Paleontological luck

Leonard Finkelman writes...

I have two stories linked by fossils and expletives.

The first story is set in Calgary International Airport, on my way home from Extinct Blog’s retreat in Alberta’s Dinosaur Provincial Park. Teaching responsibilities called me away earlier than the rest of the gang, who were still exploring the park at the time. I had just settled into a seat when my phone buzzed with a message from Adrian. It read: “So, um, on our walk Joyce and TJ sorta found a skull. We were all like, Leonard’s gonna be so mad.” A swear word escaped my lips and drifted into the air, unlike the flight that had just been canceled at the gate next to mine.

(Adrian wasn't entirely wrong on that last point, incidentally. There's no "I" in "team," but there is one in "paleontologist.")

The second story ends with the same swear word. A lab partner and I were in Oregon’s high desert, deep within a maze of eroding gullies. We climbed down there on a hunt for rare carnivore fossils. We came to a fork: he went left as I went right. By the end of the day, my search of the right fork had yielded only an oreodont skull—common by local standards—that would disintegrate during its excavation. As I climbed out of the maze, my partner let loose a joyous call from a short distance away along the left fork. “I found one!” he shouted from atop a bear-dog limb.

My lab partner's good luck exemplified.

My lab partner's good luck exemplified.

One more factor unites these stories: I blamed both outcomes on bad luck. Fossil hunting, so integral to paleontology, depends in large part upon luck. Paleontology isn't alone in this regard: lots of scientific research is determined by the researcher's good (or bad) luck. As I've reflected on the outcomes of my experiences in the field, however, it strikes me that paleontological luck is different from other kinds of scientific luck.

The species of luck

Philosophers and historians of science have recently taken more interest in the role that luck plays in scientific research. A significant proportion of research outcomes may be lucky in some sense or another, even when that research adheres to some rigorous experimental method. How exactly that luck influences research, and what that means for (say) scientific demarcation, is a matter of ongoing discussion.

Philosophers (unsurprisingly) have been thinking about luck for a bit longer. Modern philosophical consideration of luck begins (and, for many philosophers, ends) with Bernard Williams and Thomas Nagel. In the late 1970s, those two introduced a debate over moral luck into moral philosophy. It goes like this: sometimes we assign praise or blame to an agent whose actions are lucky in one sense or another, but praise or blame should only be assigned to agents who have control over their actions—i.e., whose actions don’t depend on luck. In assessing whether or not moral luck is a coherent concept, Nagel (1979) specified four particular kinds of luck: resultant, circumstantial, constitutive, and causal. Considering the differences between these will be useful in distinguishing paleontological luck from other scientific luck.

Well: considering some of them will be useful. We don’t really need to worry here about constitutive luck or causal luck.

Constitutive luck is luck due to biological characteristics or behavioral tendencies. Dinosaurs were at one point lucky in this sense: they diversified in the late Triassic period because their physiology is relatively oxygen-efficient and the atmosphere at the time was relatively oxygen-deficient. (This is the sort of luck that Stephen Jay Gould focused on in his work on evolutionary contingency.) Constitutive luck may be helpful to some scientists, but its absence doesn’t have to be a hindrance: consider Geerat Vermeij, blind from glaucoma since 1950—and a successful paleontologist since 1971.

Causal luck is luck due to the entire antecedent chain of events that led to the relevant moment. Dinosaurs were at another point unlucky in this sense: the asteroid that caused the K-PG event would have been set in motion by billiard-like events that occurred billions of years before the first true dinosaur took an oxygen-efficient breath of air. (This sort of luck was brilliantly visualized in the opening sequence of The Good Dinosaur, after which point the movie proceeded rapidly down the proverbial hill.) This type of luck affects all scientists equally, but only insofar as it affects all people equally.

There’s plenty more than can be said about these kinds of luck, but neither is relevant to the particular issue at hand. Both are kinds of luck with respect to background. Our concern is with luck with respect to outcomes.

Experimental luck and paleontological luck

Resultant luck is luck due to chance outcomes. Success in gambling is a paradigmatic example of resultant luck: I can’t guarantee (in any metaphysical or epistemic sense) that a roll of fair dice will turn up the number that I want. The result is chaotic in the mathematical sense of that term: the outcome is not necessitated by the event that sets it in motion.

In a terrific essay for NPR, Tania Lambrozo compared scientific luck with moral luck in the resultant sense. She asks that we imagine two scientists, Claire and Daniela, who set up the very same experiment in the very same conditions:

For reasons out of their control, Claire finds evidence; Daniela does not. Claire goes on to win a Nobel Prize in Physics; Daniela struggles to publish her inconclusive results. …We can further stipulate that they were equally careful and insightful, that they mentored and managed their research teams equally well, and so on. If we had changed some factor out of their control—say, the exact trajectory of a given particle on a given day—it would have been Daniela who found the evidence and went on to Nobel fame, not Claire.

This thought experiment sets up resultant luck as deeply embedded within the experimental method. A researcher can set up an experiment given perfect knowledge of the relevant state of the (scientific) art, but no experimental design guarantees the conclusion derived from its results. (This, after all, is the problem of Type I and Type II errors.) The problem is compounded if one if a regularist about laws of nature: if the laws of nature are only statistical regularities, and if the purpose of experimentation is to discover the laws of nature, then the outcome of even the best-designed experiment depends on resultant luck.

In most cases, however, paleontological fieldwork isn’t structured experimentally. The paleontologist isn’t looking for laws of nature; she’s hoping to find fossils, but she also acknowledges that she might be looking for something that isn’t there. Whether or not she finds those fossils is unpredictable, but not because of some epistemic disconnect between her understanding of theory and prediction of outcomes or because the fossil discovery isn’t a metaphysically necessary outcome of her actions. If the paleontologist must be lucky, then, in what sense is she lucky?

My lab partner was luckier than me because he went left when I went right. After that point, he and I engaged in exactly the same activities: we scoured our respective landscapes, found exposed fossils, and started digging. His happened to be a bear-dog. Mine happened to be a cautionary tale for a blog. Our actions guaranteed our respective outcomes, but those outcomes weren’t only guaranteed by our actions.

This is the mark of circumstantial luck, or luck due to environment. In this kind of luck outcomes are determined not by chance, but by (sometimes unknowable) external factors. Successful reference (in the pop cultural sense) is a good example: I might try to illustrate the concept of rationalism in an introductory course by referring to Ian Malcolm from Jurassic Park, but that teaching tool is only effective when students in the class have seen the movie. Given the same lesson plan and a different set of students, the reference might fail.

Paleontological fieldwork differs from experimentation in this regard: the outcomes aren’t a function of resultant luck. Borrowing from Lambrozo’s example, we can “rewind the tape” on the experiments run by Claire and Daniela and derive different results from the same initial conditions. We can rewind the tape on my day with my lab partner and stipulate that he and I might choose different directions, but that wouldn’t be a fair comparison: that would change conditions antecedent to the outcomes of our work. Rewind to the relevant starting point—when he and I split ways—and the same outcomes will obtain each time. His luck, and my lack thereof, was circumstantial.


Circumstantial luck is baked into the practice of paleontology. The same is true of archaeology, whose practice also depends on lucky discoveries from the field. It may also be true of field biology; however, there may be salient differences between that kind of fieldwork and the kind in which paleontologists and archaeologists participate.

Other scientists also benefit from circumstantial luck. The invention of the microwave, for example, was an outcome of circumstantial luck. But it was not the outcome of scientific practice per se: Percy Spencer, the scientist in question, was lucky, but his luck wasn't a part of any experimental design or research method. So it goes for many examples of "luck" in science: the luck in question aids science, but it isn't necessarily a part of science. Resultant luck, by contrast, might be a necessary component of scientific practice.

Similarly, paleontologist and archaeologists might benefit from other kinds of luck. Constitutive luck, for example, is a certain help: one needs sharp eyes to distinguish some fossils from the literal tons of rock surrounding them. But, again, those kinds of luck aren't a necessary component of the science's practice. If paleontology is to develop in the future, there are just going to have to be some paleontologists who get themselves into the right circumstances at the right times.