Failure is unavoidable in science, but academic institutions do not always treat it as a useful part of research. Mary Ryan, Vice Provost for Research and Enterprise at Imperial College London, says universities should be more open about rejected proposals, unsuccessful experiments and the uncertainty that surrounds scientific knowledge.
Ryan discussed the issue in the third episode of the “Failure in Science” podcast series, produced by Imperial’s Good Science Project. The episode examines how educational and research systems can contribute to failure, including through rigid assessment, grant pressures and expectations that encourage researchers to play safe.
For early-career scientists, the apparent success of established academics can make failure difficult to see as normal. Ryan notes that every academic has rejected grant proposals and papers, but says those outcomes do not necessarily mean the underlying ideas lacked value. Some ideas may have been premature or incomplete, while others may simply have been wrong. How researchers respond to rejection, she argues, helps shape them as scientists.
She also points to the history of graphene as an example of how important findings can be overlooked. In surface-science experiments, researchers used adhesive tape to cleave highly oriented pyrolytic graphite and discarded the tape after cleaning the surface. For years, pieces of graphene were thrown away because researchers were focused on the substrate rather than the material removed from it. The discovery of graphene through this technique later received the Nobel Prize.
The example illustrates how concentrating on a specific research goal can make scientists miss something significant. It also reflects a broader challenge for education: students need to understand established knowledge while remaining aware of the uncertainty at its boundaries. Ryan says this is uncomfortable for both educators and students, but that unanswered questions should have a more central role in teaching.
That challenge also applies to systems increasingly shaped by artificial intelligence. Ryan stresses that knowledge is not complete, even when it is captured and categorized by technology. Students need to ask what information a system was trained on and where the limits of that knowledge lie.
Science depends on experimentation, and experiments are often designed to test or disprove hypotheses. Ryan argues that findings that disprove an idea should be recognized as contributions to knowledge because they help narrow the possible explanations and indicate which paths should not be followed. However, such work can be difficult to conduct and publish, meaning scientific training and academic systems do not always give it the recognition it deserves.
The discussion also addressed the difference between individual and systemic failure in education. Roseanne, an Imperial student, said that educational systems can create individual failures, while individual setbacks can also reveal wider institutional problems. She emphasized that making a mistake does not necessarily mean that a person has failed, particularly when the experience leads to improvement.
Ryan added that the effects of failure are not equally distributed. Social and educational background can influence both a person’s likelihood of encountering setbacks and the consequences of those setbacks. She says educational institutions should work to reduce these inequalities and build resilience so that a single exam result is not treated as a complete measure of someone’s talent or potential.
Ryan also argues that universities should broaden the career expectations placed on early-career researchers. Although becoming a professor is often presented as the ultimate academic destination, she notes that only a small proportion of researchers will follow that path. Imperial has established a research institute for early-career researchers to help them consider careers in industry, policy, publishing and the public sector from the outset, rather than viewing them as second choices.
Entrepreneurship offers another way to create a safer environment for failure. Through Imperial’s Enterprise Lab, students can develop ideas, work with external entrepreneurs and take part in mentoring programmes and competitions. These activities are extracurricular and do not affect academic grades, allowing students to test ideas without the pressure of formal assessment.
Ryan says this setup can make failure a positive outcome when an idea does not work, provided participants learn from the experience. The aim is to create a space where students can try something, fail quickly and carry the lessons into their next project without academic consequences.
She contrasts this approach with what she describes as different attitudes toward repeated failure in start-up culture. Entrepreneurs in the Bay Area often discuss companies that failed before describing a later success, and previous failures do not necessarily prevent them from receiving funding for another venture. In the United Kingdom and Europe, Ryan says, investors place greater emphasis on a successful track record, which can make founders feel they have only one opportunity to succeed.
For scientific innovation and entrepreneurship to develop, Ryan argues that institutions need to become more tolerant of repeated attempts. That would require investors to accept greater risk, while also helping individuals develop the resilience to try again when an experiment, company or idea does not work.
Sources: Nature.