Aaron Clauset

Clauset and Larremore Lab research discussed in "A handful of universities may control flow of ideas, people in academia"

Anonymous — Tue, 11 Jun 2024 18:41:19 +0000

Clauset and Larremore Lab research discussed in "A handful of universities may control flow of ideas, people in academia" Anonymous (not verified) Tue, 06/11/2024 - 12:41

Just five U.S. universities have trained 1-in-8 tenure-track faculty members serving at the nation’s institutions of higher learning, according to new CU Boulder research.

The study, published Sept. 21 in the journal Nature, takes the most exhaustive look yet at the structure of the American professoriate—capturing data on nearly 300,000 tenure-track faculty (including where they received their own graduate degrees) at more than 10,000 university departments at 368 PhD-granting institutions from 2011 to 2020.

The study reveals that in all fields of academia, most professors come from a small number of institutions.

Off

Traditional

White

Stacking models for nearly optimal link prediction in complex networks

Anonymous — Tue, 11 Jun 2024 18:34:02 +0000

Stacking models for nearly optimal link prediction in complex networks Anonymous (not verified) Tue, 06/11/2024 - 12:34

Most real-world networks are incompletely observed. Algorithms that can accurately predict which links are missing can dramatically speed up network data collection and improve network model validation. Many algorithms now exist for predicting missing links, given a partially observed network, but it has remained unknown whether a single best predictor exists, how link predictability varies across methods and networks from different domains, and how close to optimality current methods are. We answer these questions by systematically evaluating 203 individual link predictor algorithms, representing three popular families of methods, applied to a large corpus of 550 structurally diverse networks from six scientific domains. We first show that individual algorithms exhibit a broad diversity of prediction errors, such that no one predictor or family is best, or worst, across all realistic inputs. We then exploit this diversity using network-based metalearning to construct a series of “stacked” models that combine predictors into a single algorithm. Applied to a broad range of synthetic networks, for which we may analytically calculate optimal performance, these stacked models achieve optimal or nearly optimal levels of accuracy. Applied to real-world networks, stacked models are superior, but their accuracy varies strongly by domain, suggesting that link prediction may be fundamentally easier in social networks than in biological or technological networks. These results indicate that the state of the art for link prediction comes from combining individual algorithms, which can achieve nearly optimal predictions. We close with a brief discussion of limitations and opportunities for further improvements.

Off

Traditional

White

Scale-free networks are rare

Anonymous — Wed, 23 Oct 2019 17:02:03 +0000

Scale-free networks are rare Anonymous (not verified) Wed, 10/23/2019 - 11:02

Real-world networks are often claimed to be scale free, meaning that the fraction of nodes with degree k follows a power law k^−α, a pattern with broad implications for the structure and dynamics of complex systems. However, the universality of scale-free networks remains controversial. Here, we organize different definitions of scale-free networks and construct a severe test of their empirical prevalence using state-of-the-art statistical tools applied to nearly 1000 social, biological, technological, transportation, and information networks. Across these networks, we find robust evidence that strongly scale-free structure is empirically rare, while for most networks, log-normal distributions fit the data as well or better than power laws. Furthermore, social networks are at best weakly scale free, while a handful of technological and biological networks appear strongly scale free. These findings highlight the structural diversity of real-world networks and the need for new theoretical explanations of these non-scale-free patterns.

Off

Traditional

White

Stacking Models for Nearly Optimal Link Prediction in Complex Networks

Anonymous — Wed, 23 Oct 2019 16:56:08 +0000

Stacking Models for Nearly Optimal Link Prediction in Complex Networks Anonymous (not verified) Wed, 10/23/2019 - 10:56

Most real-world networks are incompletely observed. Algorithms that can accurately predict which links are missing can dramatically speedup the collection of network data and improve the validity of network models. Many algorithms now exist for predicting missing links, given a partially observed network, but it has remained unknown whether a single best predictor exists, how link predictability varies across methods and networks from different domains, and how close to optimality current methods are. We answer these questions by systematically evaluating 203 individual link predictor algorithms, representing three popular families of methods, applied to a large corpus of 548 structurally diverse networks from six scientific domains. We first show that individual algorithms exhibit a broad diversity of prediction errors, such that no one predictor or family is best, or worst, across all realistic inputs. We then exploit this diversity via meta-learning to construct a series of “stacked” models that combine predictors into a single algorithm. Applied to a broad range of synthetic networks, for which we may analytically calculate optimal performance, these stacked models achieve optimal or nearly optimal levels of accuracy. Applied to real-world networks, stacked models are also superior, but their accuracy varies strongly by domain, suggesting that link prediction may be fundamentally easier in social networks than in biological or technological networks. These results indicate that the state-of-the-art for link prediction comes from combining individual algorithms, which achieves nearly optimal predictions. We close with a brief discussion of limitations and opportunities for further improvement of these results.

Off

Traditional

White

Pedigree and Productivity

Anonymous — Thu, 02 May 2019 06:00:00 +0000

Pedigree and Productivity Anonymous (not verified) Thu, 05/02/2019 - 00:00

Colleen Flaherty

A 2015 study found that “social inequality” across a range of disciplines was so bad that just 25 percent of Ph.D. institutions produced 71 to 86 percent of tenured and tenure-track professors, depending on field.

The effect was more extreme the farther up the chain the researchers looked, based on their own program ranking system: the top 10 programs in each discipline produced 1.6 to three times more faculty than even the next 10 programs. The top 11 to 20 programs produced 2.3 to 5.6 times more professors than the next 10. In theory, this reflects the quality of those programs. But critics say in-group hiring is also about snobbery.

Now computer scientists at the University of Colorado at Boulder who led that earlier study say academic pedigree isn’t destiny after all -- at least in terms of future productivity.

“Our results show that the prestige of faculty’s current work environment, not their training environment, drives their future scientific productivity,” says the new paper in Proceedings of the National Academy of Sciences. Current and past locations, meanwhile, "drive prominence.”

That is, when it comes to actual research output, where one works is more important than where one trained.

For this new study, researchers looked at productivity and prominence (measured in number of published papers and scholarly citations, respectively) for 2,453 tenure-line faculty members in 205 Ph.D.-granting computer science departments. The analysis was based on a matched-pairs experimental design. As opposed to a completely randomized design, matched pairs involve one binary factor and blocks that sort the experimental units into pairs.

Off

Traditional

White

'Pedigree is not destiny' when it comes to scholarly success

Anonymous — Wed, 01 May 2019 17:50:44 +0000

'Pedigree is not destiny' when it comes to scholarly success Anonymous (not verified) Wed, 05/01/2019 - 11:50

By: Sante Fe Institute

What matters more to a scientist’s career success: where they currently work, or where they got their Ph.D.? It’s a question a team of researchers teases apart in a new paper published in PNAS. Their analysis calls into question a common assumption underlying academia: that a researcher’s productivity reflects their scientific skill, which is reflected in the prestige of their doctoral training.

It’s true that faculty at prestigious universities publish more scientific papers and receive more citations and awards than professors at lower-ranked institutions. It’s also true that prestigious schools tend to hire new faculty who hold Ph.D.s from similarly prestigious programs. But according to the authors of the new study, an early career researcher’s current working environment is a better predictor of their future success than is the prestige of their doctoral training.

“Pedigree is not destiny,” says SFI External Professor Aaron Clauset (CU Boulder), a co-author on the paper. “Our analysis supports the fairly radical idea for academia that where you train doesn’t directly impact your future productivity.”

The team looked at two basic measures of academic success — productivity (how many papers a researcher publishes) and prominence (how often their work is cited) — of 2453 tenure-track faculty in all 205 Ph.D.-granting computer science departments in the US and Canada during the five years before and five years following those individual’s first faculty appointment.

“We wanted to disentangle the impact of environment on productivity and prominence, and to isolate the effects of where someone trained versus where they went on to work as faculty,” says lead author Samuel Way (CU Boulder). “On the prominence side, people do retain some benefit from having studied in a prestigious Ph.D. program. They continue to accumulate citations from their doctoral work.”

But the prestige of the training program seems to play little role in how many papers researchers go on to produce once they begin their appointments in a new place. “Someone like me, who trained at Colorado, and someone from MIT… if we both end up at Stanford, our productivity will look the same,” says Way.

The authors identify several possible mechanisms driving the increased productivity of faculty at more prestigious institutions. Selection criteria in hiring, expectations for high productivity once hired, and selective retention of productive faculty were all considered. “We only find weak evidence for each,” says Way. However, the prestige of the current work environment had a strong effect on productivity.

Identifying the underlying “forces that tilt the scientific playing field in favor of some scientists over others,” as Clauset says, is important for identifying and potentially correcting the systemic biases that may be limiting the production of scientific knowledge.

“…our findings have direct implications for research on the science of science, which often assumes, implicitly if not explicitly, that meritocratic principles or mechanisms govern the production of knowledge,” write the authors. “Theories and models that fail to account for the environmental mechanism identified here, and the more general causal effects of prestige on productivity and prominence, will thus be incomplete.”

Off

Traditional

White

Do all networks obey the scale-free law? Maybe not

Anonymous — Mon, 04 Mar 2019 23:22:48 +0000

Do all networks obey the scale-free law? Maybe not Anonymous (not verified) Mon, 03/04/2019 - 16:22

Daniel Strain

As Benjamin Franklin once joked, death and taxes are universal. Scale-free networks may not be, at least according to a new study from CU Boulder.

The research challenges a popular two-decade-old theory that networks of all kinds, from Facebook and Twitter to the interactions of genes in yeast cells, follow a common architecture that mathematicians call “scale-free.”

Such networks fit into a larger category of networks that are dominated by a few hubs with many more connections than the vast majority of nodes—think Twitter where for every Justin Bieber (105 million followers) and Kim Kardashian (60 million followers) out there, you can find thousands of users with just a handful of fans.

Key takeaways

A popular theory claims that all networks are “scale-free”—meaning that the patterns of connections coming into and out of nodes follows a precise mathematical structure called a power law distribution.
CU Boulder researchers set out to test that idea, analyzing more than 900 networks from the realms of biology, technology, transportation and more.
They found that only about 4 percent of networks met the strictest definition for being scale-free—and close to half didn’t fit the bill at all.

In research published this week in the journal Nature Communications, CU Boulder’s Anna Broido and Aaron Clauset set out to test that trendy theory. They used computational tools to analyze a huge dataset of more than 900 networks, with examples from the realms of biology, transportation, technology and more.

Their results suggest that death and taxes may not have much competition, at least in networks. Based on Broido and Clauset’s analysis, close to 50 percent of real networks didn’t meet even the most liberal definition of what makes a network scale-free.

Those findings matter, Broido said, because the shape of a network determines a lot about its properties, including how susceptible it is to targeted attacks or disease outbreaks.

“It’s important to be careful and precise in defining things like what it means to be a scale-free network,” said Broido, a graduate student in the Department of Applied Mathematics.

Clauset, an associate professor in the Department of Computer Science and the BioFrontiers Institute, agrees.

“The idea of scale-free networks has been a unifying but controversial theme in network theory for nearly 20 years,” he said. “Resolving the controversy has been difficult because we lacked good tools and broad data. What we’ve found now is that there is little evidence for classically scale-free networks except in a few specific places. Most networks don’t look scale-free at all.”

Power law

Deciding whether or not a network is “scale-free,” however, can be tricky. Many types of networks look similar from a distance.

But Scale-free networks are special because the patterns of connections coming into and out of nodes follows a precise mathematical form called a power law distribution.

“If human height followed a power law, you might expect one person to be as tall as the Empire State Building, 10,000 people to be as tall as a giraffe, and more than 150 million to be only about 7-inches-tall,” Clauset said.

Beginning in the late 1990s, a handful of researchers made a bold claim that all real-world networks follow a universal structure represented by such giraffe- and inch-sized disparities.

There was just one problem: “The original claims were mostly based on analyzing a handful of networks with very rough tools,” Clauset said. “The idea was provocative, but also, in retrospect, quite speculative.”

To take scale-free networks out of the realm of speculation, he and Broido turned to the Index of Complex Networks (ICON). This archive, which was assembled by Clauset’s research group at CU Boulder, lists data on thousands of networks from every scientific domain. They include the social links between Star Wars characters, interactions among yeast proteins, friendships on Facebook and Twitter, airplane travel and more.

Their findings were stark. By applying a series of statistical tests of increasing severity, the researchers calculated that only about 4 percent of the networks they studied met the strictest criteria for being scale free, meaning the number of connections that each node carried followed a power-law distribution. These special networks included some types of protein networks in cells and certain kinds of technological networks.

A multitude of shapes

But not all researchers use those exact requirements to decide what makes a scale-free network, Broido said. To account for these alternative definitions, she and Clauset adapted their tests to account for each of the variations.

“Wherever you’re coming from, one of our definitions should be close to what you’re thinking,” Broido said.

Despite the added flexibility, most networks still failed to show evidence even for weakly scale-free structure. Roughly half of all biological networks and all social networks, for example, didn’t look like anything close to a scale-free network, no matter how flexible the definitions were made.

Far from being a let-down, Clauset sees these null findings in a positive light: if scale-free isn’t the norm, then scientists are free to explore new and more accurate structures for the networks people encounter every day.

“The diversity of real networks presents a mystery,” he said. “What are the common shapes of the networks? How do different kinds of networks assemble and maintain their structure over time? I’m excited that our findings open up room to explore new ideas.”

Off

Traditional

White

Academic ideas are supposed to thrive on their merits. If only.

Anonymous — Wed, 24 Oct 2018 06:00:00 +0000

Academic ideas are supposed to thrive on their merits. If only. Anonymous (not verified) Wed, 10/24/2018 - 00:00

Henry Farrell

Allison C. Morgan, Dimitrios J. Economou, Samuel F. Way and Aaron Clauset are all scholars in the department of computer science at the University of Colorado at Boulder. They have just published an important new article about how ideas spread within the academy. I asked them a series of questions about their work.

People say that science is about the open spread of ideas. However, your research suggests that it really matters whether the scientist who has the idea is working at a highly prestigious university. How does prestige affect the flow of ideas inside the academic community?

Our paper examines a simple hypothesis: Ideas spread in academia by people carrying them from one university to another. This idea is reasonable because academic research is highly specialized, and most researchers spend much of their careers working on topics close to what they trained on during graduate school. So, if a researcher had started studying deep learning in graduate school, and was then hired by a university where no one else was working on it, then that person carried the idea of deep learning from one university to another. In this way, if a small set of universities train the majority of all the academics in a field, ideas that originate at those universities will be overrepresented in the field. And, it turns out that in some of our past research, we showed that prestigious universities dominate the hiring market, meaning ideas that are born at prestigious universities will tend to spread further than those born elsewhere, simply because they have enormous alumni networks. The hiring market dominance of universities like Stanford, MIT, Harvard, etc. means that the other 200+ research universities in the U.S. are likely working on ideas that originated from this tiny group of elite places. In this paper, we test the hypothesis that where an idea is born matters for how far it could spread in academia, showing that prestige can create systematic “epistemic inequality” as a result of the hiring imbalance.

You treat the spread of ideas as a contagious process, which spreads through the population a little like the flu. How exactly do ideas become infectious, spreading from scientist to scientist or university to university?

There’s a long history of thinking of ideas spreading through a population by some kind of transfer process. Memes are classic example of this way of thinking. We use a slightly more narrow definition in our study, in which we define an “idea” as academic scholarship on a well-defined topic, like deep learning or quantum computing. An idea can thus “spread” across universities if a researcher who studies that idea changes jobs from one university to another. In our analysis, we were most interested in the situation where a university “adopted” an idea by hiring a new researcher who had a track record of working on it. Of course, the range of ideas being studied in a university is much more dynamic than this simple model allows, since researchers can pick up ideas from professional meetings, reading the literature, collaborating with other researchers, or developing them from scratch. Our aim was to show that the process of hiring new researchers (typically young professors) is one mechanism by which ideas spread through the system. This way, we don’t need to account for all the different ways ideas circulate in academia. We only need to show that hiring does indeed influence who works on what ideas where.

Your results suggest that really good ideas are likely to spread, no matter who has them, but that mediocre ideas tend to spread further and stick around longer when they come from scientists at highly prestigious institutions. What implications does this have for the ways we should do science?

It’s heartening that we find that good ideas will spread well no matter where they are born. It’s a bit less encouraging that mediocre ideas can spread just as far as good ones if they originate at a prestigious university. If we want academia to act more like a meritocracy, then we should try harder to ensure that our evaluations of ideas are not biased by the prestige of their birthplace, but instead focus on their independent merits. In practice, this is much harder than it sounds, but some things do help. For instance, recent researchshows that peer review based on a double-blind system mitigates prestige biases. Double-blind review works by removing the simple signals that tend to tilt evaluations in favor of prestigious universities. Working out other ways to mitigate prestige biases, for example, in faculty hiring itself, is less clear, but perhaps even more important. That said, our findings show that the advantage of prestige in terms of influencing the circulation of ideas in academia can be very large. This suggests that many good ideas may not be getting the attention they might deserve as a result of not having a prestigious university name associated with them. Another implication is that in a system that incentivizes researchers to produce a large quantity of incremental ideas, ideas produced by researchers at more prestigious universities will tend to be more visible than similarly good, or even slightly better ideas from less prestigious universities.

As you say, it’s easy to study the spread of ideas among scientists, since there is lots of good data. What possible implications does your research have for the spread of ideas in other contexts, such as politics, or among the general public?

We suspect that similar structural advantages may exist in other systems with strong prestige hierarchies. For example, the small number of universities represented among the educational backgrounds of Supreme Court justices and their clerks may be suggestive of less diverse approaches, and less creative judicial solutions. When differences in social prestige exist, then a high-quality idea originating from the bottom of the hierarchy in a business or government will have a harder time catching on than if it came from the top, not due to any conscious or unconscious bias by individuals but rather because of the structure of the system itself. If we want to solve hard problems, of which there are many in science, technology, and society, we need to find better ways to incentivize and recognize good ideas, regardless of where they originate from, rather than relying so heavily on characteristics like reputations or affiliations of the person suggesting it.

Allison C. Morgan, Dimitrios J. Economou, Samuel F. Way, and Aaron Clauset are at the department of computer science at the University of Colorado at Boulder. Aaron Clauset is also affiliated with the BioFrontiers Institute at University of Colorado at Boulder and the Santa Fe Institute.

This article is one in a series supported by the MacArthur Foundation Research Network on Opening Governance that seeks to work collaboratively to increase our understanding of how to design more effective and legitimate democratic institutions using new technologies and new methods. Neither the MacArthur Foundation nor the network is responsible for the article’s specific content. Other posts can be found here.

Off

Traditional

White

BioFrontiers research cited in article related to updated UVa paid parental leave policy

Anonymous — Wed, 12 Sep 2018 17:57:18 +0000

BioFrontiers research cited in article related to updated UVa paid parental leave policy Anonymous (not verified) Wed, 09/12/2018 - 11:57

Ruth Serven Smith of The Daily Progress

Adapted from The Daily Progress article.

The University of Virginia on Tuesday announced expanded paid leave benefits for new parents — a move that goes beyond a state executive order and one that could help the school remain competitive with its peers.

In June, Gov. Ralph Northam issued an executive order providing eight weeks of paid parental leave to state employees. The order, which included full-time UVa employees, was effective immediately.

In January, UVa will expand leave for part-time salaried staff, as well. The expansion bumps UVa up to the nation’s median for paid parental leave, according to University of Colorado researchers.

Now, each parent who is a UVa employee can take time off after the birth of a child or placement of a child under 18 with them as an adoptive or foster parent or guardian. Part-time salaried staff will receive time off prorated to the amount of hours they work — for example, someone who works 30 hours per week will be eligible for six weeks of paid time off.

Previously, UVa employees who were tenured faculty, tenure-track or had a one-year appointment were only eligible for three weeks of leave at full pay after the arrival of a child under the age of 7. Faculty could ask for additional unpaid time off or use short-term disability leave. Staff and other employees appeared to previously be eligible for only the 12 weeks unpaid leave guaranteed by federal law.

“Spending time with a child who just joined your family is incredibly important,” UVa President Jim Ryan said in a statement Tuesday. “Besides giving parents and children a chance to bond, studies have shown that paid parental leave makes children healthier, raises productivity at work and prevents parents from having to choose between taking care of a child and keeping their jobs.”

Virginia Tech expanded its policy to one identical to UVa’s in August.

Nationally, only 5 percent of U.S. workers had access to paid family leave in 2017, according to the Bureau of Labor Statistics. In the public sector, access to paid family and medical leave is spotty. California, New Jersey, Rhode Island, New York, Washington, Massachusetts and the District of Columbia also cover some public employees or cities and counties that opt in.

When Northam signed the order, he said he hoped it would help state employees with young families who previously relied on a federal law that would protect their job while on leave but doesn’t require the time off to be paid.

Some universities have expanded leave on their own, but there is little consistency between which universities offer leave, the amount offered and if it is given to people who are not birth mothers, according to analysis from the University of Colorado at Boulder.

�鶹��Ƶ 60 percent of universities provide paid leave for new mothers or fathers, with an average duration of 14.2 weeks for women and 11.6 weeks for men, said Dan Larremore, an assistant professor of computer science who worked on the project.

“There was a huge amount of variability,” Larremore said. “I would have thought that the more prestigious places would have more of an incentive to treat new parents well, but it was all over the map.”

UVa’s eight weeks is at the median of leave offered at the 205 U.S. and Canadian universities surveyed, and below the amount offered by Princeton University, Yale University, the University of California at Berkeley and the University of North Carolina at Chapel Hill.

The group compared its findings about policies in academia with those of tech companies, some of which, such as Etsy Inc. and Spotify Technology, offer as much as six months of paid leave to men and women who become parents. Netflix Inc. offers a full year off.

But of higher importance than the amount of time off, said Allison Morgan, a graduate student who worked on the project, was making sure employees knew their school’s policy and keeping track of how the amount of leave offered might affect hiring, retention and tenure decisions.

The researchers are adding their parental leave findings to a project that looks at how factors such as gender, family makeup and an institution’s prestige interact in academia.

“We’re generally interested in institutional and perceived barriers to participation in the sciences, and one of the barriers we saw was the use of parental leave in the sciences,” Morgan said. “We want to uncover persistent inequities in science, and the data shows there are more disparities than we’d like.”

Ruth Serven Smith is a reporter for The Daily Progress. Contact her at (434) 978-7254, rserven@dailyprogress.com or @RuthServen on Twitter.

The paid parental leave project by Allison Morgan, Sam Way, Mirta Galesic, Aaron Clauset, and Dan Larremore at the �鶹��Ƶ and the Santa Fe Institute.

Off

Traditional

White

Nothing unusual about 'the long peace' since WWII

Anonymous — Mon, 26 Feb 2018 07:00:00 +0000

Nothing unusual about 'the long peace' since WWII Anonymous (not verified) Mon, 02/26/2018 - 00:00

Jenna Marshall

Since the end of World War II, few violent conflicts have erupted between major powers. Scholars have come to call this 73-year period “the long peace.” But is this stretch of relative calm truly unusual in modern human history – and evidence that peace-keeping efforts are working? Or is it a cyclical peace, destined to be broken, with few lessons for preventing interstate conflict?

A new analysis by Aaron Clauset, an assistant professor of computer science at the University of Colorado at Boulder and the BioFrontiers Institute and an external faculty member of the Santa Fe Institute, aims to answer that long-standing question using novel statistical techniques to tease out how the “long peace” stacks up against historical trends of calm and conflict.

“There’s been a debate among people who think about conflict and war, policymakers and researchers, whether or not the pattern since World War II represents a trend,” Clauset says. Resolving this debate “is important because it shapes how we think about peace,” he adds.

Determining whether we are truly in the midst of a prolonged period of peace can help us understand what is an effective deterrent to war and what is not. If this really is a “long peace,” we can then examine why — and identify the mechanisms that have contributed to that peace. But if it’s not an anomaly, we may need to use caution in ascribing the current calm to particular policies or actions.

Using data on interstate conflicts worldwide between 1823 and 2003, Clauset looked for trends in the magnitude of those conflicts and the years between them, and then used that information to create models to determine the plausibility of a trend toward peace since World War II. “The tools, the framework and the models we built in this paper haven’t been used before, and they allow us to distinguish trends from fluctuations,” Clauset says.

What he found is that this prolonged period of peace is not so unusual. “The results of the study are that at least statistically speaking, the efforts to create peace have not changed the frequency of war,” Clauset says. “These periods of peace are relatively common. It doesn’t appear that the rules that generate war have changed.”

Between 1823 and 1939, there were 19 large wars, and a major conflict occurred about every 6.2 years. Then came an especially violent period: Between 1914 and 1939, which encompasses the onsets of the first and second world wars, 10 large wars erupted — about one every 2.7 years. In contrast, during the long peace of the 1940–2003 post-war period, there were only 5 large wars — about one every 12.8 years. So essentially, the long peace “has simply balanced the books,” counterbalancing the “great violence” of the early to mid 20th century, Clauset writes.

That’s not to say, however, that the current calm is insignificant. “This fact does not detract from the importance of the long peace, or the proposed mechanisms that explain it,” such as the spread of democracy and international diplomacy, he writes in the paper. “However, the models indicate that the post-war pattern of peace would need to endure at least another 100–140 years to become a statistically significant trend.”

Clauset compares this to flipping a coin over a period of time. “If I’m seeing a low number of heads out into the future, how long will the coins need to flip before the pattern really looks different? At what point does this pattern start to look unusual?”

Clauset says he is hopeful that the study will encourage a rethink of “the long peace.”

“I hope it will encourage caution,” Clauset says. “It’s a worthwhile exercise to check our assumptions about whether there’s a real trend or not.”

Part of the reason we tend to overstate the significance of the lack of major interstate conflict since World War II may be because of a human tendency to overestimate our ability to understand complexity, he notes in the paper. “Human agency certainly plays a critical role in shaping shorter-term dynamics and specific events in the history of interstate wars,” he writes. “But, the distributed and changing nature of the international system evidently moderates the impact that individuals or coalitions can have on longer-term and larger-scale system dynamics.

The research was supported by the One Earth Future Foundation, whose mission is to catalyze systems that eliminate the root causes of war.

Read the paper, “Trends and Fluctuations in the Severity of Interstate Wars," in Science Advances (February 21, 2018)

Read the article, "Are we in the middle of a long peace—or on the brink of a major war?" in Science (February 21, 2018)

Read the article, "War may be closer than we think," in Pacific Standard (February 23, 2018)

Off

Traditional

White