In an earlier interview with China Science Daily, I answered quite a lot of questions, but when they compiled it into an article, they actually only quoted two of my sentences. Here I’m posting my original answers.
1. Someone once proposed that “failure refers to an action’s result deviating from the expected goal, or to the termination (interruption) of an action that has not achieved its intended value.” How do you understand “research failure”? The relationship between failure and success is dialectical. In your view, what kind of research counts as ‘failure’? Can “research failure” be divided into several types?
How we define failure of course depends on how we define success, so we can first discuss how many types of success there are in research. If publication of a paper counts as success, then being unable to publish is failure. If completing a task or project counts as success, then failing to complete the task is failure. These are all standards measured by external criteria. For scientists or research teams, there can also be internal standards by which they measure their own aims. For example, accumulating experience, ruling out blind alleys, pursuing truth—in this sense, work that looks like failure by external standards is not failure at all in terms of honing one’s abilities or pursuing truth. So I think the key really lies in the question of the “view of success” in research. Total failure is not something worth encouraging, but some so-called failures are merely cases where no result was produced that met a particular external standard; we should support scientists in defining their “success” more flexibly according to their own circumstances.
More specifically, we can classify these apparently “failed” successes according to what they achieved. For example: work that can train a research team’s experience; work that can contribute lessons learned or data materials to the scientific community; work that can spark new research topics (such as why it failed), and so on. These kinds of “failures” with positive significance are all worth encouraging. Of course, there are also failures caused by misconduct on the part of researchers; depending on the nature of the misconduct, sometimes serious accountability is required.
2. There may be many reasons why research failure is mentioned so rarely. For example, researchers usually only talk publicly about their failures after they have succeeded. What do you think about this? In your view, what is the main reason research failure receives so little attention?
Obviously, the main reason failure receives so little attention is that it has failed, which sounds like a tautological truism—“failure,” in a certain sense, is precisely the inability to obtain a result that attracts attention. So first of all, we should recognize that it is normal for research failure to receive little attention; in fact, in any field, we pay more attention to success and very little to failure. We don’t even pay much attention to the runner-up among successes; we only pay attention to the most successful person. That is why scientists care so much about the struggle for priority. There is nothing wrong with successful people enjoying glory and attention in itself; competition and accumulation in research unfold in just this way.
The reason we feel that research failure deserves more attention is, again, because the measure of success and failure is in fact plural. Many so-called “failures” actually have positive significance. Some people—such as journal editors, project reviewers, title-evaluation committees, the public, or the researchers themselves—hold a view of “success” that is too narrow and rigid, and this leads to many outcomes that can yield positive lessons being casually classified as failures.
On the other hand, apart from failures with positive significance, some failures arising from misconduct do need accountability. Yet these too receive relatively little attention, because many research activities lack third-party oversight mechanisms. If the research process depends entirely on scientists reporting on themselves, then scientists will often report the good news but not the bad. Take the He Jiankui事件, for example: He Jiankui believed his gene-edited baby experiment had been successful, so he went in high spirits to give a report at a conference, and that is why the whole world was shocked. But imagine that if his experiment had failed—if, say, the baby had miscarried or been malformed—he would simply not have published a paper or given a report, and if he handled it quietly in private, that would have been the end of it; others might never have known that he had carried out such serious violations of research ethics.
Allowing scientists to assess success freely, and adding third-party oversight mechanisms for scientists, are not contradictory. Research governance should not consist of only one rigid, monotonous, top-down regulatory mechanism; it needs to introduce diverse participants.
3. Compared with the era of small science, what changes have there been in the characteristics of research failure itself and the impact caused by failure in the era of big science?
The most basic change is that it has become more expensive to burn through money, and the impact has also become greater. In the era of small science, research activities were often spontaneously organized by nobles or enthusiasts, and it was enough for the work to be accountable internally. But in the era of big science, large amounts of funding are often required, and the money comes from enterprises or taxpayers; some failures may also produce serious consequences (such as dangerous substance leaks, pollution, ethical issues, and so on), so researchers have a greater responsibility to report “failures” to their supporters.
In addition, in the era of big science, the integration of science, technology, and industry is closer, and many research projects need the push of capital. Thus the logic of “market competition” and “venture capital” is also introduced into science. From the perspective of “VC,” investing in ten projects and losing nine while one succeeds spectacularly is still acceptable. But many research projects within academia cannot keep up with this pace at all; if one out of ten projects fails, that may already be unacceptable.
4. In your view, is the public disclosure and sharing of research failures necessary? Which failures are valuable to make public?
“Failures” that can produce positive lessons, or that may produce negative consequences, should all be made public.
5. In your impression, are there any cases of publishing papers because of research failure experiences? Could you give one or two examples?
I know that the phenomenon of publication bias has already drawn attention in the international academic community. It has in fact affected the accuracy of “meta-analysis” of existing research, especially in fields such as medicine, biology, and psychology. But the main solution may not necessarily be to publish more papers about failures; rather, it may be things like establishing open experimental databases, so that even if failed experiments do not publish papers, the experimental data can still enter the public record.
Some so-called “failures” actually have very high publication value and may become the trigger for overturning an established theory. But if one insists that success and failure should be treated alike in the sense that both can publish papers, that is also unreasonable. So I think ordinary research-failure experiences do not necessarily need to be made public in the form of papers; they can be recorded through public databases and other means.
6. Today our country advocates “from 0 to 1” and original innovation, which may mean that researchers in China will have to experience more failures. In such a situation, what kind of view of failure do you think is beneficial for the future development of science in our country?
In fact, innovation in application, and even simple imitation, can have a very high failure rate as well. For example, if you put the most advanced lithography machine there and ask researchers in our country to reproduce it, that is obviously not original innovation from 0 to 1, but the success rate would probably be extremely low.
As for the view of failure, I have already said earlier that first of all, the “view of success” itself should be diversified—not measured only from above, but assessed from multiple starting points: the scientists themselves, third-party overseers, capital and markets, and the public.
7. Nowadays many sciences are carried out on the basis of projects and grants. In your view, if a project fails, how can the positive value of research failure be maximized? How can researchers be encouraged to publicly share their experiences of research failure?
Then we should not let too much research be based on (official) projects and grants. We should open up diversified incentive mechanisms, including but not limited to unconstrained funds given to scientists to control autonomously; funds managed by universities and research institutions themselves; grants supervised by third-party civic groups; capital introduced by entrepreneurs; and so on. Official projects have stricter requirements and more rigid goals, which is normal. Indeed, there are some research activities whose results are easier to foresee and whose plans are easier to arrange, and whose success is easier to guarantee with sufficient material support; these kinds of projects are not supposed to encourage failure, and that is reasonable. What we need is a more diversified incentive system.
Translated from the Chinese original with AI assistance. The original text is authoritative.
Leave a Reply