction before I had left the room, I would have predicted that they could not have generated a sequence of that length by chance alone in the time available. But even after seeing the sequence on the board, I still should have rejected the chance hypothesis.
Indeed, I should know that my students did not have anything like the probabilistic resources to have a realistic chance of generating a sequence of that improbability by chance alone. In one hour my students could not have generated anything but a minuscule fraction of the total possible number of 40-character sequences corresponding to the length of the message they had written on the board. The odds that they could have produced that sequence—or any meaningful sequence at all of that length—in the time available by choosing letters at random was exceedingly low—much less than 1/2. They simply did not have time to sample anything close to the number of 40-character sequences that they would have needed to have a 50 percent chance of generating a meaningful sequence of that length. Thus, it would be much more likely than not that they would not produce a meaningful sequence of that length by chance in the time available and, therefore, it was also vastly more likely than not that something other than chance had been in play.
Decision Time: Assessing the Chance Hypothesis
Following many leading origin-of-life researchers, I came to the same conclusion about the first life and even the first genes and proteins: it is much more likely than not that chance alone did not produce these phenomena. Life, of course, does exist. So do the information-rich biological macromolecules upon which living cells depend. But the probability that even one of these information-rich molecules arose by chance, let alone the suite of such molecules necessary to maintain or build a minimally complex cell, is so small as to dwarf the probabilistic resources of the entire universe. The conditional probability that just one of these information-rich molecules arose by chance—in effect, the chance that chance is true—is much less than one-half. It is less than one in a trillion trillion. Thus, I concluded that it is more reasonable to reject the chance hypothesis than to accept it.
This was an intellectually liberating conclusion. Anyone can claim that a fantastically improbable event might have occurred by chance. Chance, in that sense, is always a possible explanation. But it doesn’t follow that chance necessarily constitutes the best explanation. And following what I knew about the historical scientific method, I wanted to find the best explanation for the origin of biological information. When I realized that I did not need to absolutely disprove the chance hypothesis in order to make an objective determination about its merits, clarity came. By assessing the probability of an event in light of the available probabilistic resources, I could determine whether it was more reasonable to affirm or to reject the chance hypothesis for that event. When I realized that it was far more reasonable to reject the chance hypothesis for the origin of functional genes and proteins, I concluded that chance was not a terribly promising candidate for “best explanation” of the DNA enigma. Chance was certainly not a more reasonable explanation than its negation, namely, that something other than chance had been at work in the origin of biological information. Indeed, when I remembered that the chance hypothesis implicitly negated both design and lawlike necessity, and that rejecting chance, therefore, meant affirming “something other than chance” at work, I began to evaluate alternative explanations.
My own reasoning about the chance hypothesis mirrored that of many origin-of-life researchers, many of whom expressed exasperation at the way some scientists used “chance” as a catchall explanation or a cover for ignorance.11 For example, after I first began reading the scientific articles about the origin of life, I found a spirited critique of chance published in Nature in 1963, just about the time molecular biologists were first coming to grips with the complexity of DNA and proteins. The paper was written by P. T. Mora, a senior research biologist at the National Institutes of Health. Here’s what he wrote:
To invoke statistical concepts, probability and complexity to account for the origin and the continuance of life is not felicitous or sufficient. As the complexity of a molecular aggregate increases, and indeed very complex arrangements and interrelationships of molecules are necessary for the simplest living unit, the probability of its existence under the disruptive and random influence of physico-chemical forces decreases; the probability that it will continue to function in a certain way, for example, to absorb and to repair, will be even lower; and the probability that it will reproduce, still lower. Statistical considerations, probability, complexity, etc., followed to their logical implications suggest that the origin and continuance of life is not controlled by such principles. An admission of this is the use of a period of practically infinite time to obtain the derived result. Using such logic, however, we can prove anything…. When in statistical processes, the probability is so low that for practical purposes infinite time must elapse for the occurrence of an event, statistical explanation is not helpful.12
I had come to much the same conclusion. Not only were the odds overwhelmingly against life arising by chance even considering all available probabilistic resources, but the chance hypothesis was usually invoked in a way that didn’t explain anything. To say that “given infinite time, life might have arisen by chance” was, in essence, a tautology. Given infinite time, anything might happen. But that doesn’t explain why life originated here or what actually caused it to do so.
Environmental Factors
There were some additional reasons to reject the chance hypothesis. The chance hypothesis for the origin of information-rich biological molecules assumes the existence of a favorable prebiotic soup in which an abundant supply of the chemical building blocks of proteins and nucleic acids could interact randomly over vast expanses of time. These chemical building blocks were thought to have been produced by the kinds of chemical reactions that Stanley Miller simulated in his famous 1953 experiment. Yet when Stanley Miller conducted his experiment simulating the production of amino acids on the early earth, he had presupposed that the earth’s atmosphere was composed of a mixture of what chemists call reducing gases, such as methane (CH4), ammonia (NH3), and hydrogen (H2). He also assumed that the earth’s atmosphere contained virtually no free oxygen.13 In the years following Miller’s experiment, however, new geochemical evidence showed that the assumptions Miller had made about the early atmosphere were incorrect. Instead, evidence strongly suggested that neutral gases such as carbon dioxide, nitrogen, and water vapor14—not methane, ammonia, and hydrogen—predominated in the early atmosphere. Moreover, a number of geochemical studies showed that significant amounts of free oxygen were also present even before the advent of plant life, probably as the result of the photo-dissociation of water vapor.15
This new geological and geochemical evidence implied that prebiotic atmospheric conditions were hostile, not friendly, to the production of amino acids and other essential building blocks of life. As had been well known even before Miller’s experiment, amino acids will form readily in a mixture of reducing gases. In a chemically neutral atmosphere, however, reactions among atmospheric gases will not take place readily, and those reactions that do take place will produce extremely low yields of biological building blocks.16 Further, even a small amount of atmospheric oxygen will quench the production of biologically significant building blocks and cause biomolecules otherwise present to degrade rapidly.17
An analogy may help to illustrate. Making amino acids in a reducing atmosphere is like getting vinegar and baking soda to react. Because the reaction releases stored chemical energy as heat, it occurs easily. (It is an example of what chemists call an “exothermic” reaction.) Trying to make biological building blocks in a neutral atmosphere, however, is more like trying to get oil and water to mix.18
Scientists investigating the origin of life haven’t tried to adjust their probability calculations in light of these developments. But they have recognized that these developments do complicate matters further for the chance hypothesis. To make matters worse, an accumulating body of geochemical evidence has shown—perhaps, not surprisingly, in light of the previous discussion—that there likely never was a prebiotic soup. Two leading geochemists, James Brooks and Gordon Shaw, argued that if an ocean rich in amino and nucleic acids had existed, it would have left large deposits of nitrogen-rich minerals (nitrogenous cokes) in metamorphosed Precambrian sedimentary rocks. No evidence of such deposits exists, however. In the words of Brooks: “The nitrogen content of early Pre-Cambrian organic matter is relatively low (less than .015%). From this we can be reasonably certain that: there never was any substantial amount of ‘primitive soup’ on earth when Pre-Cambrian sediments were formed; if such a soup ever existed it was only for a brief period of time.”19
Given my own deliberations about what constituted a substantive rather than a vacuous chance hypothesis, this seemed significant. A substantive chance hypothesis must invoke a definite process that produces the outcome in question with some regular or statistically predictable frequency—just as a roulette wheel produces various outcomes with a predictable frequency. The chance hypothesis envisioned DNA and proteins arising from a random process of chemical “roulette” in a favorable nitrogen-rich prebiotic ocean. If no such environment had ever existed, then whatever specificity the chance hypothesis might have once had was now lost. If there was no “chemical roulette” in which life would emerge as an inevitable if improbable outcome, chance could no longer be considered a substantive hypothesis; it would instead be just a vacuous notion that at best concealed ignorance of the true cause of life’s origin.
Additionally, I knew from my Ph.D. work that there were other significant, if less quantifiable, problems with the idea that information-rich biomolecules had arisen by chance from a prebiotic soup. Most origin-of-life researchers recognized that, even if there had been a favorable prebiotic soup, many destructive chemical processes would have necessarily been at work at the same time.20 Simulation experiments of the type performed by Stanley Miller had repeatedly demonstrated this. They have invariably produced nonbiological substances in addition to biological building blocks such as amino acids. Without intelligent intervention, these other substances will react readily with biologically relevant building blocks to form biologically irrelevant compounds—chemically insoluble sludge.21 To prevent this from happening and to move the simulation of chemical evolution along a biologically promising trajectory, experimenters often remove those chemicals22 that degrade or transform amino acids into nonbiologically relevant compounds. They also must artificially manipulate the initial conditions in their experiments. For example, rather than using both short-and long-wavelength ultraviolet light, which would have been present in any realistic early atmosphere, they use only short-wavelength UV. Why? The presence of the long-wavelength UV light quickly degrades amino acids.23
Zero Hour
I began to wonder if the odds of life arising by chance alone, at least under the circumstances envisioned by advocates of the chance hypothesis, weren’t actually zero. Imagine that a casino owner invents a game in which the object is to roll 777 consecutive “sevens” with a set of dice. He asks the odds makers to calculate the chances of any one contestant winning. They are, of course, infinitesimally small. But now he gives the odds makers some additional information. The dice are made of white chocolate with dark chocolate spots on the faces, both of which will melt as the result of the glare of the lights over the game table and repeated handling by the game players. Now what are the odds of turning up 777 sevens in a row? Clearly, they have diminished further. In fact, they are not just effectively zero, but under these circumstances with destructive processes inevitably predominating they are actually zero. Seven hundred seventy-seven “sevens” will never appear, because the faces of the dice will be destroyed in the attempt to generate them. The destructive processes will ensure that the desired outcome will never occur. I wondered if the same problem didn’t afflict the chance hypothesis for the origin of life.
In the face of these and other difficulties, most origin-of-life researchers have decided to consider other theories that do not rely heavily on chance. In the next chapter, I examine one of the other main contending approaches: self-organization. Since the odds of a purely random process producing life are “vanishingly small,” many scientists have concluded that some nonrandom, lawlike process must have been at work to help overcome these odds. If chance is insufficient, then perhaps “necessity” will do the job. Christian de Duve expresses the reasoning of researchers who favor this approach: “A string of improbable events—drawing the same lottery number twice, or the same bridge hand twice in a row—does not happen naturally. All of which lead me to conclude that life is an obligatory manifestation of matter, bound to arise where conditions are appropriate.”24
Nevertheless, I should note that a few theorists have attempted to either retain or resuscitate a role for chance. They have done so in one of two ways. Either they have attempted to lower the complexity threshold that random processes must first produce or postulated an increase in the probabilistic resources available to such processes.
Some theorists, notably those proposing an initial “RNA world,” have sought to retain a role for chance by suggesting that natural selection might have played a key role in the origin of life, even before the origin of a fully functioning cell. They propose combining chance with natural selection (or other lawlike processes) as a way of explaining how the first cell arose. In doing so, they suggest that random processes would have had to produce much less biological information by chance alone. Once a self-replicating molecule or small system of molecules had arisen, natural selection would “kick in” to help produce the additional necessary information. In Chapter 14, I evaluate theories that have adopted this strategy.
Others theorists have attempted to resuscitate chance theories altogether by postulating the existence of other possible universes beyond our own.25 In doing so, they have attempted to increase the probabilistic resources available for producing biological information and life itself. These other universes would presumably provide more opportunities to generate favorable environments in which random processes could generate the vast number of combinatorial possibilities necessary to give life a realistic chance of arising somewhere. Since this idea depends upon highly speculative and technical cosmological models, I have decided to evaluate it separately, in Appendix B. There I argue that although it is impossible to disprove the existence of other universes, postulating such universes does not actually provide either a satisfying explanation or, still less, the best explanation of the origin of biological information. As I explain, there are several reasons that such models fail to do so. For now, however, it is fair to say that most serious origin-of-life researchers have not found speculations about other possible universes terribly relevant to understanding what actually happened on the early earth to produce the first life on our planet. So in the next chapters, I turn my attention to evaluating the other leading contenders I have encountered in my investigation of the DNA enigma.