Newshot

‘I had felt overconfident’: senior scientists reveal mistakes that aided discovery

Summary

Nature, Published online: 20 November 2026; doi:10.1038/d41586-026-03590-2 Five experienced researchers, including a Nobel prizewinner, revisit the dead ends, bloopers and moments of stubborn certainty that shaped their science.

Original text(via Nature)
A portrait of Prof. Peter Shawhan wearing a hairnet and lanyard at the LIGO Livingston Observatory site.
A portrait of Prof. Peter Shawhan wearing a hairnet and lanyard at the LIGO Livingston Observatory site.

Physicist Peter Shawhan and his team perfected not finding gravitational waves at the Laser Interferometer Gravitational Wave Observatory (LIGO) facility in Livingston, Louisiana — until they did. Credit: Igor Andreoni/Peter Shawhan

To err is human, science endeavours included. Accounts of missteps in the laboratory or field abound among early-career researchers, as do tales of what they learnt from them. But senior scientists make mistakes, too — and not just early in their careers.

Sometimes, reaching a dead end was the only way around a problem. Other times, graduate students saved their advisers from throwing away ‘wrong’ data. Senior scientists also know the dangers of overthinking and how it can lead to experimental paralysis. “Every year, I devote a meeting with students and postdocs to discuss failures and how errors of judgement or methods are part of academic life,” says Catherine Badgley, a palaeontologist at the University of Michigan in Ann Arbor. “I hope this helps them to see they can learn from mistakes rather than be devastated by them.”

Nature’s careers team spoke to Badgley and four other senior scientists to hear about their ‘failures’ and the lessons they took from them.

PETER SHAWHAN: Learning to find nothing is a skill

Physicist at the University of Maryland in College Park and a deputy spokesperson for the Laser Interferometer Gravitational Wave Observatory (LIGO) scientific collaboration.

For the past 26 years, I have been searching for gravitational waves with LIGO, which has a gravitational-wave observatory in Livingston, Louisiana. These waves are echoes of massive events such as the merging of black holes or the collision of neutron stars. They produce ripples in space-time and, by the time these tiny ripples reach Earth, they are so faint that scientists need extra help to ‘hear’ them, because they are extremely hard to detect by just using light, or telescopes, to see them. For example, observations of a collision of two neutron stars with telescopes has revealed the presence of gold and platinum atoms, solving a long-standing question: what is the origin of about half of all elements heavier than iron? The detection of such an event with LIGO would get us even more information.

In the first tests of LIGO, the team found nothing. Later, with improved detectors, we still found nothing.

In fact, we got very good at reporting nothing. We became experienced with knowing what we could rule out. By the time we made our first wave detection, in 2015, a finding that earned LIGO founders a physics Nobel in 2017, we had published about 100 papers reporting null results.

One such null case was in 2010. We were completing a science run, in which data are collected for a set period of time, and had a candidate for a striking event — it looked like we’d finally got there. But we knew it might be a ‘blind injection’, which is when a false signal is inserted into a data set by authorized colleagues to test our analyses.

We finished the analysis, drafted the paper and then asked the LIGO director to open a sealed envelope to tell us whether this was a real event or not. It wasn’t, but the exercise showed a real problem: the sky location our analysis recovered didn’t match the signal’s known location. Two of the detector’s mirrors had moved in the wrong direction. We found the glitch in about a day. This showed we could efficiently self-correct when detecting a real sign — which happened five years later.

I tell my students that scientists make lots of mistakes. We’re careful with cross-checks and analyse results carefully, so we catch almost all of them. Some are caught in minutes, some in days and some take more work. This one, because we’d built a test designed to assess itself, took about a day. That’s really the goal: not avoiding mistakes, but rather building enough ways to catch them quickly. Knowing what it means to find nothing is important; it’s a real scientific skill.

Omar Yaghi poses with students for a group photo in 2024.
Omar Yaghi poses with students for a group photo in 2024.

Chemist Omar Yaghi (second row, fifth from right), with his team at Tsinghua University in Beijing, says that professors can learn important lessons from their students if they listen to them.Credit: Yaghi Group Photo

OMAR YAGHI: A fixed mindset is the enemy of discovery

Chemist at Tsinghua University in Beijing and a co-recipient of the 2025 Nobel Prize in Chemistry for his work with metal–organic frameworks.

Back in the 1990s, I was an assistant professor at Arizona State University in Tempe. My team had begun research trying to make inorganic porous materials. These are solid materials with a network of tiny, connected pores that act like molecular sponges to trap molecules, such as wastewater contaminants and toxic gases, on their surfaces. We were successful: we had a paper1 published in the Journal of the American Chemical Society and, later, in Science2.

Then a doctoral student came looking to join my lab. She didn’t sound very confident, her English was not so good and it wasn’t the right time of year to look for a PhD adviser. I made up some reason not to take her. But she kept coming back and was adamant about working with me.

So I accepted her into my lab group, and she started working on metal–organic compounds, another type of porous material. I felt that this was a waste of time: researchers had known about these structures since the 1950s, but nothing useful had resulted from them. Once you emptied their solvent-filled pores to trap other molecules, such as gases, they collapsed and became useless. I used to call them sculptures: they had beauty, but no function.

She kept at it, and I held onto my scepticism. After a year, we sat down to discuss her results. That was when I realized that there was an opportunity there: we could make robust metal–organic compounds, after all.

Metal–organic frameworks (MOFs) are also materials full of tiny pores. But what makes them particularly interesting is that scientists can adjust the size and shape of those pores, as well as the chemical properties inside them. This makes MOFs more tunable than inorganic porous materials such as zeolites and much better at trapping some gases. Some MOFs can drive chemical reactions, some conduct electricity and some are stable enough to be reused.

My student found strong chemical bindings that gave stability to early versions of these structures and, within a very short period, our findings were published in Nature3. It had worked.

Decades later, I met with my former student, and she told me that she had been struggling back then and wanted to try her luck with me, a new professor at the time, because no one else would take her in.

I think science is the greatest equalizer. One discovery can take you from humble beginnings, as it did for us, and propel you to the top of your profession.

I had an opinion about an area of research and didn’t listen to my student, and, to that student’s credit, she persisted. To my credit, I changed my thinking. Neither the student nor the professor can afford to do science without the other.

Prof. Chris Martine gestures in a patch of bloodbone tomato in northeastern Western Australia.
Prof. Chris Martine gestures in a patch of bloodbone tomato in northeastern Western Australia.

Plant geneticist Chris Martine’s misidentified sample led to the identification of a new species of bush tomato plant in northern Australia. Credit: Jason Cantley

CHRIS MARTINE: Overconfidence can be a trap

Plant geneticist at Bucknell University in Lewisburg, Pennsylvania, and president of the Botanical Society of America.

I’ve studied Australia’s bush tomatoes for about 25 years. My team was the first to describe Solanum ossicruentum, nicknamed the blood bone tomato because it seems to bleed when cut open. It got us some press coverage and was a big deal for our laboratory in 20164.

In 2017, Angela McDonnell, a postdoc, joined us. We started a big genetics project to build an evolutionary tree to understand the family history of the bush tomatoes we had been studying for years. We had two samples of S. ossicruentum from two locations: the Tanami Desert and the Keep River National Park, both in northern Australia. When the results of testing came back, genomic differences landed them in totally different places on the evolutionary tree of the group.

In my mind, there was no way it was possible. I’d identified the specimens myself. I assumed we’d mislabelled something. I asked Angela to check everything carefully, but she was certain that wasn’t the case. To me it didn’t add up — I came close to saying, “let’s just pull the misplaced one and call it an error”.

Two years ago, while looking through the folders of S. ossicruentum specimens at the Northern Territory Herbarium in Palmerston, Australia, I noticed some that had strange dots on the backs of their leaves. At first, I thought it was mildew, but then I wondered whether these dots were extrafloral nectaries, or nectar-producing glands found outside the plant’s flower. Then I realized the plants with such dots were mostly from one part of Australia, the northern edge of the Tanami Desert — exactly where I was heading next.

When I got there, plants with these little dots were dripping with nectar with ants harvesting it, an interaction I hadn’t seen to this level before. I checked whether Angela’s big phylogeny project had anything from that region, and there it was — the misplaced plant. The two samples of what I thought were S. ossicruentum belonged in two different places on the evolutionary tree because they were two different species. Indeed, that initial sample’s evolutionary tree placement was the best evidence for the new species that we later described: Solanum nectarifolium5. My first thought was to thank Angela for not pulling that specimen out of our analysis. I was honestly thrilled.

Before that, I had felt overconfident in what I knew, and Angela was certain that this was a carefully curated data set and a rigorously achieved genetic result. As I’ve become more senior, I’ve become more comfortable with uncertainty: when something doesn’t fit our preconceived ideas, that’s an interesting result — that’s how you find more new things.

Now I realize how important it is for people who are more junior than myself to speak up and share their ideas, especially when they run counter to convention. If we look back at the history of science, that’s the only way we’ve progressed.

Portrait of Catherine Badgley holding a fossil tibia.
Portrait of Catherine Badgley holding a fossil tibia.

Palaeontologist Catherine Badgley says that a key skill for scientists is knowing when to stop pursuing a good idea that simply isn’t working. Credit: David Bay

CATHERINE BADGLEY: It’s important to know when to stop

Palaeontologist at the University of Michigan in Ann Arbor.

In the 1990s, I was an associate research scientist in the University of Michigan’s Museum of Paleontology. Serge Legendre, a colleague who was then at the University of Montpellier in France, showed me a new way, at the time, of explaining body-size distributions in fossils of extinct mammalian species. He used cenograms — a graphical technique that ranks mammalian species in a community by body size.

In well-watered ecosystems, there was a fairly continuous line: all species were closely spaced in size. But habitats with long dry seasons had a big gap in the middle where medium-sized mammals would be — an interesting mystery. Legendre found that many animals in that size range are fruit-eaters, and fruit isn’t available in arid or very cold environments all year long. If the theory was correct, we might be able to infer past climate from the sizes of mammals in fossil communities.

I had a lot of climate data for mammal locations worldwide, so the two of us, as well as palaeontologist Philip Gingerich, who was then also at the University of Michigan, decided to build a much larger database of locations and add climate variables. We wanted to determine whether climate variables could predict the size gap we saw. I was excited enough that I wrote it up as a book proposal and received a contract from the University of Chicago Press.

But the more data that we collected, the fuzzier the picture became. After more than a year, I had written four chapters. But I had to admit we had fewer results than when we started.

I knew the data weren’t flawed because I had used this enormous mammalian fossil database successfully for other research projects. For example, undergraduate students used it to show that the length of the growing season, as a climate variable, could be used to predict the number of fruit- and plant-eating species at a given time.

I eventually realized that each continent has its own evolutionary history and its own climatic circumstances — no one is exactly like another. Our largest data sets, from Africa and North America, have almost completely non-overlapping climates, so one was never going to be a good model for predicting the other.

Enjoying our latest content? Log in or create an account to continue

  • Access the most recent journalism from Nature's award-winning team
  • Explore the latest features & opinion covering groundbreaking research

doi: https://doi.org/10.1038/d41586-026-03590-2

Subscribe to Nature Briefing: Careers, an unmissable free weekly round-up of help and advice for working scientists.

References

  • Yaghi, O. M., Sun, Z., Richardson, D. A. & Groy, T. L. J. Am. Chem. Soc. 116, 807–808 (1994).Article Google Scholar

Yaghi, O. M., Sun, Z., Richardson, D. A. & Groy, T. L. J. Am. Chem. Soc. 116, 807–808 (1994).

Article Google Scholar

  • Li, H., Laine, A., O’Keeffe, M. & Yaghi, O. M. Science 283, 1145–1147 (1999).Article PubMed Google Scholar

Li, H., Laine, A., O’Keeffe, M. & Yaghi, O. M. Science 283, 1145–1147 (1999).

Article PubMed Google Scholar

  • Yaghi, O., Li, G. & Li, H. Nature 378, 703–706 (1995).Article Google Scholar

Yaghi, O., Li, G. & Li, H. Nature 378, 703–706 (1995).

Article Google Scholar

  • Martine, C. T. et al. PhytoKeys 63, 19–29 (2016).Article Google Scholar

Martine, C. T. et al. PhytoKeys 63, 19–29 (2016).

Article Google Scholar

  • Martine, C. T., Brennan, K., Cantley, J. T., Webb, A. T & Newton, G. PhytoKeys 268, 183–199 (2025).Article Google Scholar

Martine, C. T., Brennan, K., Cantley, J. T., Webb, A. T & Newton, G. PhytoKeys 268, 183–199 (2025).

Article Google Scholar

Download references

Subjects

  • Scientific community

Latest on:

  • Scientific community
  • How to turn impostor syndrome into a superpower Career Feature 20 NOV 26

How to turn impostor syndrome into a superpower

Career Feature 20 NOV 26

  • Conference networking made simple, in three short videos Career News 18 NOV 26

Conference networking made simple, in three short videos

Career News 18 NOV 26

  • A grant to hand out more grants — and build a research area Career News 18 NOV 26

A grant to hand out more grants — and build a research area

Career News 18 NOV 26

  • How to turn impostor syndrome into a superpower Career Feature 20 NOV 26

How to turn impostor syndrome into a superpower

Career Feature 20 NOV 26

  • Conference networking made simple, in three short videos Career News 18 NOV 26

Conference networking made simple, in three short videos

Career News 18 NOV 26

  • What postdoc ads ask for and what they leave unsaid Career Column 09 NOV 26

What postdoc ads ask for and what they leave unsaid

Career Column 09 NOV 26

  • How AI is affecting science careers Career News 20 NOV 26

How AI is affecting science careers

Career News 20 NOV 26

  • US science academy urges journals to fix errors more quickly News 20 NOV 26

US science academy urges journals to fix errors more quickly

News 20 NOV 26

  • How to turn impostor syndrome into a superpower Career Feature 20 NOV 26

How to turn impostor syndrome into a superpower

Career Feature 20 NOV 26

Jobs

  • Associate or Senior Editor (Computational and mathematical biology), Nature Communications Job Title: Associate or Senior Editor, Nature Communications (Computational and mathematical biology) Location: Shanghai, Nanjing, Beijing, Pune,... Shanghai, Nanjing, Beijing, Pune, New Delhi, Hybrid working model Springer Nature Ltd

Associate or Senior Editor (Computational and mathematical biology), Nature Communications

Job Title: Associate or Senior Editor, Nature Communications (Computational and mathematical biology) Location: Shanghai, Nanjing, Beijing, Pune,...

Shanghai, Nanjing, Beijing, Pune, New Delhi, Hybrid working model

Springer Nature Ltd

  • Associate or Senior Editor (Biomaterials), Nature Communications Job Title: Associate or Senior Editor, Nature Communications (Biomaterials) Location: Shanghai, Nanjing, Beijing, Pune, New Delhi, Hybrid working... Shanghai, Nanjing, Beijing, Pune, New Delhi, Hybrid working model Springer Nature Ltd

Associate or Senior Editor (Biomaterials), Nature Communications

Job Title: Associate or Senior Editor, Nature Communications (Biomaterials) Location: Shanghai, Nanjing, Beijing, Pune, New Delhi, Hybrid working...

Shanghai, Nanjing, Beijing, Pune, New Delhi, Hybrid working model

Springer Nature Ltd

  • Emerging Scholars Forum 2027 | 19-20 January 2027 | Singapore The College of Engineering at NTU invites outstanding senior graduate students and postdoctoral fellows to apply for the Emerging Scholars Forum 2027. Singapore (SG) Nanyang Technological University, Singapore

Emerging Scholars Forum 2027 | 19-20 January 2027 | Singapore

The College of Engineering at NTU invites outstanding senior graduate students and postdoctoral fellows to apply for the Emerging Scholars Forum 2027.

Singapore (SG)

Nanyang Technological University, Singapore

  • TT Faculty - Protein Engineering develops and applies innovative computational tools, particularly AI/ML-driven approaches to protein engineering and small-molecule discovery Cleveland, Ohio Case Western Reserve University/Pharmacology

TT Faculty - Protein Engineering

develops and applies innovative computational tools, particularly AI/ML-driven approaches to protein engineering and small-molecule discovery

Cleveland, Ohio

Case Western Reserve University/Pharmacology

  • Assistant/Associate Professor - Pharmacology This is a renewable, 12-month, non-tenure-track position, with the majority of time devoted to teaching and advising undergraduate students. Cleveland, Ohio Case Western Reserve University/Pharmacology

Assistant/Associate Professor - Pharmacology

This is a renewable, 12-month, non-tenure-track position, with the majority of time devoted to teaching and advising undergraduate students.

Cleveland, Ohio

Case Western Reserve University/Pharmacology

The text above was extracted automatically from the source page and may miss images, tables or code blocks. The original is authoritative.

Published by Nature
This site only summarises and aggregates; copyright stays with the original author. Headlines and summaries follow the source — where they differ, the original wins.
Read the original ↗