The Mechanics of Discovery — Eureka and Processing Phase

Scientific progress occurs in two main phases: a eureka phase, in which an initial breakthrough discovery is made, and a subsequent processing phase, in which all the potential applications of that breakthrough are explored and employed.

As an example, a eureka would be the discovery of the CRISPR-Cas9 system, a discovery that can be used to manipulate our DNA and genes in groundbreaking ways. In the subsequent processing phase (which is still underway today), CRISPR-Cas9 technology is being optimized and adapted to cure genetic diseases.

Notably, the relationship between the two phases is not symmetrical. The processing phase relies entirely on the eureka — but not vice versa. You cannot invent a carriage if you don’t know what a wheel is. In many technological branches, we scientists find ourselves in a prolonged but highly productive processing phase, in which it can feel like we are accomplishing a lot. In the long term, however, controlling the frequency of eurekas is the rate-limiting step for scientific progress. The more eurekas and the faster they come, the greater the progress and the quicker its pace.

But increasing the frequency of such eureka discoveries is extremely hard, as they cannot be predicted. A true eureka can only be identified in hindsight. No one knew we needed the wheel before we invented it. No one was missing penicillin until it was discovered.

If eureka discoveries are the time-limiting step for scientific progress, how can we increase eurekas’ frequency? The dominant idea in academia is that eurekas are found by special individuals: the geniuses. Some major discoveries were indeed made this way. But examining the lives of the Marie Curies or Albert Einsteins reveals an obvious fact: their genius was only visible once they had made their discoveries.

The Nobel Prize committee initially intended to award the prize to Marie Curie’s husband instead of her, because scientists could not fathom that a woman could accomplish something like that. Albert Einstein could not find work in academia because, for the genius-searching universities, his math was not good enough. So he worked for a patent office when making his groundbreaking discoveries.

But even if some people appear special, genius does not determine breakthroughs. If individual genius were the main factor driving eurekas, one would expect that once we have identified such a genius, that person would find multiple eurekas. But most scientists who make one eureka discovery do not make another. They can have an edge in the specific processing phase of their eureka and make great further downstream discoveries, but the next eureka comes from somewhere unsuspected.

Only three individuals have received more than one Nobel Prize for inventions, for example (and in all of these cases, one could argue that the second prize could also be seen as part of the respective processing phase). Instead, eurekas are regularly found by “underdogs.” For example, the eureka for the gene scissor system CRISPR-Cas9 was found by a poorly recognized Spanish researcher, Francisco Mojica, and the mRNA vaccines that saved countless lives in the COVID pandemic were discovered by the completely neglected Kati Kariko.

Incentives Created by a Competition of “Geniuses”

Whatever the reasons, academia is reluctant to accept that eureka discoveries are not primarily dependent on individual genius. Moreover, scientific institutions cannot accept that even if individuals’ unique attributes play a role in a discovery, it is almost impossible to identify these attributes in advance.

Instead, the main incentives in academia all push scientists toward one goal: prove that you are a genius. The consequences are devastating.

First, since proving oneself to be a genius is an inherently individualistic activity, it incentivizes not collaborating on scientific questions. Recognition is greatest when you pretend you did everything on your own.

These perverse incentives are built into the academic system. For the acquisition of grants in life sciences, the only relevant measure today is so-called “last authorships.” This means that if you participate in a big study that, for example, investigates COVID patients in ten different countries, only one of the ten researchers can get the most prestigious recognition as “last author.” One can technically “share” the last authorship, but the reality is that only the literal “last author” — the last one in the list of authors on a paper — gets this recognition.

If you as a scientific researcher participate in such a study but don’t get the last authorship, you may have wasted your time, since your goal is to get grants to do more research, and only last authorship advances that goal. Worse, middle authorships are not clearly defined according to their contribution. Some big-name academics are regularly put in author lists for no work at all, as a nice gesture or because they gave a resource that they had monopolized. And coauthorships that are not the first or last authorship are largely meaningless.

Second, scientists are incentivized to find ways to inflate their perceived genius by showing that they publish in highly recognized publications in multiple fields. The easiest way to accomplish this: claim ownership over the work of subordinates and outsource risk.

Once you are a well-known professor, you can hire many junior researchers who do the actual work. But as the well-known professor, you get the main credit when it comes to the publication and presentation of the work. This is possible because in life sciences, the “reward” that junior researchers get — namely, first authorships (your name is the first in the list of authors of a publication) — is completely detached from the “reward” for the professor: the last authorship. It doesn’t matter if the junior researcher is responsible for 99 percent of the publication or 50 percent. In both cases, the junior researcher gets the first and the professor the last authorship.

Since this is true for all subordinates, even a professor who does not contribute anything relevant can look as productive as a dozen researchers with very little expenditure of resources. The first authorship, however, is not that valuable — it is read as something more akin to “This diligent worker executed the genius ideas of the senior researcher well.”

Conversely, while success gets bundled into the senior researcher, the opposite happens to risk. The tenured professor of a large laboratory has no problem if half or even three-quarters of their staff completely fail in their research. Since the measurement of the professor’s academic success is not adjusted to the number of workers responsible for the work (a desirable metric would be “publications per junior staff”), risk is completely carried by the workers.

Junior researchers sink or swim in giant laboratories. If they sink, they are on their own to find a new path forward. If they “swim,” the professor profits.

By hiring a large group of junior researchers, leading scientists can massively increase their output — while having “outsourced” and individualized the risk of failure to their workers. If this sounds similar to capitalist workplaces, that is because it is. Laboratories operate on the same principles as other workplaces. Managers, leading scientists, and shareholders rarely carry much risk; workers, meanwhile, get dismissed if companies or a line of research falls short.

Eurekas Are Dependent on the Number of Autonomous Scientists

Exploitation of junior researchers by senior scientists reeks of unfairness. But why does it matter for eureka discoveries?

Since junior researchers are completely dependent on their superiors, they cannot decide what they work on, limiting their autonomy. And the unfair conditions lead many junior researchers to favor high-paying industry jobs over staying in academia. Together, this means that our current system allocates power and resources to a small number of researchers. This in turn decreases the absolute number of autonomous (meaning independent) scientists in academia.

The problem: it is this absolute number that influences the chances a eureka is found.

Since eurekas cannot be predicted, their discovery is “stochastic,” or random. The best way to increase the probability of a stochastic event occurring at least once is to increase the number of attempts. The chance of rolling a six with a single dice roll is 16.67 percent. The chance of rolling at least one six in ten dice rolls is 84 percent. Decreasing the number of autonomous scientists is the worst idea if we want more eureka discoveries.

If eureka discoveries are in fact mostly stochastic, then science policy should focus on increasing the number of autonomous researchers. What measures could be taken to do this?

Instead of focusing resources on a few individuals, access to the resources needed to conduct research should be broad. This can be achieved by capping the number of funds and subordinates a single professor can have. Instead of rewarding the most exploitative professors, senior scientists should be incentivized to build their subordinates up so they can become autonomous researchers quickly.

An important tool to accomplish this would be to measure success and eligibility for grants in terms of the number of junior scientists: the most exploitative laboratories would look very different if grants were given out according to publications or citations per staff.

Instead of leaving it to the discretion of the “genius” professor how long subordinate researchers need to graduate or publish (many professors keep researchers for as many years as possible), all PhD and postdoctoral programs should be strictly time-regulated. After a certain time, junior scientists should have the right to mentor their own staff to build them up for their next career step.

Moreover, funding should not be given out in “competitions” that are thought to identify our societies’ geniuses. We have to accept that we cannot identify beforehand who will produce a eureka discovery. Currently, when a junior researcher starts their own lab, they have to immediately participate in and win these competitions or lose their job. Instead of having to prove themselves immediately, the funds for the first years for any new laboratory should be provided and guaranteed.

Of course, such policies should not end in academia, as the favoring of the few at the expense of the many begins much earlier, in charter schools and elite programs for young students. Instead, the focus should lie on increasing the number of people with access to a rich, public education in science and technology, since it is mostly that number or quantity (not the quality of the ones that make it) that will drive our future progress.

Science and Socialism

Obviously, the problems I’ve described here are not unique to academia — they are problems that can be found throughout capitalist societies. Capitalism produces a whole range of problems that you can read about in this magazine; in the case of scientific discoveries, capitalism is not as efficient as other systems could be.

Fifteen percent of adults and a whopping 45 percent of American scientists believe that scientific achievements in the uber-capitalist United States are the best in the world. But when measuring scientific output per capita, the most social democratic countries — Norway, Sweden, and Denmark — are all in the top ten, while the United States is number thirty-nine.

Moreover, when looking at Nobel Prizes in the same way, the United States has won 1.2 Nobel Prizes per million people (rank eleven), while Norway, Sweden, and Denmark are in the top five and received 2.4, 3.2, and 2.2 prizes per million respectively. Even judging by the current scientific metrics of competitions, the hypercompetitive US isn’t stacking up. Countries with much greater equality and a broad access to a good education ensure that more people can become autonomous scientists.

In fact, the number of people who are granted access to self-actualization and scientific education — and not the monopolization of resources — determines the bursts of technological progress. The Renaissance and the Scientific Revolution were the result of opening science up to larger and larger numbers of people and breaking the monopoly on research and study previously held by a small handful. Modern capitalism, which focuses on allocating resources and opportunities to a small elite, is designed to keep most of our society’s potential untapped.

By opening scientific research up to a wider range of people, socialism can jump-start another such scientific revolution. A “scientific socialism,” to put an old phrase to new use, could dramatically accelerate the discovery of eurekas and open up new and unimaginable technological progress. Fighting climate catastrophe, winning new generations to socialist politics, and pushing our society forward depend on our ability to articulate just such a scientific socialism for the twenty-first century.