_{Ideally, this will be a living document. I just hit publish on this because I’m impulsive but it probably needs some serious editing. This was last updated on 1/7/2022.}

In most research settings, research scientists are the central characters. In industrial research, experienced research scientists and engineers are well paid and prestigious. In national labs, the main career path is as a research scientist. But in academia, research scientists are few and far between. In many fields, a faculty-as-manager paradigm dominates and the most experienced scientists are engaged in the management responsibilities of the modern professor. Instead, most research is executed on by graduate students.

There are a lot of difficult questions about how we structure science. This doesn’t seem like one of those difficult questions. This seems obviously bad and I’d like to do what I can to fix it. I would love to hear arguments to the contrary. Most arguments I’ve heard focus on the coordination problems or the history of the system rather than any fundamental reason that the system should work this way.

Before continuing, let’s get our terminology precise. For the sake of this article, a research scientist has a PhD-equivalent level of experience and is involved in the day-to-day execution of a research agenda. By comparison, a principal investigator is often aloof from the execution and is involved in discussion, supervision, grant writing, teaching, presentation.

In order to dig deeper on this question and works towards solving the current lack of senior research scientists in academia, it would be good to understand more about what a research scientist position is and what such a position could be and many others facets like a map of where and how these positions currently exist.

Why are research scientists essential for a productive lab?

• Productivity: Productivity increases dramatically through the early career of a scientist. Most successful research in academia is done by late stage graduate students and by postdoctoral researchers before they leave hands-on research for a faculty (management) position. By the time a graduate student defends their dissertation, they are often the world expert in their subject. What if some of those researchers could continue building their skills and executing on high impact projects instead of being consumed by the responsibilities of a management role?
• Continuity: A common refrain across academia goes “Oh, that project died because the grad student left and the knowledge was lost.” Long-term employment of research scientists who were involved in the day-to-day of lab operations would provide a reservoir of knowledge on prior projects and techniques.
• Guidance: An expert software engineer can immediately spot problems in an engineering plan that might cost a project many months of time. A lab expert might foresee problems with equipment that would otherwise require a year to identify and resolve. Having senior hands-on technical staff is critical to developing successful plans that can be executed on quickly.
• Daily collaboration: Many graduate students do almost all of their work alone, only sharing results and discussing strategy with their advisor every couple weeks. Working closely on a daily basis together produces better results and faster. A hands-on technical research scientist could work together with graduate students in the lab or write code together with the student. Watching others work is an excellent way to learn the tacit knowledge behind a field or skill.

Why are research scientists not more common?

1. Research scientists are more expensive than graduate students or postdocs.
2. Research scientists are hard to fund because our granting system and universities are built around a workforce dominated by graduate students. Granting agencies like the NSF and NIH have a lot of power over these incentives.
3. Talented scientists view academic research scientist positions as low prestige (especially compared to a faculty position).
4. There is not a path for private sector professionals to enter or return to academic research. Such people are viewed as having “quit”.

What am I doing?

I’ve been working on these problems from a few different angles. First, I’m working as a research scientist/engineer myself. Second, I put together a proposal for a type of technical research consultancy that I think would fill a helpful niche. Finally, I am working on proposals oriented at federal law makers and agency staff on how we could modify our grant funding system to employ more research scientists in academia.

What is a senior individual contributor or staff engineer?

If we look at the tech sector, in contrast to the more traditional people management career path, senior individual contributor (IC) career paths have been developed over the last few decades. These IC career paths allow people to grow their influence and responsibilities and income while remaining mostly in a hands-on technical role. Many senior software engineer ICs are in a purely technical execution role. However, if someone advances to a “staff” engineer position, the most common role is that of the “tech lead”, guiding the technical decision making and execution of a team. Other paths might involve prototyping and architecting a critical new system or diving in and solving an emergency problem. For more details on this kind of career, take a look at the Staff Engineer book.

Why is academia based on a graduate student workforce?

I’m not an expert on the history, but as I understand it, the graduate student workforce of academia exists the way it does because of the rapid growth of scientific effort in the US in the 1940s and 1950s. The demand for scientists dramatically outstripped the supply of trained scientists, so we built a workforce based on untrained and partially trained scientists. Unfortunately, that system became entrenched and we haven’t changed the basic structure since then.

What do you have against graduate students?

I have nothing against graduate students! An apprenticeship period seems like an essential stage in training for most scientists. But, as is, many graduate students in fields dominated by the faculty-as-manager paradigm don’t apprentice for a master so much as they walk blindly in the dark while having a meeting every other week about which dark alley to walk down this time. I would’ve benefitted a lot if I could have worked closely with experienced scientists working on similar topics. I don’t mean the typical academic meaning of “working closely”, but actually sitting near-by on a daily basis and doing things like pair programming or asking quick questions.

Why do the best scientists not currently want to be research scientists?

I think answers to this fall into two categories:

1. Inertia and lack of options.
2. Prestige.

Imagine you’re super excited about your research plans and you’re phenomenally good at what you do. Why would you choose to stop doing hands-on research and become a faculty member? I think the good answer to this question is “Because I think I can have a force multiplier by supporting a bunch of other researchers who do the work and I think I’ll be a capable manager.” But that’s not the most common answer. I think the most common answer is something like “Becoming a faculty member is just what you do!” or “Anything besides a faculty position feels like failure.”

Status within academia is a major problem. Research scientists and software engineers are treated like second-class citizens compared to faculty members. Pay is substantially lower and more uncertain. Many universities deny “PI rights” to non-faculty. That means that a non-faculty member can’t submit their own grants or build their own team even if they are able to get external funding to do so. It’s not a surprise to me that talented scientists avoid a career path that means they have a poor and unstable salary and are arbitrarily prevented from running their own research programs.

Examples of research scientists in academia.

This is summary of institutions that employ research scientists or people in interesting academic research scientist positions. The point here is not to be exhaustive but to demonstrate some of the models that currently exist.

Many of the research scientists in academia that I have met are in abnormal situations. I’ve left most of these anonymous.

• one research scientist is mostly funded by oil and gas money because his earthquake research also touched on issues relevant to oil companies.
• another research scientist working on numerical linear algebra has continuously pieced together a few months of funding here and there for almost 15 years. These days, most of his funding comes from a large hardware company that wants him to demonstrate the performance potential of their products.
• Two friends did postdocs and were retained as research scientists when their postdocs ended because they were too essential to let go.
• a research scientist transitioned to being an independent contractor providing GPU and high performance computing support after his request to go remote was rejected. Ultimately, he worked on the same project but was paid a lot better…
• the Wofsy-Munger group and the Harvard atmospheric chemistry modeling group both involve more engineers and research scientists than most and are built on collaborations between a faculty member and a “research fellow”!

There are also a variety of more “normal” academic or academia-adjacent research scientist positions:

• national labs, including the Department of Energy, US Geological Survey, etc.
• the Research Software Engineering group at Princeton. A similar organization existed at Harvard but has been reduced in scope over the last few years.
• While PIs at HHMI Jenalia are not really, strictly speaking, serving in a research scientist position, I think it’s interesting that they are required to participate in hands-on research.
• Mathematics professors write proofs throughout their careers and surgical PIs are in the operating theatre on a regular basis. Not all fields of science have a faculty-as-manager design.
• the Institute for Advanced Study is arguably a place where famous scientists are allowed to be full time research scientists.
• Many biology and chemistry labs employ lab managers and engineers.
• many fields have a faculty-as-research-scientist model as opposed to the faculty-as-manager model. This is common in math and in the social sciences. There are various reasons for this including faculty that have more department funding or a greater degree of independence on the part of graduate students. Or, for example, in the case of a surgical faculty member, it just wouldn’t make sense to not be doing hands-on work on a regular basis.

The lines between research scientist and research engineer are very blurred, but much of what we’re talking about here also applies to engineering in an academic setting. I’ve seen universities try and fail to build a software engineering team surrounding their high performance computing clusters. They paid less than a third of what a similar role would be paid in the private sector. I never got the sense that the engineers were particularly capable or motivated.

The lack of consistency in these paths is a problem in and of itself. What is a fifth year graduate student to do if they are confident that they want to be a research scientist? The only remotely consist path seems to be doing a postdoc with a lab that you want to work with long-term and then making yourself indispensable. But even that seems like a major gamble. So most grad students that don’t want faculty positions just leave academic science.

Why should we fix academia?

If there are other areas where research scientists are readily employed (national labs, non-profit labsprivate sector), why should we try to fix academia instead of just focusing future efforts on these other institutional models? Would new institutional models be better? As an aside, the Overedge catalog lists many new institutional models that are worth paying attention to.

I think these are reasonable points and I’m very supportive of new institutional models but the inertia of our current system is huge. Academic research is a little over half of all research in the US and much more than half in terms of influence because most academic research is not secret. I’m also optimistic that changes are possible. Even if those changes are small, the impact of those changes could be large. An enormous portion of academic research is centrally funded by the federal government. Changes in the incentive mechanisms set by that funding could have huge effects.

For example, the current system incentivizes graduate student funding over research scientist funding. Including graduate student funding in an NSF grant means that you can claim “broader impacts” because you are training the next generation of scientists. Graduate students are also much cheaper than a research scientist. So, you can A) fund two or three graduate students and get a boost in the likelihood of approval or B) fund one research scientist and not get a boost in approval chances. Given the extremely competitive grant process, most scientists will choose to fund the graduate students even if that’s not the best way to achieve their research goals. Small shifts in the grant approval process could lead to a lot more grants that are funding research scientist positions.

Questions that I would like an answer to

• What is the average salary for an academic research scientist? How much stability is there in that pay?
• Are research scientists ever viewed as “competition” by PIs?
• Does anyone have a good argument why academia should not employ research scientists? Almost everyone I talk to thinks that this part of the system is obviously broken. Despite that, there is slow progress.
• Why don’t more scientists return to academia from the private sector? I think this one-dimensional flux of scientists from academia into the private sector leaves academia with many blind-spots. I don’t think compensation is the whole story for why people don’t return to academia.

_{Thanks to many people for their help in thinking about writing about this. In particular, Sasha Chapin, Milan Cvitkovic, Liz Santorella and Yuri Vishnevsky have been helpful!}