Shiro Takagi

On Research

I’m an independent researcher in machine learning. In this article, I will write about my current thoughts on the problem of the research community. I would very much appreciate feedback from researchers.


One of the reasons I decided to become an independent researcher is because I want to do research in the style that I think is best. Many researchers have various ideas, but it is difficult to realize them due to many restrictions. I would like to experimentally launch a prototype of a new research style from a standpoint independent of the existing research system and not bound by any restrictions.

A person I respect once told me that whenever there is a problem, we should not attribute the cause to a person, but aim to improve the system. All of the following statements are written with that intention. I believe that when someone does something in a system, there is always a structure behind it that induces that behavior. I hope to provide a draft of the proposal to think about how to design a system that results in better scientific activities.

The ideal research environment in the sense that “anyone can do research in good mental and physical health, regardless of their social and economic backgrounds and circumstances

The ideal research environment for me is one in which anyone in good mental and physical health can do research, regardless of their social and economic background and circumstances.

I feel that not all research environments always meet this requirement. For example, in terms of mental health, the problem of harassment at universities is still deep-rooted, and the mental health of graduate students has always been an issue. Many universities address equality issues regarding race, gender, nationality, and financial power. However, we still face many problems that need further improvement.

There are many possible reasons for these problems. These may be a closed nature of academia, financial issue, or scarcity of employment.Below, I would like to share some of my personal thoughts on these issues. I would like to reiterate that what I would like to write is a discussion on how to solve these problems not by requiring individual efforts but by establishing the system to solve these problems by design.

On harassment issues

Harassment in the academic community is one of the most urgent problems to be solved. I believe there are structural factors that cause harassment.

Structural causes of harassment

The first is related to the structure of the laboratory. A laboratory in a university often consists of a PI (the head of the laboratory), staff, and students. In many cases, the PI has the final decision-making authority in the laboratory. In this sense, there is an imbalance of power. This imbalance results in several factors inducing harassment.

First, the number of PI members is much smaller than in a company. Thus, it would be easy to identify the accuser. Second, because the PI is involved in all choices, building a good relationship with the PI is vital to survive in academia. Hence, students are discouraged from reporting harassment. Third, changing laboratories in the middle of a career is a burden for student mentality and practically.

In addition to these factors, the scarcity of employment in academia also contributes to the occurrence of harassment. Because getting a job is hard, some students have to obey PI even if they feel uncomfortable with the PI. Some PI might even use the research position as a hostage to initiate harassment. The weaker the researcher’s power, the weaker the community to resist harassment.

The importance of criticism in research could also be a factor that could spur the occurrence of harassment. I believe that criticism can easily tend toward accusations and insults if one is not careful. This creates room for PIs to accuse researchers of using overly negative words.

Given these circumstances, I believe that it is not realistic to leave the solution to harassment problems to individual victims. To create a system to discourage harassment, feedback from outside of the laboratory is essential.

What matters to discourage harassment?

In recent years, with the widespread use of social networking services (SNS), collective monitoring by society of the injustice made by people and organizations has begun to function. In such a society, the more the activities inside a laboratory are disclosed, the more people criticize harassment. Therefore, designing incentives to disclose the inside of the laboratory is important to solve this problem.

It is also necessary to have an anti-harassment organization independent of the university. Many universities have established anti-harassment organizations, and each organization does its best to eliminate harassment. However, I have seen many cases in which organizations cannot strictly handle misconduct since they are within the organization. I think the existence of an independent anti-harassment organization with legal force would help us better deal with such problems.

On financial issues

The importance of correcting economic disparity

I believe that the correction of economic disparity is an urgent issue. If this is not resolved, more privileged people will join the research community. This means that the economically privileged would be supported by the non-privileged since much research is supported by taxes. This system would just enhance the economic disparity.

The problem will get worse if the government and companies prioritize the top class of talented people and allocate funds to them without correcting the situation. The talented people have already obtained funds from various sources, so providing them with new funds will further develop economic disparity. Instead, I think it would be more desirable to widely provide funds to those working hard but forced to give up research because of financial issues.

The importance of new money flow

Creating a new flow of money to academia is indispensable. To that end, it is important for academia, from the individual to the organizational level, to be more connected to society and to secure diverse funding sources.

Currently, a major source of research funding is the national budget. This has created a situation in which researchers are forced to compete to obtain the limited government budget. I believe that more constructive and positive discussions about funding will increase if the community pays more attention to how to create new cash.

I believe research relatively close to the business sector should be funded by the private sector as much as possible. Then, you can allocate more national budget to basic research. I think it is also important to ask for help from foundations and philanthropists.

In addition to the above, I am also considering monetizing academic activities themselves. Specifically, I am wondering if the research process can appeal to more people. The existence of athletes, professional gamers, and other forms of entertainment can be considered “useless” in the same sense that basic science can be considered “useless.” The same could be said of works of art. But they excite many people and form a large market of their own. Research activities also generate enthusiasm, at least among those who are involved in research. By considering what these two types of enthusiasm have in common and what is different, and what we can learn from the monetization of the former, I believe there is ample room to consider whether similar monetization is possible for research activities as well.

Of course, researchers have always been active in communicating their fascination to people who are not engaged in research activities. However, I think that this has mainly been done in a way that depends on the attractiveness of the research subject, such as the universe, history, or brain. On the other hand, I believe that researchers are attracted to the act of research itself as much as they are enthusiastic about the subject of their research. I think these attractions may appeal to people not involved in research. If it is done well, more people will be attracted to it, and research can be conducted with the money of those who are attracted to it.

I am sure many people are much more knowledgeable about business than I am. So, if there are businessmen interested in this topic, I would be happy to hear from you.

On the fact that much research is supported by tax

In the two paragraphs above, I mentioned that academia and society should be more connected. I would like to change the subject a little and explain another reason why I think researchers should interact with society.

I believe that researchers should be accountable to society. I also think the knowledge gained through research should be open to everyone as much as possible. This is because much research is supported to a greater or lesser extent by taxpayers’ money. Before I tell you why I think so, I would like to state that I also think basic research should be supported by government funds.

Why should research be supported by tax?

I believe that research is an act of taking over the knowledge of the world that people have been weaving since ancient times, creating new knowledge, and passing it on to future generations yet to be discovered. Thanks to the knowledge accumulated and shared little by little over a long time, we can quickly reach places that cannot be reached from scratch by a single person. The transmission and development of such knowledge is not something that can be done by a single person. I think it is significant that we have invented a system in which a large community, such as a nation-state, supports these activities.

Or, to look at it from another perspective, our present life is already supported by this inherited knowledge, whether we want it or not. In that sense, we have some obligation to bear a burden to enjoy its benefits. From this perspective, supporting research through taxation can be considered a natural mechanism. For these reasons, I support the support of research activities through taxes.

The importance of accountability and open access

So far, I have explained why research should be supported by the state. On the other hand, I also believe that it is not a given that tax supports research.

Providing funds for something implicitly takes away funds for something else since the funds of a nation are finite. From this perspective, it is not necessarily obvious to everyone whether investing in research is an act that takes precedence over everything else that is important.

For example, I mentioned above the idea of passing on knowledge to the future. However, this may not be of any benefit to people living in this time. Some people are so desperate to live in the present that they don’t even have enough to eat tomorrow. Given that, I think we should discuss well as a society whether we should support the inheritance of knowledge by spending their money as well.

I have also presented the idea of paying for the benefits we already enjoy. However, taxpayers cannot choose to be born into such a world. Since it is not possible to choose to live in a world without such knowledge, there should be at least some explanation for paying taxes on that knowledge.

For these reasons, I believe that researchers who conduct taxpayer-funded research are accountable to the taxpayers for the research they conduct. It is natural for taxpayers to want to know what their taxes are being used for, and it is not natural for them to not be able to provide an answer to that question. For the above reasons, I believe that researchers need to be more proactive in communicating with society and research results should be accessible to all who support them.

On the importance of a growing research ecosystem outside of existing research ecosystem

I have described my personal thoughts on research ecosystems. In many ways, I believe a research ecosystem should emerge and grow outside the current system to address the issues above.

As I mentioned above, I believe that a society in which “anyone who wants to do research” can do so is desirable. However, in the current situation, some people want to do research but have to give up their desire to pursue or remain in academia due to financial or social reasons. Some have had to leave due to harassment or emotional distress. If there were a research environment outside the current system, more of them might stay involved in research.

If there is an environment where researchers can conduct sufficient research without having to remain in academia, the mental burden of graduate students would be greatly reduced. This is because many people are worn down by the pressure to produce results to remain in academia.

As the external research ecosystem matures, I believe that the research environment in academia itself will improve. For example, those who have had no choice but to ignore what they do not agree with to remain in academia may be able to speak out. This is because they can do research even if they cannot remain in academia. Suppose that the ecosystem is further developed to the point where an external environment is preferable for certain people. This might encourage academia to proactively pursue reform from within. This is because the outflow of human resources from academia is considered a problem that affects the survival of academia. I believe this will create an incentive to create a more attractive environment to retain human resources, which may lead to a more favorable research environment in academia.

Even if the research environment outside academia cannot play the role I have described above, it is still desirable to have that system. This is because researchers can have multiple choices of research systems to participate in. If there were a more desirable research environment, conducting research under more mentally stable conditions can be possible by choosing the environment good for you. For these reasons, I believe it is preferable for many people to have a thriving research ecosystem outside of the current system.

On the ideal research environment for sharing and advancing human knowledge

I mentioned that it is good to see the development of an external research ecosystem. In this connection, I would like to briefly describe another reason I think exploring new ways of research is necessary.

I believe that when one goal is to advance human knowledge, there is no guarantee that the current general way of doing research is always optimal. For example, a person who has achieved great results in research commonly becomes a PI. However, the duties of a lab PI are to obtain funding and to do administrative and educational tasks. Just as an excellent engineer is not necessarily a good manager in a company, a person who is distinct in research is not necessarily suitable as a lab PI.

In addition, research is now generally conducted by a small group of people called a laboratory. This may lead to a lack of sharing of knowledge and know-how across laboratories. Also, competition among laboratories may lead to hiding one’s research ideas. My personal value is that these competition-based systems are not necessarily optimal in light of the objective of sharing and advancing human knowledge.

Thus, I believe there is still room for further study on how research should be conducted. Management of research on GitHub, as described in another article, may be an alternative solution to these problems. In any case, I do not think the currently common research practice has necessarily developed to be optimized for a purpose. Repeatedly confirming the objectives to be achieved and considering from scratch the best way to conduct research to achieve them is necessary.


What I wanted to convey is to think about the ideal research from scratch. A better research ecosystem may advance human knowledge at an unprecedented rate. More people may be able to conduct research with a sense of psychological security. To that end, discussing desirable research environments with everyone, including those involved in research and those not, is critical.