other

Working in Universities vs Companies

Over the past year, I’ve been involved in a number of conversations with other researchers about the relative merits of working for a university vs doing R&D work for a company. Below I set down some general observations (not recommendations!!) about this. Keep in mind that both companies and universities are very diverse, so all general comments should be taken with a large grain of salt. Working for Google is not like working for a startup, and (for junior academics) working for a US university with a high-pressure tenure track is not like working for a UK university with a relatively low-pressure probation system.

Some of the below is informed by my own experiences (I’ve mostly worked in academia since getting my PhD, but have also spent time at two NLG companies, CoGenTex and Arria), but mostly its based on experiences of friends, colleagues, and students.

Control

University academics have a lot of freedom to choose their research areas, while company R&D work usually needs to be justified as potentially providing benefit to the company (either via better products or increased sales, or via “softer” metrics such as public relations). Its worth remembering that companies can and do change their minds about what they are interested in. I have seen cases where a researcher moved to a company to work on an exciting research project, which later on was cancelled, and the researcher was told that if he wanted to stay at the company, he would need to move to a non-research software development role.

A perhaps related point is that academics are free to discuss their research with anyone they wish to interact with; indeed academics have an incentive to widely discuss their research because this improves their citations. Commercial researchers often face restrictions on what they discuss outwith the company, either because the company wants to patent ideas before they are published, or because the company does not want researchers to say anything negative about the company’s technology (eg, it can be hard to publish negative results).

Resources

On the other hand, commercial researchers usually have more resources than academics, in all sorts of ways: higher salaries, better equipment, more time to spend on research (no teaching), and also bigger teams to support their research. Indeed in NLP we are seeing that some areas (such as building very large language models) are dominated by companies, because they are resource-intensive and universities find it hard to compete with companies.

A related point is that commercial researchers often find it easier to make a difference in the “real world”, especially in the short/medium term. They have the resources and backing to create products and services which can help people. Academics of course can come up with great ideas that make a difference to society, but it can take quite a while to get these ideas out into the world.

Switching between academic and commercial tracks

Of course in the 21st century most people do not work for the same organisation for their whole life. They switch jobs, and that can include switching from universities to companies or vice-versa.

I suspect that its usually easier to move from university to company than from company to university. At least in computer science, most successful academics have the skills needed to succeed in industry. Also, because we live in a capitalist society, most academics understand the basics of how commercial world works: companies need to create high-quality goods and services which people want to buy, and corporate researchers need to justify their research as helping the company achieve its goals.

On the other hand, it can take a while for people moving from companies to universities to understand how to be successful academics. From a research perspective, success mostly comes from publications and citations, which means impressing academic peers (who review submissions and cite papers). I have seen people coming from industry to academia struggle to understand what academics value, especially since this can depend on what is trendy/fashionable, and also is different for different academic communities (eg medicine vs CS, or HCI vs AI).

In other words, in both worlds you need to successfully “sell” your ideas. But while everyone living in a capitalist society has a basic understanding of what it means to sell ideas in a commercial context, people outwith academia often dont understand, at least initially, how to sell ideas to academic peers.

Timescales

Academics tend to work on slower time-scales than companies. At least in IT, there is an expectation that products will move from concept to sales within a year or two. In universities, if I have an idea for a research project, it can easily take 1-2 years just to get the project started: eg, 6 months to write a proposal, 6 months to find out that it was approved, and 6 months to hire staff and get them in post. And it may take another few years of actually conducting and writing up the research before I have solid research publications which advance my academic field. In contrast, someone moving from academia to industry may be able to start producing results and having impact within a year, because companies work on a faster time-scale.

In other words, it may take several years for someone moving from industry to academia to become fully “productive” (at least as measured by high-impact publications), but much less time is required for someone moving from academia to industry.

Final Thoughts

Whether you’re better off in a company or a university is up to you! I’ve listed a few of the difference between academic research and commercial research above, but of course your decision will depend on what you care about, what you want to achieve, your personal circumstances, and the specifics of the organisations you are thinking of working for.