The Failure in Science Conference at Imperial College London

Dr Stephen Webster, director of the Good Science Project at Imperial College London, opened the conference by welcoming colleagues from Imperial and participants from outside the university.

The event was developed with Magdalena Skipper, editor-in-chief of Nature. After Skipper suggested holding a conference about failure in science, Webster and Skipper worked together to bring the idea to life.

In April 2026, Imperial College London welcomed 150 delegates to discuss a topic that is rarely addressed openly: what happens when science goes wrong.

The conference examined mistakes, unsuccessful projects, missed opportunities and valuable ideas that may otherwise be overlooked. It also formed the basis of a six-part podcast series exploring different types of failure in science.

Rather than treating failure as an individual problem or a matter of bad luck, the series considers whether failure can also be institutional. It asks whether universities should view failure as part of the creative process and a necessary component of good science.

The discussions also explore how institutions can respond more constructively to failure when rapid progress and successful outcomes are often treated as the highest priorities. Another focus is what can happen when scientific knowledge leaves the laboratory and is applied in wider society.

Why Failure Is Still a Taboo in Science

Magdalena Skipper began by thanking Imperial College London for following through on the idea of a conference about failure—a subject she described as taboo.

Failure is part of success and one of the best teachers available, Skipper said, provided that people learn from it and avoid repeating the same mistakes. In scientific research, failure is an everyday occurrence. Most experiments do not produce the expected result, while career plans and research projects can also take unexpected turns.

Those experiences can provide valuable information. Researchers may need to reconsider their assumptions, change their methods or move in a new direction. Yet despite the usefulness of these lessons, scientists often do not discuss failure honestly.

“If you failed, you’re a loser, and that’s why nobody wants to talk about it.”

— Magdalena Skipper

Skipper argued that failure is part of both professional and personal life. Everyone fails at something, and the crucial question is how much they learn from the experience.

Learning from failure can build resilience. It can also encourage people to examine a situation more deeply and understand it in a more detailed and nuanced way. Failure is therefore both a natural process and a powerful teacher.

Failure Happens at Individual and Institutional Levels

The Failure in Science conference examined failure at several levels. One was personal: how individuals experience failure, how they recover and what they learn from it.

Another was institutional. The conference considered how universities and other organisations respond when research projects fail, funding is rejected or scientific decisions do not produce the intended results.

The discussions also extended beyond academia. Science and research take place in the private sector and in society more broadly, where attitudes towards failure can differ significantly.

Changing Career Direction Is Not Necessarily Failure

Skipper described a personal example of how the meaning of failure can depend on perspective.

After completing a PhD and a postdoctoral position, she decided to leave active scientific research and become a scientific editor. Looking back, she considers it one of the best professional decisions she has made.

However, people who believe that the only valid path for a researcher is to remain in active research interpreted her decision differently. To them, leaving academic research represented a failure to become a successful academic scientist.

For Skipper, the same decision represented success: she chose a career that suited her better. She noted that she has been described as a “failed scientist”, just as she could be called a failed firefighter because her childhood ambition to become one never materialised.

The important point, she said, is not whether someone has failed at everything. It is whether they can learn from past experiences and continue as a more knowledgeable and resilient person.

Fear of Failure Can Prevent Scientific Risk-Taking

A culture that treats failure as unacceptable can make people play safe. If researchers are afraid of failing, they may avoid taking risks, asking ambitious questions or developing hypotheses that could lead to unexpected discoveries.

In science, risk-taking can involve addressing major questions, proposing bold hypotheses or seeking funding for long-term projects whose outcomes are uncertain.

Projects of this kind may fail. However, they also have the potential to produce major advances, open new areas of research, create new disciplines or reveal different ways of investigating important questions.

Skipper questioned whether current research funding systems are designed to push the boundaries of knowledge or whether they encourage researchers to take only small, incremental steps.

Researchers applying for grants often already know that parts of their proposed work are likely to succeed because they have demonstrated them on a smaller scale. This can make it difficult to secure funding for work that is genuinely uncertain or exploratory.

The result may be a system that rewards projects built on established foundations rather than research that investigates the unknown.

What Science Can Learn from “Fail Fast” Culture

Skipper noted that different industries have different attitudes towards failure. She referred to a practice she understands to be relatively widespread in technology companies, particularly in Silicon Valley.

In some technology settings, employees may be financially rewarded for failing quickly. The purpose is to encourage risk-taking without allowing unsuccessful projects to continue indefinitely.

In a research setting such as Google X, Skipper suggested that rewarding only positive results could unintentionally discourage ambitious work. If employees believe they will be rewarded only when they succeed, they may avoid projects that carry a high risk of failure.

Rewarding fast failure changes the incentive. It recognises that stopping an unsuccessful project early can prevent further investment of time and money, while also encouraging people to test new ideas.

Skipper contrasted this approach with academic research, where projects may continue for a long time even when it becomes clear that a different decision should have been made earlier.

One reason may be emotional attachment. Once researchers have created a project, it can become their “baby”. Continuing with it may feel easier than acknowledging that it should be stopped or redirected.

Failure may be unavoidable on the way to success, but knowing when to stop can be just as important as persevering.

When Rejected Research Funding Leads to Greater Success

Skipper also discussed a metascience study—research into how science itself works—that examined the long-term careers of researchers near a funding cutoff.

The study used data from a United States funder and compared two groups: researchers who narrowly received funding and researchers who narrowly missed it.

The researchers followed both groups over time. The study found that those who just missed receiving funding later performed better in terms of career development and contributions to science than those who had just received funding.

Skipper emphasised that the study reported an observation, while explanations for the result remain open to interpretation. One possibility is that researchers who strongly believe in an idea may find another way to pursue it after funding is rejected.

That setback may encourage inventiveness and provide additional motivation. Failure can become a source of determination, particularly when researchers remain convinced that their work is worthwhile.

Scientific Progress Depends on Self-Correction

Science advances through self-correction and repeated attempts. Researchers working in laboratories, computational settings and the field understand that many experiments will fail.

When that happens, they adjust their expectations, change parameters or modify the research question. This process of trial and error is not random. It is an intentional part of scientific investigation.

Yet public descriptions of science often leave out this iterative process. Scientific discoveries may be presented as if they emerged fully formed, rather than through a sequence of experiments, failures, revised hypotheses and improved explanations.

According to Skipper, the reality is more complex. Scientists follow the scientific method, but the public does not always hear about the stops, starts and corrections that are part of research.

What the COVID-19 Pandemic Revealed About Scientific Uncertainty

The COVID-19 pandemic showed why communicating scientific uncertainty is so important.

During the pandemic, advice from scientists and decisions by policymakers changed as new evidence emerged. Recommendations about face coverings, isolation and other measures appeared to shift over time.

For people who understand the scientific process, this development was expected. Researchers were dealing with a situation that no one alive today had previously experienced on the same scale. As evidence changed, recommendations also needed to change.

However, many members of the public became confused or frustrated when guidance appeared to change direction. Without an explanation of how scientific knowledge develops, changing recommendations can look like incompetence or contradiction.

Explaining the role of uncertainty, evidence and correction could make these changes easier to understand. A failed prediction or revised recommendation does not necessarily mean that science has failed. It can demonstrate that researchers are responding to new information.

Could Openness About Failure Increase Trust in Science?

Skipper argued that scientists should communicate more openly about the process behind their conclusions. That includes explaining how researchers test ideas, encounter problems, revise hypotheses and improve their understanding.

When these steps are hidden, scientific communication can sound more certain than the evidence allows. It may even resemble a statement of belief rather than a conclusion reached through an iterative process.

Greater honesty about failure could help the public understand why scientific advice sometimes changes. It could also make science appear more transparent and trustworthy.

Trust in science is a growing concern. Showing that scientific knowledge is built through evidence, correction and learning may help address declining confidence.

Communicating failure does not mean presenting science as unreliable. Instead, it can demonstrate why science deserves trust: conclusions are tested, challenged and updated when better evidence becomes available.

Can a Conference About Failure Be a Success?

Skipper considered whether a conference about failure should itself be judged as a success or a failure.

Her conclusion was that the discussions would be valuable because they encouraged researchers and institutions to examine an important but neglected part of scientific work.

The conference highlighted that failure can be personal, institutional and societal. It can be painful and costly, but it can also lead to resilience, creativity, better decisions and new discoveries.

“Success is the ability to go from one failure to another with no loss of enthusiasm.”

— Winston Churchill, as quoted by Magdalena Skipper

For researchers and institutions, that may be the central lesson: scientific progress does not require the absence of failure. It requires the ability to learn from failure, recognise when a project needs to change and continue pursuing meaningful questions with informed enthusiasm.