Vrftsjtryk

In CS field, an entry-level software engineer in top-tier tech
companies (e.g. Google, Facebook, etc.) could earn as much as a
20-year experience professor....Why do so many PhDs still choose to be a professor while they have the choice to go to the industry?

First, your assumption is wrong. Most PhDs end up in industry. I don't have any source but I think this is from 90 - 99%.

If you limit your question to the 1% - 10% that become professors, they are all very successful, i.e. they have plenty of papers, promising research direction, strong network etc etc. And there is a reason for their success: they have passion, and when you have passion, money is likely not the most important thing in life.

Except for machine learning, in most areas in CS, you need to stay in academia to do research. And there are many benefits that you can only have when working in academia.

• You take credit for what you have done. Products in industry are developed by a large team, and nobody can take full credit for it. But researchers can take full credit for what they do in their papers.


• Reputation: you are invited to give talk, become PC members, etc etc, and everybody will know you. I would be excited to meet an author whose paper I have read. I'm not excited at all to meet a Google employee. (I'm living in Mountain View, a small city with 80,000 residents, but more than 20,000 Google employees)


• Do interesting jobs. You always work on new things in research, while the majority of tasks of a software engineer are maintenance, fix bugs etc.

(I'm a software engineer if you are curious)

Update:

Regarding many comments about research in industry.

@Aåkon

Plus all the companies even disney have their research division and
increasing using the slogan "research without limit" to attract
researchers from academia.

Yes, yes, I'm aware of those industrial research labs. However if you compare the number of PhDs in industry who still publish vs the number of PhDs who do not, I guess it is less than 0.1%, even if you compare the same company (again, no source). I'm using the Z3 solver developed by Microsoft Research, and I notice that group have the same 3 people in more than 10 years.

@xuq01

but people in industry do regularly publish in top conferences (off
the top of my head, at least they do publish quite a lot in SOSP/OSDI,
PLDI, OOPSLA, nowadays even in POPL). People in industry do care quite
a bit about systems research for example

Ha, I guess we work in a closely related fields. You should point me to the group that publish at those conferences. Look at Domagoj Babic's group at Google, last time I checked they hadn't published anything for years. Look at the the Facebook Infer group, they just won POPL 2019 most influential award for a paper they published when they were professors. But the lead authors, Calgano and DiStefano, haven't published anything since joining Facebook in 2013.

I've worked in industry science, and in academic science (as a PhD for both - I'm back in academia now).

Industry Science

• The problems you're working on are tangible and usually very interesting


• You have resources available


• There are good things about a real HR department (never had a missed paycheck in industry science; never really worked more than 40 hours a week)


• Most of your coworkers, even the dumb ones, are pleasant enough at work.


• You will spend at least half your work week in meetings. Most of which will be valueless


• Corporate middle management is usually staffed by people who are arrogant and self-serving enough to escape actual work, but too dumb to really do anything important. These people will usually supervise you. Daily.

Academic Science

• The problems you're working on may be interesting or they might not be. Academic freedom is bullshit - if you want to be successful, you're still as constrained in what you can do as industry science (IMO)




• You are more resource limited, but usually have more manpower available




• There are good things about a fake HR department (you can finish all your annual trainings in 20 minutes on a computer)




• You may spend a lot of time on administration, depending on your role




• You may spend a lot of time teaching, but it's generally somewhat rewarding unless you're teaching intro stats to 600 people who would rather be high




• Most of your coworkers, even the unpleasant ones, are smart and provide an intellectual environment that's usually positive




• You usually don't report day-to-day to a idiot chimpanzee




• You can make fun of corporate-speak without having a meeting about being a team player.

So I mean... there's good and bad on both sides. The pay gap exists but isn't as big as people think in most fields (when I left my last industry job for this academic job, the salary was pretty much the same). Just do what makes you not want to stab people every morning.

As far as I see it, the main advantage that academia has over industry is freedom. In industry, you generally work on what your employer tells you to work on. As a professor, even at the entry level (assistant professor), you have quite a bit of leeway to work on what interests you, with the constraint that you have to find an agency that will give you money for at least some of it. Many postdocs and even some graduate students have the ability to come up with their own project ideas and pursue them, as long as they are somewhat consistent with their mentors' funding streams. The constraint of being able to find funding is a big one, but I still think that I have much more freedom than my friends in industry. As a professor, I only talk to my boss about what I'm doing once or twice per year. In academia, usually your boss is happy as long as you bring in grants.

Another reason to stay in academia is inertia. Once you've spent 5 years getting a PhD, you know a lot about what a career in academia looks like and have resources to help you move forward. Finding how your skills might fit in industry is less obvious and you might not know how to start.

Socially acceptable answers:

• Becoming a professor is unlikely to be the goal of the majority of Ph.D. students.




• Can demand higher salary upon joining company




• To pursue their own interests and passion without pressure from company




• To teach others and continue to learn.

Socially unacceptable answers:

• Pathway to immigration into wealthy, Western societies. Count the number of Chinese, Indian and Iranian Ph.D. students in your research field and their overwhelming white, Anglo research advisors and calculate this ratio.




• Pathway into high paying companies, especially by people who did not study a lucrative major. I cannot tell you how many people I know from civil engineering or chemical engineering have used their Ph.D. as a way to learn advanced software courses and join Google or a bank afterward.




• They cannot cut it in the competitive industry world. Industry routinely does their own research but they usually hire the really big names: inventors, people who have written books, etc.

A few other points.

• Not everyone has an enormous industry to fall back on. The body of your post mentions CS, but your title doesn't limit to science fields. There is no high-paying industry for most people in the humanities, subfields of linguistics, psychology, imaging.




• Personally, I'm glad I don't work for a private company and "produce value for shareholders."




• Many universities are still good places to work in terms of benefits; good retirement, medical, and other perks, especially in public schools that can offer government benefits. My university has a pretty generous vacation schedule, even if a lot of people don't take their days.

I wish I could just "chose to be a professor". But you are correct in your assessment that, for people who chose to stay in academia, money could not be the motivation.

But, doing a PhD is not just obtaining another diploma, just another step in one's education. The knowledge (i.e. collection of facts related to your field of study) obtained during a PhD is so specific that it is not easily transferable. The transferable skills a PhD candidate trains for have to do with understanding scientific literature, being able to make connections, conclusions and get new ideas from it. After obtaining their PhD, one should be able to do that reasonably well with literature from any (sub)field related to theirs, after a short period of time needed to come up-to-speed with a new topic.

On the other hand, skills required for a software engineer are a different set of skills entirely. A masters education will sometimes familiarize you with the basic tools (programming languages) and concepts needed, but you eventually need to develop skills related to writing legible, repeatable, reusable code, which you mostly obtain by a lot of practice. Judging by some of the industry interviews I had after my PhD (in parallel with the postdoc interviews), I would've probably done much better in them before my PhD, or even before finishing my masters.

So, they're jobs requiring different skillsets. I chose Computer Science because it fulfilled me, unlike e.g. medicine or law. Taking this further, I want a research job because it gives me a sense of fulfilment I wouldn't get from a software engineering position. Pure research positions in top-tier companies are quite competitive, so there's good chance somebody who can't succeed obtaining a permanent faculty positions will be able to secure a research position in industry.

Some of the following points are what is important to me, rather what might be important to everybody. Comparing research positions in industry and academia (though my experience in industry is limited to what I got from a few interviews, and then a few months where I was forced to take an industry job while waiting for my immigration documents):

• 
  

  Most research positions are much more flexible with working hours. People understand that, doing creative work, some days you just can't get anywhere, and some days you're on a streak and don't want to stop after 10 hours. People that are really not morning people can occasionally arrive late for lunch, and nobody says a word.

  
  
  

  In industry, I've heard arriving to work at 9:30AM is considered as having "flexible work hours".

  
  


• 
  

  Even in case of very strong research ties to industry, holding an academic position allows you to chose the problems you want to work on (possibly amongst several industry collaborations, but the choice is still made according to your research interests and research group).

  
  
  

  In industry, you need to work on the problems as dictated by the market, interesting be damned in favour of profitable.

  
  


• 
  

  In my academic interviews, people were interested in the problems I was tacking.

  
  
  

  In industry, people were interested in my problem solving skills. Nobody cared about what I applied them to, just whether I can apply it to the problems they would present me with.

  
  


• 
  

  Academic jobs are some of the most travelled jobs out there. During my PhD, I wouldn't be surprised if the amount the University covered for my travels would have easily closed the wage gap to industry. Conferences and professional visits are an integral part of academic research.

  
  
  

  Travelling with industry always has a promotional purpose.

  
  


• 
  

  I feel free to discuss my ideas with anybody, anywhere. I reach best conclusions through discussion. The taught of chatting about a topic, having an idea and having to bite my tongue is a bit terrifying to me. The worst that could happen is somebody poaching an undeveloped an untested idea and researching it themselves - not a terrible loss since I don't get to develop all of my ideas anyway.

  
  
  

  In industry, one needs to constantly think of confidentiality, and which details of their work they are allowed to disclose.

  
  


• 
  

  The goal of publications, as the "tangible products" of academic research, is to share your ideas, results and findings with the world. I enjoy the idea that my work is public and adding to the collective of human knowledge (let's not discuss paywalls right now...).

  
  
  

  In industry, all the ideas have to be intellectually protected before publishing: patented or whatever. The goal of publishing is not, precisely, to share your approach with the public. It is again promotional; to boast about the results you obtained with your new method; the details of which you might (need to) try and keep vague.

  
  

In my head, the list of contrasts goes on. It's little things and big things, but all in all, all the freedoms that staying in academic research allow me are worth more than the difference in monetary compensation industry could offer. If somebody wanted me to work a 9-5 job, working on their problems which I find marginally interesting, and not discuss my work with anybody not on their payroll, they would have to offer me much more than the industry standards. Since I'm not a research rock star, I guess I'll stay in adacemia if I can, and leave an industry engineering job as a fall-back option.