 Whenever I go dancing, my motions are more or less at random. You might say that I tripped the light stochastic. Ah, the origin of species. Everyone knows the story. Charles Darwin, a brilliant scientist, bravely voyaging on the HMS Beagle and exploring the Galapagos Islands, where his careful observations of finches led him to an epiphany about evolution and the world was forever changed. That's generally how we're taught the history of science. We learn the names of those giants whose shoulders we now stand atop. Those geniuses whose blazing intellects allowed them to see further than anyone had ever seen before and beat a path for the rest of us humble mortals. But there's a bit of a problem with that story. You probably haven't heard of Alfred Russell Wallace, another British scientist and contemporary of Darwin's, who also voyaged in faraway lands to research and catalog exotic species. One year before Darwin published on the Origin of Species, Wallace submitted a paper to the Linnaean Society of London titled, On the Tendency of Varieties to Depart Indefinitely from the Original Type. Hmm, the exact sequence of events is up for debate and it's possible that some internal politics at the society were involved. But by almost any account, Wallace developed the idea of evolution and speciation through the process of natural selection independently and was fully aware of all the implications of that idea. Now, Darwin had been editing his book for nearly 20 years by that point. But wouldn't you know it, the great scientific hero and visionary whose name we still use to describe this very important idea ended up publishing it immediately after Wallace's paper. If this were a singular event, you might think to yourself, oh, ha, ha, what a coincidence. But mutual independent discovery of the same revolutionary idea happens all the time, often nearly simultaneously. Evolution, calculus, oxygen, the telegraph, a staggering number of Nobel Prize winning discoveries in physics, lightning definitely seems to strike multiple places at once. The narrative that large scientific advances are solely due to singularly brilliant individuals doing something that no one else could is called the heroic or great man model and it's what we learn in school. But it seems that historically, it's very rarely the case that only one hero could have possibly figured these things out. Maybe we can look at this history in a different way. Maybe the people involved aren't really that important at all. In the multiple discovery model, it's suggested that the thing that drives large scientific advances isn't the genius of any particular individual, but some underlying mechanism of civilization itself. From this point of view, there's a gradual accumulation of knowledge which is intrinsic to society that makes important discoveries more or less inevitable once it reaches a certain level. The jeopardy board slowly fills in with letters and at some point, anyone with half a brain knows exactly what the answer is. Very smart people can get us there a little bit earlier than we would otherwise. That's all. It's not like they should be revered as timeless heroes because they can guess the word jabberwocky with one fewer vowel. That model would explain all the times that we see independent discovery at the same idea. Darwin and Wallace were both smart guys, but maybe the conceptual groundwork for evolution had already been laid and was just waiting for someone smart enough to put the pieces together. But there's an issue here too. Independent discoveries of the same important idea sometimes happen centuries apart, like with fine-cut brass gears or universally programmable computers. And there doesn't seem to be any real reason that their potential wasn't realized the first time. I mean, if the next step is obvious to anyone with half a brain, In 2003, Dean Keith Simonton advanced a third model for how scientific discovery might work, something which he felt accounted for both the irregularity of scientific progress and the multiple discovery phenomenon, as well as a bunch of other data which he amassed for his paper. You probably remember the term stochastic from your high school chemistry class. It's a term used to describe chemical reactions which occur at random. It's really absurdly unlikely that two hydrogens and one oxygen will happen to collide at precisely the correct angle and speed to fuse together into a water molecule. But if you stuff a million of them into an enclosed space together, then sooner or later, you get that exact reaction. Simonton suggests that we imagine ideas as particles bouncing around in people's brains and epiphanies as those chance occurrences where two concepts happen to line up perfectly and fuse into something new. In his model, scientists iteratively combine existing concepts more or less at random and check the results for viability, rejecting combinations that don't make any sense and investing in the ones that might. Now, if you're a skeptically-minded sort of person, your first reaction here is probably to ask, okay, so what? How could we know if this stochastic model was right or wrong? What testable predictions would it imply? Those are great questions. Let's get into it. First, let's do a sanity check to make sure that it agrees with what we already know. A scientist can only try to combine ideas that they're aware of, so no matter how smart they are, they aren't gonna develop any key insights about, say, quantum physics if they don't know anything about it. Check. The more ideas which are being tried at once and the more reactive the reagents added to the mix, the more frequently we'd expect to see useful new combinations. Scientific discovery is accelerating as the field expands and communication increases. Check. Two people trying combinations from the same set of ideas could very well happen upon the same reaction, especially if they're communicating with each other. Check. Okay, so far so meh. But there are also some riskier predictions which can be made with this theory for which we need to look at some data. For example, let's think about it in the context of publication rates and impact. Some labs and researchers release dozens of papers every year, while others only release one every few years. Presumably, everyone's working hard and angling for a paper which will get thousands of citations and advance the field in significant ways. So what could account for that discrepancy in output? Well, if we assume that scientists can't iterate combinations of ideas significantly faster or slower than each other, it seems likely that the answer is just about signal to noise and editing. Maybe some scientists are keen to share every little tidbit that they come up with, while others will only publish if they know they found something important. If that were true and both styles share the same base rate of discovering something awesome, then the number of high impact papers shouldn't depend on how many papers that someone publishes. And that's exactly what we see. So much so that there's a name for the phenomenon, the Equal Odds Rule. Whether you publish 10 papers a year or one every five years, you're equally likely to happen upon something truly awesome in a given period of time. Speaking of time, what about the distribution of high impact papers over a scientist's career? Well, this curve is called a Poisson distribution and it's what we use to model things which have a very small chance of occurring but a large number of opportunities to occur. It's the same curve that we use to model stuff like radioactive decay or the number of mutations in a given length of DNA. Statistically driven phenomena where it doesn't really matter how many times something has happened already, it always has the same odds of happening in a given timeframe and it happens to match the distribution of high impact papers over a career perfectly. There aren't really any significant trends of runs or lolls. You might occasionally see two important discoveries happen close together or far apart but not with any sort of predictability. It appears that publishing a really high impact study is an independent and random phenomenon compared to other publications just like the stochastic model predicts. These are just some of the examples of the kind of data that Simonton pulls from for his thesis. I encourage you to read the whole paper. It's relatively approachable and it does explain a whole lot of stuff. The way that we usually learn the history of science, the heroic model, suggests that important discoveries like gravity or evolution are solely due to a unique sort of brilliance that geniuses come up with genius ideas and the rest of us not geniuses should just sort of wait around and make things comfortable for them in the meantime. Not only is that demonstrably untrue but the stochastic model suggests that the most important part of this whole discovery process is the number of combinations tried, how much thinking someone is willing to do and the kinds of ideas they're putting together. In order to get those landmark papers, it's necessary to try novel combinations that nobody's thought of before. It's obviously important to learn as much as possible about the field in question to see what's already been tried and what concepts are available for combination. But every researcher has a unique set of experiences that they can grab and mash against those existing concepts to produce novel combinations, even things like their personal experiences, Newton and his apple, Wallace and Darwin and their voyages, maybe you and that weird hobby that you have. And the data supports that idea. Multidisciplinary groups produce more novel output. Researchers who are involved in multiple disparate fields and who are widely read tend to be the ones who mix things up. Just spend a lot of time learning about the field you want to affect, spend a lot of time learning about random other stuff and combine concepts as quickly as you can. In other words, don't stop thinking. Now, where have I heard that before? Do you think that the stochastic model adequately explains scientific discovery? Please leave a comment below and let me know what you think. Thank you very much for watching. Don't forget to blah, blah, subscribe, blah, share and don't stop thunking.