‘When things don’t work out,that’s business as usual’: Scientists’ perception of setbacks and failure in experimental research

Abstract

Although experimental science is renowned for its discoveries, conducting research often involves dealing with problems and setbacks in the research process. Such research endeavors are often marked as failures, even though they have produced new knowledge. This paper explores researchers’ perspectives on these issues, shedding light on how they navigate setbacks in the research process. By conducting participant observation and qualitative interviews within two research groups, one in biochemistry and one in optical physics, our study reveals that experimental scientists tend to reject the notion of failed research when faced with unexpected results or unanticipated challenges. Instead, they recognize these instances as learning opportunities rather than failure as an inevitable outcome. However, the concept of failure is applied selectively, particularly when recognition for scientific efforts is lacking. Using Bourdieu’s framework, we illustrate the nuanced categories and classifications that experimental researchers use to understand and articulate their experiences of failure in science.

On problems, setbacks, and failures in research

Recent discussions in social studies of science have highlighted concerns regarding the non-publication of negative findings, non-working experiments, and exaggerated promises of success in research proposals, all of which relate to how failure is addressed in science. This has prompted various efforts to conceptualize failure within scholarly work. Some focus on failed research endeavors, as opposed to successful ones (Lehmann et al., 2022; Philipps et al., 2024), often measured by (high-ranking) publications, numerous citations, and new discoveries (Fam and O’Rourke, 2021). Others study the implications of failure in academic career paths (Bearer and Molloy, 2022; Kraimer et al., 2019). Yet others discuss failing in the research process itself: experiments that do not run as planned, instruments which do not work as they should, and results that are less significant than expected (Hon et al., 2009a). Notably, failing in the research process is discussed more ambivalently, resulting in proclamations of ‘acceptable failure’ (Firestein, 2023: 417) or ‘successful failures’ (Gross, 2023: 405) when failing is seen as an integral part of research processes and has to be accepted as such (Delamont and Atkinson, 2003). This perspective on failure is particularly at odds with its conception in everyday language, where it is linked to moral assumptions and accusations (Carroll et al., 2017: 14).

While most of the research on failure in science approaches the topic in a theoretically conceptual way, there is a distinct lack of empirical data that provides insights into researchers’ perception of failure in its various settings regarding their research work. Even though there is a comprehensive body of literature offering conceptualizations of failure in science ‘as an etic, analytical category’ (Carroll et al., 2017: 3), it is still unclear what scientists mean by failure ‘as an emic, lived reality’ (Carroll et al., 2017: 3). How do scientists perceive research that does not run as expected, proves ineffective, or yields no usable results? How do they categorize and classify non-functioning experiments, running into dead-ends, or the lack of scientific recognition for their work? Do they refer to the same conceptualizations of failure regarding research processes and concluded research projects?

To gain more empirical insights, we conducted a qualitative study within two research groups in experimental sciences, inquiring about failures in their research work by participant observation in the laboratories and subsequent interviews. The data collected shows that most researchers reject the idea of failed research processes when projects do not run as planned and objectives are not met. Notably, they did not categorize research efforts that became apparent as solvable problems or minor setbacks as failed. Instead, they emphasized the learning effects of such instances. However, the researchers did not completely discard the classification of failure, but they applied it in settings like when they struggled to gain recognition for their research efforts. It turned out that the researchers maintain very specific and distinct categories and classifications for when things go awry.

The following section begins with an overview of concepts regarding failure in science and failing during the research process. Subsequently, we apply Bourdieu’s framework of categorization and classification to our research focus. We then turn our attention to our study of research groups in biochemistry and optical physics, examining how the researchers describe problems and setbacks in their research processes, as well as failure in a lack of scientific recognition, and how they perceive, categorize, and classify these issues.

Conceptions and findings about failure in science

For nearly three decades, science studies have been increasingly concerned with failure in science (Hon et al., 2009a; Horowitz and Janis, 1994; Jungert and Schuol, 2022b). Existing research provides different perspectives on failure. One of them focuses on research processes and examines how they proceed, which problems occur, and how researchers adjust their research processes accordingly. From this viewpoint, failing refers to whether errors, mistakes, or something unexpected occurs during the research process. Another perspective takes into consideration the failure of whole research projects or lines and tasks of research, as well as the question of whether they realized stated objectives and expectations. It also takes into account the consequences that such failure has for the recognition of the scientific work performed. In this paper, ‘failing’ and ‘failure’ are therefore differentiated as done in Kurunmäki et al. (2023) wherein ‘failing’ means the process of something going amiss, and ‘failure’ refers to an ‘inevitable outcome’ (p. 163).

Failing as part of the research process

This understanding of failure in science puts the research process in the center and starts from the observation that research ‘does not always do what it is supposed to do’ (Fujimura, 1987: 271). Such circumstances produce uncertainty about the course of the research process and doubt that ‘decreases doability because it inhibits researchers’ abilities to plan ahead’ (Fujimura, 1987: 276). Accordingly, ‘much work is carried out on ad hoc bases’ (Fujimura, 1987: 276). In this regard, Hon et al. (2009b) speak of research that goes amiss. They argue that errors and failing occur in multiple contexts (related to the abilities of researchers, the functioning of technical devices, and the nature of the objects of research themselves), shape research processes, and therefore play an ‘epistemologically productive role’ (Hon et al., 2009b: 2). Similarly, Clark and Sousa (2020) say that failing occurs in ‘every stage of the research process’ (p. 1): the research design, the methodological frameworks, the data collection and analysis, and the community engagement.

In a slightly different vein, Knorr-Cetina (1981) describes scientists as adapting their objectives at each step of the research process to the direction in which the investigation is moving, trying to make the experiments work by making reasonable modifications of the research process. Firestein (2016) similarly speaks of scientists’ efforts to get research projects to work as ‘farting around’, as ‘dabbling, fiddling around, puttering, tinkering’ (p. 119). Young (2019) finds it a ‘serendipitous failure [when] things don’t go as planned’ (p. 368). To her, this kind of failing represents ‘one of the more common types of failure in research’ (Young, 2019: 368).

From this perspective, failing is part of a normal research process (Alexander, 2023). Firestein (2016) conceives of modifications during the research process as ‘a practice, where it soon becomes apparent that failure is the most common outcome’ (p. 66). Nunes et al. (2022) argue similarly: ‘failure is hard-wired into the scientific method’ (p. 24). To put it plainly, the conclusion to be drawn from these concepts of failing is that it is a part of science and that failure is the greatest teacher (Webber et al., 2022) because in science, ‘failures are as informative as successes, sometimes more so, and, of course, sometimes less so’ (Firestein, 2016: 18).

Failure due to unachieved research objectives

In this view, failure manifests as the inability to accomplish the objectives set in the initial research proposal. It rests on two premises: First, failure is assumed to be the opposite of success. In this understanding, it is grounded in the ‘inseparability of success and failure’ (Lehmann et al., 2022: 237). Second, this notion assumes that scientists set out with definite research objectives and clear expectations that should be met by the end of the research process. Consequently, failure means that the intended aim of the research could not be achieved and that requirements remained unfulfilled. Young (2019), for example, combines both assumptions, defining failure as ‘a lack of success, but also as a “nonoccurrence of something due, required or expected”’ (p. 366). Lehmann et al. (2022) share a similar understanding: ‘When researchers do not achieve the goals of their research process or research programs’ (p. 239), then they fail. According to Jungert and Schuol (2022a), failure requires a goal, an alignment, or a willful intention.

The concept of failure as the opposite of success at achieving goals can be limiting, as it makes failed research seem to be deficient and unfit for further exploration. Pushing this concept too far would not suit science, as it is well known that ‘without failure there can be no discovery, no new theories, no new revolutions in thought’ (Young, 2019: 366). For the same reason, O’Rourke and Fam argue that researchers should ‘harness failure and make it a productive input into an epistemic process that enhances prospects for understanding’ (O’Rourke and Fam, 2021: 6).

A more productive approach may be to adopt an understanding of failure in which the occurrence of results that are wholly unexpected and do not align with existing research is not necessarily indicative of failure. Such results sometimes actually represent novel discoveries, which facilitate the construction of new research fields. Firestein (2016) advises that one ‘think of [failures] as discoveries’ and see that ‘failing is a part of succeeding’ (p. 9). Young (2019) also identifies a form of failure that ‘allows for the circumstantial discovery of counter-intuitive findings’ (p. 368). Altogether, the literature leaves a rather mixed impression of this concept of failure in science, with failure being understood as the opposite of success yet also as an avenue to success. This ambiguity arises from the logic of science, in which counterintuitive and not-easily-explicable outcomes can pave the way to great new scientific findings.

Failure as a lack of recognition for performed scientific work

The concept of failure as a normative judgment of unachieved goals raises the question of its consequences for the recognition of scientific work performed. We already know from earlier studies, such as Latour and Woolgar’s (1979) Laboratory Life, that day-to-day experimental research differs in specific ways from the outcome it produces in the form of scientific publications. Although the omission of failed research from scientific publications is a well-known problem (e.g. Springer Nature, 2025), relatively little research has been conducted on the consequences of failing and failure for individual scientists, and how they perceive these consequences. There is a consensus that failed research limits opportunities to publish and to gain reputation (Lehmann et al., 2022), as well as leading to social insecurity, especially for researchers on fixed-term contracts. ‘In times when their employment security is undercut by organizing research in ever more competitive and temporary projects’ (Sigl, 2016: 348–349), failure can become a threat to an academic career. In Sigl’s (2016) study, late PhD students or early postdocs mentioned that ‘imagining a failing experiment often coincided with imagining the end of a career’ (p. 354). They find it particularly threatening that they are not ‘able to produce any valuable output’ (Sigl, 2016: 355) from this research, which is why they lack publications.

Depending on the composition of the scientific staff and the funding models, different strategies are developed to reduce the social risks of failure, such as preventing young researchers from working on projects with a high risk of failure (Sigl, 2016). Timmermans (2011) reports on another strategy when scientists put wanted results in a new context so as to turn them into publishable results. This is seen as the only way to gain recognition for the performed scientific work. Researchers thereby ensure that negative results are turned into ‘reputational impact’ (Timmermans, 2011: 551). Motivated by the ‘search for recognition’, they transform ‘failure into success’ (Timmermans, 2011: 567) by modifying the initial research project. However, there is no research on how scientists themselves perceive failure to receive recognition for their work and how they normatively judge this failure.

In fact, various scholars connect failure with certain socially shared norms. O’Rourke and Fam (2021) understand failure as ‘relative to a norm or ideal’; hence, ‘what constitutes an outcome as a failure is that meeting the norm is desired, intended or expected, which means that the effort has led to disappointment’ (p. 4). Similarly, Alexander (2023) argues that failure can be understood as a ‘normative judgement’ (p. 9), and Appadurai (2016) asks ‘who is authorized to make it [and] who is forced to accept the judgment’ (p. xxi). In this regard, Kurunmäki et al. (2023) introduce the concept of different failure regimes entailing ‘the sets of practices that form and fashion the idea of failure itself’ (p. 163).

However, these norms and regimes are based on ideas from outside the laboratory about how research is conducted and are thus unrelated to what experimental scientists personally consider failed research. Such acknowledgment (and revocation) processes do not provide insights into how scientists deal with the various problems, setbacks, and failures they encounter in their research. We ask, how do they perceive these events, and when do they problematize them?

Categorization, classification, and instances of consecration

Our study examines all three identified settings of failed research: malfunctioning processes, unmet aims and objectives, and a lack of recognition for research work performed. Previous studies have typically addressed only one or two of these settings (e.g. Firestein, 2023; Lehmann et al., 2022; Young, 2019). Furthermore, analyses of failed research often remain descriptive, omitting the experimental researchers’ perceptions and construing failure as merely the opposite of success. Such a narrow conceptualization does not fully capture the researchers’ perception of failure. To address these issues, we draw on Bourdieu’s (1990) concept of the ‘logic of practice’ to apprehend processes of perception, evaluation, and classification. Bourdieu posits that the perception and interpretation of social practices stem from individuals’ own categorizations and classifications, rather than from predefined theoretical conceptualized definitions. This approach emphasizes valuing individuals’ perceptions and judgments, as these are relevant to how individuals react and act, thus creating social practice.

Applied to our study, we thus examine how researchers categorize and classify problems and setbacks in research, and how they interpret failure. Thereby, the researchers explain and justify how they deal with failings and failures in their research practice. In order to ascertain these categorizations and classifications, it is imperative to transcend the ‘preconceptions and presuppositions’ embedded in quotidian discourses and to eschew the ‘logic conformism’, which characterizes many theoretically conceptualized definitions (Bourdieu, 1998: 136). Instead, sociologists have to provide individuals with the opportunity to articulate their unique perspectives, thus rendering their actions comprehensible (Bourdieu, 1993: 915). The methodological instrument for such an approach, as Bourdieu suggests, is to ask, to interview the individuals about their perceptions and judgments. In the context of our study, this means investigating the researchers’ categorizations and classifications of failure and failing and analyzing the perceptions and interpretations inscribed in them. Given the highly performative nature of failure and failing in everyday discourses, it is crucial to examine which setbacks researchers do not categorize and classify by these terms. This entails a careful analysis of how researchers designate and interpret these challenges and setbacks.

In addition, we adopt Bourdieu’s (1996) concept of ‘instances of consecration’ (p. 225) in order to examine failure as a lack of recognition for scientific work performed. This approach does not entail the reconstruction of researchers’ perception of setbacks and the failure of their own work. Rather, it focuses on how they perceive and categorize the instances by which their research performance is classified. Bourdieu (1996) introduced the concept of instances of consecration in his study on rules of art in the field of science, where ‘true specific apparatuses of consecration’ (p. 49) exist. We apply this framework to evaluate the instances in which performed scientific work receives recognition.

A characteristic feature of the instances of consecration is that they acknowledge scientific work as worthy and award scientific recognition to the performers – in our case, the scientists. The instances of consecration include the academic graduation process for conferring doctoral degrees, postdoctoral qualification for a professorship, as well as various forms of scientific quality control, such as the peer-review process and all other evaluation procedures. Academia’s instances of consecration also encompass the different modes of recruiting and appointing candidates to professorships. Finally, there are the academic award ceremonies, such as the bestowal of science prizes, appointments to academies, and procedures such as the inaugural lecture. These procedures contain criteria for both the recognition of scientific performance and the rejection of performance as unworthy of recognition (Barlösius, 2024). They define which scientific work is recognized as a success and which is not – the latter therefore being considered a failure.

In this study, we focus on experimental researchers’ perceptions of failure, both in terms of their research process and the instances of consecration that give them scientific recognition for their research performance. We ask how they categorize and classify failure in these two different yet interwoven settings and when they thus begin to problematize failure.

Data and method

Our research rests on data collected by participant observation and subsequent interviews with scientists from two highly productive and recognized research groups within the same German university. We examine and compare the scientists’ experiences with and their handling of unforeseen events in the research fields of biochemistry and optical physics. Both teams engaged in hands-on laboratory work; thus, frequent encounters of setbacks and problems in their research processes were to be expected (Rheinberger, 2009). Prior research suggests differences in how physicists and life scientists perceive and categorize failed research, with the former being more invested in explaining the reason behind it, and the latter adopting a more pragmatic stance (Fochler and Sigl, 2018). The comparative approach was therefore chosen to sensitize us to the characteristics of each research field, which we examine through the lens of the other.

The group of optical physicists researched the interaction of light with various materials, specializing in theoretical as well as experimental physics. The biochemists focused on particular bioprocesses to set up a production line for delivering a new product with possible applications in pharmacy, among others. The physicists’ team varied in size, generally encompassing around 25 members, including a mix of professors, postdocs, PhD candidates, and undergraduates, fluctuating as members joined or left. Conversely, the biochemists’ group operated with a more stable core of eight, amid a dynamic laboratory environment shared with other researchers, leading to an ebb and flow that peaked around 40 individuals. Both research groups were internationally mixed, with English and German being the common languages used in the labs. ¹

From October 2020, ethnographic fieldwork commenced in the optical physics group, extending to the biochemistry group in April 2022. Our methodology involved participant observations and ethnographic interviews, executed by one researcher who immersed themselves in the scientists’ daily routines. This approach, in line with Breidenstein et al. (2020), enabled us to develop comprehensive ethnographic protocols, which were collaboratively analyzed during scheduled breaks between observation phases. To supplement this data, we conducted 20 semi-structured interviews (Flick, 2021) with the members of both research groups and associated researchers from other German universities (see Table 1), asking about their experiments, their educational and research background, their experiences with and handling of unforeseen events during their research, and their understanding of success and failure. The audio-recorded interviews were fully transcribed. All participants in our study were informed about our research in the groups, and they signed consent forms.

Table 1.

List of all interviews and selected interviewees’ characteristics.

Research groups	Gender		Academic degree
	Female	Male	Undergraduate students	Graduate students	Postdocs	Professors
Optical physics	3	9	0	5	3	4
Biochemistry	1	7	2	3	1	2

The authors jointly interpreted the observational and interview data in several meetings. In our analysis, which conformed to Flick’s (2021) thematic coding procedure, the observation protocols were first compared with the researchers’ corresponding interview statements. This approach sensitized us to what we saw and noted and what our interviewees emphasized. It also enabled us to conceptualize how research processes in the observed groups evolved, as well as their specifics.

In a first analytical step, we examined the extent to which the interviewees perceived, described, and assigned failures and setbacks to similar research settings. It turned out that they consistently evaluated research challenges in similar ways, that is, neither arbitrarily nor situationally, often using the same (or similar) terms and ascribing the same performative meaning. This observation holds across the two disciplines in our sample despite their differences in daily research routines and lab work. Even across different professional levels, researchers exhibited similar perceptions and interpretations, whether or not they were well versed in handling failure and setbacks in their research work. ² Hence, we suggest that researchers perceive, categorize, and classify setbacks and problems during the research process in remarkably similar manners. This is reminiscent of Fleck et al.’s (1979) concept of thought collectives, which scientists from both fields in our study seem to form regarding their perception of failed research.

To explore this further, we assessed the interviewees’ perception of problems and setbacks in their research, focusing on how they categorize and classify the different types of research going awry. We then analyzed the researchers’ classifications, examining how and when they consider something as failed. For categorizations, we focused on the descriptions of the various problems and challenges and how the researchers distinguish these distinctly. For classifications, we examined the researchers’ evaluative judgments when things went wrong. With emphasis on the researchers’ own perspective, we adhered to their terminology to precisely map out their categorizations and classifications.

Researchers’ perception of setbacks and failures

In the following section, we present our findings structured into three different categories of ‘failed’ research according to the distinction made by the researchers interviewed: first, the confrontation with problems and setbacks during the research process; second, the encounter of dead-ends or limits of possible research; third, the lack of recognition for the research work carried out. In line with this systematization, we introduce the categorizations and classifications that we developed on the basis of the interview material and participant observation.

Encountering problems and setbacks in research processes

Observing scientists in their laboratories, we gained insights into their daily research work and noted various occasions on which experiments did not work out as planned. In one research project, for example, an undergraduate and a postdoc in the optical physics department jointly worked on a laser device. The task of the undergraduate was to set up a laser and use manufacturing processes to verify whether it worked; the postdoc supervised and occasionally provided assistance. In a ‘crisis meeting’, according to the postdoc, the undergraduate reported problems he had had with the manufacturing process the week before, even though – as he said – ‘everything had actually been set up correctly’. In the laboratory, they went through the experiment, first inaugurating a test phase and recalibrating the laser when necessary. Shortly after the experiment commenced, the device aborted the test phase, indicating that something was incorrect and in need of manual adjustment. The student and the postdoc pondered why it had not worked and adapted various parameters, but the error message persisted. They switched to another device chamber, where the test phase concluded without incident. However, more problems occurred during the subsequent experiment. First, the pane of the chamber was too dirty for the laser to work properly and had to be replaced by engineering staff. Then, the rotation of the target malfunctioned, leading the postdoc to break off the experiment temporarily. On the next day, it turned out that only a fuse had blown. They replaced it quickly and were able to resume the experiment.

Such technical malfunctions were not restricted to experiments by optical physicists. We became aware of similar ones during our stay in the biochemistry department, where, for instance, an undergraduate student and a PhD student were working together on animal-cell cultivation using an automated device. Normally, the apparatus’ robotic arm would take samples and add substances when necessary, effectively replacing the human hand and minimizing the risk of contamination. While adding substances during one of the cultivation sequences, however, the robotic arm jammed, pausing the whole process. The two students immediately started trouble-shooting, going through the most recent similar instances of the device’s malfunction (e.g. the previous day, the previous weekend) and trying out past solutions. They tried reinitiating the program, occasionally suspending the procedure and even completely turning off the device in order to fix the problem, to no avail. The robotic arm even unlocked the lid that sealed one of the experiment’s 24 cultivation vessels. The lid had to be replaced by hand, a maneuver that contaminated the culture in that vessel.

In both cases, the researchers had to contend with typical technical problems that disrupted their experiments. Other sorts of interference were epistemic (e.g. anomalous measurements), organizational (e.g. search for devices), and social in character (e.g. other students needing help). In all these incidents, we found optical physicists and biochemists pragmatically trying to find the cause of the problem and solve it. In some cases, senior scientists were consulted to understand the roots of the trouble. But most of the time, experimenters (usually undergraduates, master’s-degree students, and doctoral students) handled such setbacks in a pragmatic manner to get the experiment running, even though it took a lot of time and effort.

The interviewees also mentioned additional disruptions, setbacks, and unforeseen events relating to their experiments. The optical physicists and biochemists did not have to go far to come up with a story about such issues during their research process. A PhD student in the optical physics department, for example, gave the following account of his latest research project:

Shortly after I began, the laser was not functioning; [. . .] not exactly the laser itself, but there are many devices involved [. . .]. The laser had a lot of technical problems that delayed me for at least a year and a half, I would say, if not longer. And when [. . .] the laser was halfway under control, things really started to look up. Then I was able to deal with the experimental problems. (I_PH03_mDoc: 535–562) ³

Optical physicists often spoke about problems with laser devices and how they affect the research process. By contrast, scientists in the biochemistry department mentioned contamination as the main recurring problem. We heard many stories about unforeseen events like the following one:

Yes, for example, we very often had contamination during the first cultivations. In other words, we prepared the experiment with a preculture for two weeks, then prepared the main culture for another week, and then went into the experiment [. . .]. You buy an item from project partners, or you get one that is sterile. In your naivety, you don’t check it. You install it; you see that it was probably not sterile and [that] you have destroyed the whole process. You can’t just repeat it quickly; you need all this preparation time again. (I_BC04_mDoc: 275–283)

In subsequent reports, the interviewed scientists mentioned a variety of other problems and things that can go wrong during the research process. In other words, their descriptions confirmed what we observed in their labs: the ubiquity of problems and setbacks in experimental research. For the scientists, this meant much trial and error, which took time and effort.

Principally solvable setbacks

A closer look at how the interviewed optical physicists and biochemists described these unforeseen events and problems reveals that they did not speak of failing, as one would expect from everyday preconceptions, but also from many theoretical concepts. Instead, their statements contained expressions such as ‘does not work’ (I_PH01_fPost: 874; I_BC01_mPost: 269), ‘is not functioning’ (I_PH03_mDoc: 535; I_BC02_mProf: 814), ‘is not running well’ (I_PH12_mProf: 385; I_BC07_mBA: 261), or ‘goes wrong’ (I_PH07_mProf: 966; I_BC03_fBA: 413). What, at first glance, appeared to be moments of failing were actually categorized by the interviewees as malfunctioning processes. They described procedures that seemed to follow a certain course but ceased progressing at some point. They did not speak about the entire research process as ‘failed’ but rather about a procedure that was not doing what was expected of it and was therefore ‘not working right’ (I_PH10_fDoc: 426; I_BC08_mProf: 290). This viewpoint includes technical equipment that ‘breaks down’ (I_PH03_mDoc: 749; I_BC07_mBA: 231–232), ‘malfunctions’ (I_PH06_fDoc: 463), or proves ‘unstable’ (I_PH05_mDoc: 267) and cell cultivations that become ‘contaminated’ (I_BC05_mDoc: 287).

When the expectations of the scientists had not been met and ‘project goals [. . .] not achieved’ (I_PH04_mPost: 1217–1218), scientists still refrained from categorizing these occurrences as ‘failure’. Instead, they called the research outcome ‘poor’ (I_PH05_mDoc: 485–486; I_BC07_mBA: 261–262) or ‘wrong’ (I_PH02_mDoc: 829–830). Within this framework, one can return to the procedure, search for the problem, and perhaps fix it. These incidences were frequently referenced as ‘problems’ (I_PH07_mProf: 832; I_BC07_mBA: 581), ‘difficulties’ (I_PH12_mProf: 141; I_BC06_mDoc: 470), ‘challenges’ (I_PH03_mDoc: 751; I_BC01_mPost: 152), ‘puzzles’ (I_PH04_mPost: 496), and ‘setbacks’ (I_PH09_mProf: 929). Thus, the researchers categorized such occurrences in a way that presents them as principally solvable.

In a nutshell, the researchers’ categorizations indicate that they are used to talking a great deal about problems and setbacks and that they have a distinct understanding of how to interpret them: Such things simply happen during the research process. Furthermore, if problems or malfunctions occur in research processes, they are seen as impediments that cause delays and require solutions but do not mean complete cancelation and loss.

Learning rather than failing

In this section, we focus on the scientists’ interpretation of these incidents, that is, how they classify them. Their reasoning shows that research, to them, generally implies setbacks and problems, which they classify as learning opportunities that affect future research.

First, the interviewed researchers emphasized the ubiquity and normality of the problems and setbacks portrayed. For example, when we asked a PhD student about his experiences with unforeseen turns in research, he emphasized: ‘When things don’t work out, for me, that’s business as usual. It’s normal, so that’s nothing out of the ordinary now. For me it is quite normal. For me, it is not a failure’ (I_PH02_mDoc: 687–689). Similarly, the professor in biochemistry explained, ‘This is actually our daily bread to carry out experiments and see that what we had in mind doesn’t come out at all’ (I_BC08_mProf: 275–276). Second, negative results, setbacks, and solvable problems are aspects that interviewees associate with the chance to learn something about their research subject. They do not give up when confronted with problems. Instead, dealing with such challenges is described as a ‘joy of puzzling’ (I_PH01_fPost: 952), as solving the problems and getting things to work again. As stated by an undergraduate in biochemistry, ‘When I realize, “Yo, that’s not working,” I optimize something and it works after all. So you learn from your mistakes bit by bit’ (I_BC03_fBA: 410–412). In this quotation, she described the process from the moment of experiencing an experiment that does not work to the step of optimizing it until ‘it works after all’. This optimization is essential and classified as a learning process. The classification as learning is also true for the optical physicists:

In the meantime, I’ve come to the realization that it’s important to just do it and to run in the wrong direction for a week. There’s an enormous learning effect in that, and you can’t take that kind of learning effect away from anyone (I_PH04_mPost: 1071–1073).

These interviewees described how the learning effect achieved through dealing with problems and setbacks is an important part of the research process and cultivates a valuable skill that scientists acquire. Such statements from the interviewees lead us to suggest that from their perspective, the classification as failed would not correctly frame such research problems. They emphasize that these problems are solvable and that experiments are principally doable (Fujimura, 1987), and on this basis, they classify them as a normal part of research processes, which opens learning opportunities.

Research limits and dead-ends as acceptable failure

Of course, not all problems in research processes can be solved. There are occasions when scientists cannot continue their research despite having tried different approaches:

You always come to a certain point where you first have a limit. And then you have to ask yourself, ‘How do I go on now? Or is this now somehow the end?’ So, I really get absolutely nowhere here. Of course, it can also be that you try something that just doesn’t work. Like now, for example, this absorption, which I wanted to measure at the beginning, where I just never saw anything. At a certain point, of course, you have to say, ‘I think I’m going to abandon this. I have to do something else’. (I_PH03_mDoc: 660–667)

A postdoc in biochemistry recounts a similar incident:

So, about my doctoral thesis: I did a project start for the first twelve months, which didn’t play a role in my doctoral thesis in the end. It just didn’t work at all. I was working with genetic methods to find new targets for [a certain organism], and I set up a method. We tested more than 20,000 samples, and I didn’t have a single activity that I could have worked on. And then after twelve months I turned to a new topic. (I_BC01_mPost: 313–318)

These examples show scientists experiencing an impasse, when they could not complete or had to abandon their research due to insurmountable problems. Upon exhausting all other research possibilities, they describe it as ‘reaching a limit’ (I_PH03_mDoc: 66; I_BC04_mDoc: 288). The importance of exploring every avenue before accepting limitations is stressed further by the doctoral student of optical physics:

I think I’m going to abandon this. I have to do something else. But you can’t give up too soon. You first have to make sure that you’ve tried everything you can think of or that’s somehow doable, and if you think you’ve tried everything and it still doesn’t work, then maybe you have to make the decision not to continue. But you also have to try it first. (I_PH03_mDoc: 666–672)

Or in the words of an interviewed biochemist:

Throwing everything away and then not even thinking about what could be improved or what the problem could have been – that would be inadmissible for me. I simply could not stop an experiment immediately and quit trying altogether when there is still so much potential. (I_BC07_mBA: 459–466)

When, after trying everything, it becomes clear that the set research aim cannot be met with the existing methods, scientists have to consider whether continuing this particular research is reasonable. In these incidents, they uttered phrases which evoke associations with finality and nothingness, such as ‘I really get absolutely nowhere here’ (I_PH03_mDoc: 663, our italics), ‘it just didn’t work at all’ (I_BC01_mPost: 314–315, our italics), ‘[we] never saw what we wanted to see’ (I_PH08_mPost: 311, our italics), or ‘ultimate end [of the research project]’ (I_PH03_mDoc: 884, our italics). Thus, using again specific wording to mark this stage they see as a terminal point, the interviewed scientists categorized such research experiences as research limits or dead-ends.

It is important to note that the category of research limits is distinct from those principally solvable problems in research processes, as described in the preceding sections. Rather, research limits and dead-ends pertain to the entire research endeavor, suggesting the possibility of ending it.

If our interviewees recognized that they had hit the prevailing limits of doable research and possible epistemic gain, they switched to the classification of failure. According to a doctoral student of biochemistry, ‘But a true failure would actually only be if nothing at all, if you couldn’t even get your experiments through somehow, so actually such a real extreme case’ (I_BC05_mDoc: 502–504). He detailed how reaching a limit of possible research is tantamount to failure to him, establishing real failure as an ‘extreme case’. This classification was also applied in the optical physics department. As a PhD student stated: ‘If, of course, you then just see, okay, this doesn’t work and we can’t make anything of it and it’s all trivial anyway and has nothing of significance, then you can also see that as a failure somehow’ (I_PH02_mDoc: 840–843).

Examining the interviewed researchers’ approach to classifying this type of failure reveals their use of descriptors to rationalize that persisting with the research project is unfeasible. Terms such as ‘failing for good’ (I_PH02_mDoc: 889–890), ‘failing fundamentally’ (I_PH03_mDoc: 939), ‘real failure’ (I_BC05_mDoc: 502), and ‘ultimate failure’ (I_PH03_mDoc: 884) were common among the optical physicists and biochemists of our study. It reflects their acknowledgment of the project’s infeasibility and the exhaustion of all viable research avenues. Hence, this classification of failure serves to explain and support their decision to discontinue their study, reminiscent of Firestein’s ‘acceptable failure’ or ‘real failure’ (Firestein, 2023: 417), a failure that requires no justification, as nothing went wrong and is thus as scientifically valuable as success. Indeed, our participants shared the same understanding of this type of failure, to which they explicitly did not assign any negative judgment. Some interviewees, upon reaching an impasse in their research, still highlighted the scientific insights and understanding they gained through the process:

In this respect, of course, the intended aim is often not achieved. But that doesn’t mean that no epistemic gain was made, and I wouldn’t perceive it as a failure. (I_BC08_mProf: 478–480)

I mean, ultimately, the realization that it doesn’t work is also a scientifically valuable epistemic gain, so failure is somehow the wrong word in my opinion. (I_PH05_mDoc: 400–402)

In conclusion, acknowledging these research limits and dead-ends could yield important scientific knowledge for both the researchers and the broader scientific community. From their viewpoint, this epistemic gain signifies that they have not truly failed as scientists. Although they have pursued every possible route, they ultimately encountered the persistent limits of current research possibilities. As a result, they regard this as an acceptable form of failure without a negative performative sense.

Lack of recognition as failure

Last but not least, a third setting of failed research emerges as the interviewed scientists themselves categorize their research performances, by pointing to the time, money, and effort they have invested. From their point of view, there is a clear discrepancy between their own perception of their research performance and the requirements for having this performance recognized by the instances of consecration:

I think if you’ve invested a lot of time and money in something because you thought it would work and in the end it doesn’t work and you can’t publish anything, then you’ve failed to some extent because you’ve invested a lot in it. Yes, that is failure. (I_BC03_fBA: 449–452)

Well, if you work on a project for three years, have nothing to publish and, at worst, can’t write up a doctoral thesis that would, of course, be the worst-case scenario. You always have to have accomplished something in order to be able to submit your dissertation. If you couldn’t do that at all, then that would be failure for me. (I_PH06_fDoc: 597–601)

In both cases, the interviewees give exemplary scenarios of research classified as failed by emphasizing the inability to show their efforts and what they have accomplished. Whether it be prerequisites for publication, academic qualification, or solicitation of projects, the interviewed biochemists and optical physicists both see the demand to deliver a great story of successful research. As a professor of biochemistry criticizes:

When you try to acquire third-party funding or grants these days, you always have to tell a great story in order to get any funding at all. And in my opinion, these stories are sometimes too big to be able to fulfill everything that these stories imply, so to speak. [. . .] because it’s simply not realistic to save the world with every project. But now, unfortunately, you always have to tell that story and claim that you’re going to do it. In other words, the stories are always bigger, often bigger than what is actually behind them. And I actually find this tendency harmful, to be honest. (I_BC08_mProf: 234–256)

A postdoc of optical physics specifies further:

And you can only get funding if you have also published. Because you also have to show that you have expertise in the subject matter and that the German Research Foundation, the DFG, doesn’t just throw its money at anyone for anything, but that [the research project] is also reasonably promising. [. . .] Especially in the early stages of your career, you haven’t achieved that much, you want to achieve something, but of course people expect you to be able to show a certain level of expertise in the field. It’s a bit of a catch-22. (I_PH01_fPost: 407–417)

Especially in the context of obtaining funding, research processes are often presented as linearly plannable and particularly promising stories of success. Beyond that, scientists are expected to demonstrate their expertise by publishing again stories of successful research to be eligible for funding, further underpinning this narrative. Nevertheless, such stories do not accurately reflect the conditions of research processes that are prone to going amiss, as we have demonstrated in the previous sections. Although the researchers interviewed have carried out their research carefully, tried out different approaches, and met the criteria for good research, they do not succeed in gaining recognition for their scientific work (Lehmann et al., 2022). They categorize their own research performance according to these criteria and therefore think that recognition would be appropriate.

If this is not achievable, they perceive the lack of recognition as a failure: They are not able to make the results of the research process visible outside their laboratory and therefore cannot be credited with the scientific achievements contained therein. They fail because of academia’s instances of consecration and the fact that only certain scientific findings are considered worthy of publication and recognition.

In the interviews, the researchers openly criticize that the practices of granting scientific recognition are confined to a specific understanding of failure and success in research, which does not correspond to the nature of research processes. While researchers categorize their performance according to how they have conducted their research and whether the criteria of good research have been met, the instances of consecration are designed to look first at the outcomes and at stories of big success. According to this criterion, they decide whether the research performed deserves recognition, for example, by publishing the work, approving the application for third-party funding, or accepting the dissertation. If the researchers do not succeed in getting recognition by the instances of consecration, then they classify this experience as failure.

In these situations, they associate the classification of failure with a negative performative sense. However, this classification also contains a critique of the instances of consecration, which is derived from the way they themselves categorize research performance – according to the nature of research: not everything works out and stories of linearly successful research endeavors should not be the norm. When researchers criticize this restricted conception of recognizable research performed, they are in line with Firestein’s comment: ‘failures and negative results don’t get funded’ (Firestein, 2023: 419).

Categorizing and classifying according to the practical logic of research

In this paper, we have examined how researchers in the fields of biochemistry and optical physics categorize and classify when research processes go wrong, when the hoped-for research aim is not achieved, and when research performance does not translate into scientific recognition. In short, when something happens that is usually considered a failure. We have focused on the researchers’ perspective and how they perceive and judge such challenges and setbacks. In this way, we wanted to find out how they orient themselves in their categorizations and classifications – on assumptions and presuppositions, on the performative sense of failing and failure, on how they describe the nature of the research practice. With this approach, we wanted to break with the ‘logical conformism’ (Bourdieu, 1998: 136) inscribed in failing and failure, which is the prerequisite for reconstructing the researchers’ perception. This ambition distinguishes our study from most studies on failure in science, which have posited a conceptual understanding of failure (Hon et al., 2009b; Horowitz and Janis, 1994; Jungert and Schuol, 2022a; Lehmann et al., 2022; Nunes et al., 2022; Young, 2019). Rather, our approach aligns with the notion of the emerging interdisciplinary failure studies as postulated by Barbera and Rees Jones, that ‘[w]hat is needed is a clearer thinking about what failure really means, a better understanding of the mechanisms that generate, reproduce and terminate it as a normal way of working of the system’ (Barbera and Rees Jones, 2024: 3).

In our analysis, we identified three distinct categories of research going amiss, which are associated with different settings in research and are matched to three different categorizations and classifications (see table 2). First, the interviewed researchers perceived setbacks and problems during the research process as a normal part of research and categorize them as principally solvable. Such setbacks do not represent failing to them. Rather, it is in the nature of research processes that something does not work. Accordingly, they classify such principally solvable problems as learning opportunities that affect future research.

Table 2.

Interviewees’ perception of varying settings of failed research.

	Setting	Categorization	Classification
Research process	Setback, problem, non-functioning experiment	Principally solvable problem	Learning opportunity
Overall research endeavor	Reaching a limit, the ultimate end	Dead end, but epistemic gain	Acceptable failure
Instances of consecration	No (high-ranked) publications, funding	Lack of recognition for research work performed	Failure

Second, as for cases in which research was not working at all, it became apparent that the researchers had encountered a dead-end. In these circumstances, it was crucial that they exhausted all possible avenues before concluding that their research process had to be ended. Only then would they have scientific proof that they have reached a limit, which cannot be overcome with the current possibilities of research. Such experiences of research limitations and dead-ends were thus classified as acceptable failure. In this regard, they emphasized the epistemic knowledge gained from research processes that do not work at all. This perception is reflected in the way they classify such cases as failures: without a performative sense of negativity, but rather by describing that the research is currently going nowhere. As with the problems and setbacks, the scientists’ perception of these incidents is keyed to the specific nature of research, which is that innovative research comes up against limits and sometimes leads to dead-ends.

The third setting refers to how the scientists categorize their research performance, and how they classify the lack of recognition of their research performance by the instances of consecration. Here, their perspective shifts from the nature of research processes to the evaluative criteria of research work, which consider only certain results worthy of granting recognition. While researchers categorize their performance according to how they have conducted their research and whether the criteria of good research have been met, the instances of consecration are designed to look first at the outcomes and then decide whether the research performed deserves recognition. If the researchers do not succeed in transforming their research performance into recognition, then they classify this experience as failure. In these situations, they associate the classification of failure with a negative performative sense. However, this classification also contains a critique of the instances of consecration, which is derived from the way they themselves categorize research performance – according to the nature of research: not everything works out, and stories of linearly successful research endeavors should not be the norm.

Implications arising from the researchers’ perception of failure

As our study expands the existing research on (conceptual) understandings of failure in science by adding empirical data, we were able to provide complementary insights into the researchers’ own perspective. Examining their specific categorizations and classifications of problems, setbacks, and failures allowed us to identify different settings of research wherein the scientists voiced their struggles with failure in varying degrees. While challenges and problems in research processes are seen as ‘usual business’ and reaching limits or dead-ends can still yield important scientific knowledge as long as all possible research avenues have been exhausted, the (lack of) recognition for the scientific work performed is when failure becomes truly problematic.

Our study shows that there is a perceived discrepancy between how the experimental scientists understand research processes and failure therein and how the instances of consecration that grant scientific recognition attribute failed research. A rethinking of the culture of failure in science, as proposed by Firestein among others (Firestein, 2023; Gross, 2023; Nunes et al., 2022; Young, 2019), would therefore have to start where the instances for recognizing research work run counter to the nature of research.

Bourdieu addressed a similar problem in the field of art, for which he developed the concept of instances of consecration. The value of a work of art is not inherent but is derived from its recognition by the instances of consecration: they produce the ‘value of the works’ (Bourdieu, 1996: 229). This is precisely where the critique of the researchers we interviewed begins: The research process itself has little significance for the valorization of scientific work. The criteria of the instances of consecration apply here, which, however, pay too little attention to the nature of research (see also Philipps et al., 2024). Even when failed research generates new knowledge, this is still not considered valuable. It would require a change in the evaluation criteria that are set out in the instances of consecration. Breaking with the normative judgment of such ‘failed’ research could generate more visibility and facilitate the recognition of scientific work performed according to the criteria of good research over stories of successful research outcomes. The focus would thus need to shift from research outcomes to research processes and the quality of research work. It remains to be seen whether and how such a shift can be achieved. Even so, this criticism by the researchers interviewed is already being voiced outside the scientific context, stipulating a ‘need to deal with the normality of failure in a threefold sense; failure is endogenously normal, ubiquitous, and morally legitimate’ (Barbera and Rees Jones, 2024: 3).

Despite offering valuable insights, our study is limited by its focus on biochemistry and optical physics within a German university context. While the lack of major differences between these fields suggests broader relevance across experimental sciences, further research is needed in more diverse, international settings, especially in non-experimental sciences. As social scientists, we would expect different perceptions of failing pertaining to our research process because it varies greatly from day-to-day laboratory work. However, regarding the lack of recognition, we would argue in favor of scientists forming a thought collective across disciplines, as we understand the instances of consecration to follow universally applicable parameters of successful and failed research. In addition, examinations across other disciplines or even non-scientific fields could yield novel insights into understanding and navigating failure in various settings. These insights would provide the groundwork for identifying the contexts in which failure is problematized and where a process of rethinking can even begin.