The True Cost of a PhD: Giving Up a Family for Academia | Hacker News Reader

The True Cost of a PhD: Giving Up a Family for Academia

jamesgmartin.center

68 pointsEndXA69 comments

“colleges and universities must find a way to allow those pursuing academic careers to obtain the basic goods that most people seek in life: marriage, a family, and a career to support themselves. The price of pursuing a life of the mind shouldn’t be the rest of one’s life.”

Wait, why is this the universities problem? If too many people want to get phds than our society can support, then less people should get phds. That or they’ll get stuck in the current academic grind.

Any PhD student should be well aware of the career difficulty that follows, particularly if they want to be a professor. There’s no lack of articles like this one pointing out that fact.

I got my PhD but went in well aware that I don’t want to go thru the ringer trying to be a professor. I’d love to be one, but not willing to sacrifice so much to be one.

So I choose to do industry from the start. The PhD wasn’t the most financially advantageous move but I loved it for the 6 years I was there and wouldn’t trade it for the “sunk cost”.

You would think being cognizant and forward thinking enough to evaluate the prospect of a 6 year endeavor is a requisite for prospective PhD students.

And yet...

> Wait, why is this the universities problem? If too many people want to get phds than our society can support than less people should get phds. That or they’ll get stuck in the current academic grind. Any PhD student should be well aware of the career difficulty that follows, particularly if they want to be a professor. There’s no lack of articles like this one pointing out that fact.

But I think there very much is a systemic effect that everyone is missing because it happens so slowly.

Fundamentally, if the number of academic positions remains constant, then each professor only needs to train one PhD student looking to enter academia in their entire career.

OK, maybe 1.1 to deal with deaths or other sorts of natural attrition. But think about that for a minute, and then think about how many PhD students a typical professor actually trains in their career. That level of "production" is OK when the number of academic positions are expanding at a similar rate. But they're not expanding at any where near the same rate these days; instead, there are longer and longer "holding patterns".

And this introduces a pattern where the "holding patterns" get longer and longer, but people don't really notice. Professors who got their PhDs in the 70's didn't do any post-docs, or maybe only a year or two. Professors who got their PhDs in the 80's or 90's did a few years, but not many. PhDs getting jobs now in some fields can expect to have post-docs for a decade; but people getting PhDs now will have even longer waits.

So PhD students "should be well aware", but 1) the people advising them almost certainly had a much easier time than the people currently in the pipeline, and 2) it will almost certainly be much worse by the time they actually want to get a job. Add to that that it would be nearly impossible to perform modern academic research without the army of PhD students and postdocs being paid slave wages, so there's very little incentive to give students a realistic picture of what their chances actually are.

So yes, it should be up to universities and research institutions to set up a sustainable system.

Two counterpoints:

1) There’s _tons_ of other reasons to get a PhD than to become a professor. So the number is a lot higher than 1.1.

2) I can only see it as naive on the part of students to think they can become one unless they’re able to be in the top ~5% of grad students and then work incredibly hard for 10 years. We should be telling everyone this. And encouraging professor aspiration only for PhDs with the heart for this. If we did that then the universities would work things out. But expecting the universities to fix the problem is backwards because they have zero incentive to.

> There’s _tons_ of other reasons to get a PhD than to become a professor.

Sure; that's why I said, "PhD student looking to enter academia". CS PhDs for the most part go into industry (as I did). But if you're a CS prof, you should basically default to telling all of your students except maybe three in your entire career to plan to go to industry. The same thing should hold for physics/chemistry/biology/history/whatever profs: if you have more than one PhD student per decade going on to get a post-doc, you're failing your mentorship duties.

> But if you're a CS prof, you should basically default to telling all of your students except maybe three in your entire career to plan to go to industry.

Except it's not like 3 PhD students from $topschool become $topschool professors while 3 students from $randomschool become $randomschool professors.

What happens is the top students from top schools become top schools professors. The middling students from top schools become professors at random schools. The students from random schools are unlikely to become professors anywhere.

So if you're an MIT professor you can encourage more students into academia. If you're a random school professor, you should encourage them all into industry.

> Any PhD student should be well aware of the career difficulty that follows

There's an important disconnect about the job market. Most PhD students know general facts about the job market: one in ten PhD holders becomes a professor. Very few people have a good idea about that applies, specifically, to their own career prospects.

As for postdocs, treating it as "a holding pattern" is, IMHO, the root of the problem. There's been a massive push, especially from the NIH, to speed up the "pathway to independence", but that's only really a goal because postdocs' job conditions aren't great (low salary, massive job insecurity). Research is hard and it's absolutely bizarre that we want to transition people away from doing it as soon as they are trained, and towards supervising it, a role for which they get virtually no training or mentoring. If I were King of Science, I'd move towards a model with a more gradual career path and less trainee mania.

At least for CS, it was pretty clear to me that PhD and post-doc is not about becoming a professor. Instead, you are getting much more interesting projects, but also a much lower salary.

This is obviously not for everyone, but there are plenty of people who would like to do this for a few years (or much more -- then they end up as research scientists and non-tenure-track professors)

There's a subtle insidious effect of professors being childless. Like it or not, for practical intents and purposes, academia serves as arbiter of truth for our society. If most of the professors are childless, they will be subconsciously incentivized to arbitrate truth in ways that are biased accordingly.

yes, because the vast majority of our society is run by professors...

Almost all of the ruling class in the USA went through the same few schools.

Academia has prestige and its viewpoints diffuse across the rest of society.

This topic is difficult to address in general and world wide. What I can say is: My wife and I graduated in Germany, we both had a net salary/scholarship beyond 1200€/m, so we could easily afford anything baby-related. This would even have been possible with only a single salary -- having a baby is not expensive if you don't follow your urgent need for a house or car. In a German city, one can raise a kid in a single room apartment and study half time -- and this does not meet the national (or my subjective) definition of "poverty".

The actual problem is time. Having a baby, I felt the lack of time every day and night. I could not do as much for science as I wanted to. That definitely reflects in the grades and scientific publications.

Did you have access to health care? In the US, having a baby while in graduate school was profoundly risky from a financial standpoint, and there was a strong bias against women who had kids. I knew families that were bankrupted. Also, they had to completely drain their finances in order to qualify for some forms of public assistance.

As far as I know, it's quite uncommon an any OECD state other than the US to be without quite good coverage for health care bills and medical bankruptcy is even less common—I hesitate to write "unheard of" because I'm sure there are some exceptions somewhere, but it fits the colloquial use of "unheard of", certainly.

It would be "unheard of" in the United States for a Ph.D. student in a normal program to not have good health insurance and the birth paid for. The stipend in the US is ridiculously low, and I can see having trouble affording to take care of a baby, but it is not going to be the medical costs that get you.

Do universities pay for health insurance for PhD students? In my experience, an uneventful birth costs $3k to $5k, with insurance, and it all depends on if the pregnancy costs occur in the same calendar year to avoid resetting the out of pocket maximum limit.

In Germany healthcare is mandatory by law and affordable - for most employees except ultra-high-income, it's a countrywide fixed percentage of income.

Yes, In Germany, health insurance is proportional to your pay ( as long as you do not go private). There are laws called "Mutterschutz" to protect mothers. The mothers are eligible for 65% of their pay for 1 year (parental leave) or 300€/m if they are not working. The child gets about 200€/m from the Govt. The education is free. If you are on a contract that might end when your child is younger than x years, then the contract will be extended etc.

>Yes, In Germany, health insurance is proportional to your pay ( as long as you do not go private).

Not if you are getting a scholarship for a PhD. A scholarship does not count as work income.

Then you have to pay a flat fee close to 200€/month for insurance, like someone who does not work at all

In most cases, the scholarship also pays more if you have a kid. Yes, you have to pay around 200€ if you have a usual scholarship.

> if you don't follow your urgent need for a house or car

Unfortunately these are not the sole concerns in the US. Costs for rentals are spiraling out of control in many places, so whether you are renting or buying is often a moot point. And healthcare is a huge problem as well.

This is happening to a lot of European countries as well. The cost of rental homes/flats is getting higher and higher. Even social rent is at the point that it is often just below 50% of their net income (assuming 1 person with a fulltime job) not sure how that compares to the US. But it is a worrying thing overall that is happening in the western society.

I think it's generally recommended to be at 1/3 of your net income, which simply isn't doable in most areas here in the Netherlands.

It’s also very difficult for mothers because the US doesn’t offer any decent time off, so all the pumping for breast milk and stress that goes along with it is discouraging.

The only reason to pursue a PhD is interest in the subject matter. You don't have to have all consuming passion and love but you have to be genuinely interested. Those who do it because their parents have phds or their colleagues, or they want "dr" in front of their name, in other words, for vanity, prestige or to further their career usually never do well. Either because they start at a later point in their lives, lack the dedication or drive to go beyond the call of duty. A phd is always about doing 110% of what is expected.

In my field (physics), the expectation is to finish a phd within 6 years of completing your undergrad. If you have a good project or work hard 5 years is doable, 7 if you are a good student but got unlucky. This means the average age out of graduate school is between 26-31, depending on what country you received your undergraduate in. That is more then enough time to still start a career or work on other life goals. It's when you take longer then this timeline that things start to go wrong, for both futures in academia and industry and your life and mental health in general.

A mediocre phd is worse then not doing a phd and there are plenty of those around but those who do well (not exceptional geniuses by any means!), a phd will open many doors even if you do not pursue a further academic career.

Man, seemingly every week on HN there’s an article attacking PhDs as monastic pyramid schemes for the foreign, financially gullible, and unemployable.

As a 5th-year computer science PhD student (graduating next semester!) I think this characterization is unfair.

With summer internships, I have made about 40k per year over my PhD, which was more than enough for anything I wanted. I only had to TA for one year. The jobs available to me now are much more interesting than the ones available to me 5 years ago. With the help of a great advisor, a few other hosts and mentors, and fellow grad students and postdocs, I’ve learned a lot about (and now, made some contributions I’m proud of) to a couple of areas of research. As someone who enjoys research, that has been a very rewarding process.

This is not a universal truth. Things might be much worse if I had dependents, had a bad stipend, had a bad advisor, or did not like research. But in my case I had none of these complications, and I’m glad I did a PhD — not just to have the credential, but to have had the fun experience, grown, and now have a research future I’m pretty excited about.

This is pretty similar to my experience, but I also have two kids both born during graduate school. Their addition to our family will delay my graduation a bit, but they're worth it :)

Your experience is atypical for Ph.D. students though not out of the ordinary for CS Ph.D. students. There are other fields that are similarly well situated like Economics, with plentiful internship opportunities and great options after the Ph.D. that aren’t academia.

Even for you in CS and the others in fields where the average position is ok a Ph.D. is a pretty bad bet monetarily but if the consumption value is high enough, great, it’s worth it even if you never even wanted to be an academic. But you gave up a lot in forgone earnings and in work experience and if you’re lucky you’ll make it up in a decade. The work experience might be a wash depending on the field you end up in. The forgone earnings will not be a wash.

If you’d decided to do a Ph.D. in Comparative Literature there would have been no internships, a poor stipend, wild competition for academic jobs and no real demand for your skills and expertise outside academia.

If a Ph.D. was worth it for the consumption value for you that’s great and it’s fantastic that there exist fields that have similar teaching demand, stipends and exit opportunities. But the Ph.D. hate is to tell people

Are there any jobs in this field or demand for its graduates? If not it will be bad. Stay away.

My impression is that a PhD in CS can be beneficial in industry, but the benefits generally derive more from the kind of work you get to do, not in the salary you earn. Some positions available "only" to PhD holders may allow you to bypass many of the unpleasantries in software engineering (technical whiteboard exams, SCRUM, limited autonomy, deadline pressures), but they don't necessarily pay more. And if you do go for the SWE position, a PhD won't keep you from having to go through the SWE hiring gauntlet.

I say this based mainly on personal experience and some browsing of positions and salaries with limited data. I'm a onetime doctoral student who left with a MS and went over to Sun Labs for a bit. Salary-wise, the newly minted PhDs didn't really earn much more than the newly minted MS students[1] - and almost certainly less if you compared them with MS holders who had spent the last 5 years working. But I did get put in giant due diligence projects where I had to recreate and verify ever SQL query in a giant operational software system, while the PhD researchers pursued more interesting projects with greater autonomy.

And if you're the sort who will burn out and decay to poor performance if you're bored, it could make a difference. We can compare MS to PhD generally across a population, but we can't really compare them for the same person. Fortunately for me, I actually like digging into long, complex, chaotic data pipelines, and if you find you enjoy something high impact and complicated that bores people, well then, that's not a bad way to get paid. But if you don't like it, that's a recipe for misery.

I believe, strongly, that we greatly overproduce PhDs, even in employable categories like CS (mainly because I think many of these CS PhDs would have found equally if not more fulfilling work at a potentially higher salary without the long years in grad school). Some people just need to do the PhD style work if they're going to succeed in life, and for them, more time in grad school combined with no clear financial advantage may hide a number of benefits they get from these jobs.

[1] one aside - salaries were notably lower for engineers and scientists in tech companies back then. This was back in the 1999/2000 - I was a grad student at Berkeley, and I was shocked by how much lower all engineering starting salaries were, at the MS and PhD level, than salaries out of the law and business schools, especially since the companies that were responsible for this pay disparity were lobbying congress fiercely to do something about this critical shortage of US citizens studying graduate engineering and science. My impression is that this gap has closed considerably since then.

You're likely in one of a few fields where this is true. A PhD in classical history, not so much.

This is nowhere near the norm outside of CS.

Many advisors don't allow students to intern. Many fields don't offer internships. Many fields don't have enough funding so people TA nearly every year during their PhD. Many fields don't have industry prospects like CS does.

A CS PhD is pretty good. Pay is usually okay. Eventual job prospects are very strong. Leaving with a masters degree provides a great fallback. Other kinds of PhDs are nothing at all like this.

"At least with medical school, Peter said, the job prospects are better. New doctors begin making $60,000 and their salary increases by $20,000 a year during residency."

Actually not true. A resident's salary might increase by $20k between the start and end of their training (esp for intensive 7-8 surgical programs), but most programs only increase salary by $3,000 per year. In the following tables, PGY stands for post-graduate year, and extends from residency into fellowship if you choose to do one. Usually the salaries are the same for all residents, regardless of whether you are training to be a neurosurgeon or a psychiatrist.

https://med.virginia.edu/gme/program-resources/salary-benefi...

https://www.hopkinsmedicine.org/emergencymedicine/em-residen...

I read that and immediately thought 'that'll be news to my surgical resident friend'.

This is an interesting article, but I feel like it only vaguely addresses the root of the problem of grad school in the US(in my opinion). The root problem is that your grad school experience is highly dependent on what university you go to and who your advisor is.

I recently finished my PhD in physics from a top university earlier this year. During my time there, I was paid enough to live comfortably (~35k/year) and had amazing free health care. My advisor was great for the most part, and I had amazing opportunities and experiences that I would not have had otherwise.

On the otherhand, there are other grad students who have terrible advisors and/or at institutions with less resources and opportunities.

> American universities graduated 100,000 new PhDs but only created 16,000 new professorships

Here we go again - this weird idea that the only purpose for doing a PhD is to become a professor.

Agreed. CS PhDs don't typically identify themselves with a Dr prefix, so our population (I have one, I don't use a Dr prefix) is rather hidden. People don't know how many of us there are. When I was at Google, about 25% of googlers had PhDs.

In many fields, that's actually not far from the truth, though. Yes, PhDs in computer science, engineering, and natural sciences can find jobs in industry, but it's not like anyone getting a PhD in 19th century German literature can easily find a job using that other than in academia.

However, computer science is also one of the fields with the largest need for professors partially because of the competition with industry jobs. My experience from a Canadian perspective was that computer science PhDs also tend to be better funded than those in the social sciences so the overall risk is much lower.

The risk may be lower but the forgone income is a lot greater. The median admit to a CS Ph.D. programme, even the worst ones in the US, can probably get a job at Google and can definitely get a job programming for a living if they have the right to live in the US. If they’re from a less developed country a Ph.D. stipend and the possibility of staying afterwards probably looks pretty great.

> The median admit to a CS Ph.D. programme, even the worst ones in the US, can probably get a job at Google

Hah I think the hiring bar for Google is higher than the bar for a PhD viva!

For real? Because people have been hired directly from Lambda School and App Academy to Google and one would get the impression that https://elementsofprogramminginterviews.com/ and leetcode would do, which is maybe six month’s work from a standing start.

Well I know a lot of people with PhDs who couldn't pass a Google interview. Of course it's harder to tell how many can pass a Google interview but can't pass a PhD, because people don't tend to simply 'fail' a PhD except in extreme circumstances, and people who drop out of a PhD weren't necessarily going to fail.

I have a PhD and I don't think I could pass the famous Google interview! I'm just not that good.

This is very true about income. Although I think it's safe to say that the lifetime earnings of a PhD graduate in CS are likely to be well above average across the whole population even if significantly lower than those who skipped the PhD.

A Ph.D. is professional training for being a professor, or at a push, a researcher. If you don’t get yours from a top programme your chances of getting a professor job are quite low in the US, though other countries vary.

A Ph.D. will at absolute minimum take three years on top of a Bachelor’s in the UK. In the US it’s common for it to take six years including a two year Master’s portion. Many fields have large numbers of people taking nine years, like literature or anthropology. Even three years is a huge cost in forgone income never mind six or nine. Then in many fields there’s a post doc before you can get a shot at a professor job and then you spend six years trying for tenure, which you often won’t get. That’s fine if you’re willing to live like a monk because you love what you do for most of your twenties and if you could be guaranteed a job at the end of ten years plenty of people would take that offer but many very talented people spend close to a decade, or more, in training for a job they never get.

A Ph.D. for fun, for consumption purposes is fine if you come from money or are supported by a working spouse with a real job but for a great many people they end up with nothing after enormous effort.

> A Ph.D. is professional training for being a professor, or at a push, a researcher.

It's professional training for being a researcher. You can do that in industry or academia. It doesn't make a huge difference, really.

Becoming a professor is something that only a tiny minority of people are ever going to do, perhaps after another decade or more after their PhD.

Unfortunately, there are also not a huge plethora of jobs for researchers in industry either. At least, it's hard to find posted job openings. It may be that a lot of the people in industry who are doing researchy work started out as clerks and worked their way into their positions over time.

And yet the system is set up for a time when every Ph.D. who wanted a job as a professor would get one. That ended in the 70s. Now you have people doing three post docs before giving up to go into industry as they could have straight after the Ph.D. which many would never have done if they’d known where they’d end up. And you have humanities Ph.D.s doing jobs that in no way require the training and social science Ph.D.s doing jobs they could have ended up in if they’d never gone to graduate school though it’s not quite as bad because at least they often end up getting jobs that require the six or so years of experience they spent in grad school, earning a lot less money.

Industrial research positions are collapsing. Even in CS, what is left outside of ML? MSR is declining. Fujistu labs and IBM research aren't doing much.

I wouldn't agree with that - I think industrial research is flourishing it's just moved from separate labs into the mainstream parts of companies. For just one example the V8 team at Google at a product group but they're doing research - they have PhDs, they publish top-tier papers, they have students. They happen to ship their research, that's all. Seems like a good thing, really?

I'm a researcher at Shopify. I don't work in a formal lab, but I'm doing research, writing papers, reviewing papers.

It's certainly true that there are reasons for getting a PhD other than becoming a professor, but it's also true that in many fields having a PhD is also completely unnecessary. And yet in many fields there seems to a qualification arms race where, due to the glut of PhDs and lack of jobs where PhDs are truly required, jobs for which a PhD is not necessary or beneficial are starting to become expected. I've definitely begun to see this in the fields of machine learning and data science. I'm sure other fields are similar.

A PhD is like a vocational arts school for being a professor/researcher. The fact that you can do something else with it is a bonus, not a purpose.

> > American universities graduated 100,000 new PhDs but only created 16,000 new professorships

> Here we go again - this weird idea that the only purpose for doing a PhD is to become a professor.

Or that professors need PhDs.

Time spent getting a PhD deprives future educators of skills beyond academic ones that will serve many students better than academic skills.

It's the one job that requires a phd so it's not weird

I don't think that's really an absolute requirement - there are full professors without PhDs.

I think the characterization is a side effect when students ask "What do I do next?" After about four years pursuing a non-vocational degree, a lot of students have job prospects inconsistent with a meritocratic opportunity worldview. But not just students, their professors also recognize that barista is a median outcome for many new graduates without connections...the people who most depend on meritocracy to provide access to good jobs.

The Phd as a route to professorship is a way of maintaining the idea of meritocratic access to stable and fulfilling careers. The belief meets everyone's need to believe that people will just give money to the deserving intellect.

I wanted to share my experience. I was a foreign student doing a Master in the US, with several other schoolmates. 3 of them decided to pursue a PhD after their master. We all understood that a Master is done in 2 years, a PhD in 3 years. So they all completed their PhD in 3 years, not a month more. But I met many students, in the same field, some with the same advisors, who were 5+ years into their PhD with no end in sight. I think a lot of it came from the state of mind. My friends had a very specific time lime to finish their PhD in 3 years and they followed it, pushed their advisors when needed, etc. There was never any doubt that they would be done after 3 years. Whereas others started an open-ended journey with no clear goals.

"Solve for the equilibrium": Only the tiniest fraction of superstars, those with nothing better to do, and foreigners from poor countries will go for a PhD. Beyond that, the talented will avoid the PhD.

This article and the sad reality it portrays reminded me of Episode 16 of Timesuck: “Is we getting dumberest?”[0] The most obvious conclusion to be drawn is that the brightest and most educated members of our species are incapable of producing offspring because our society does not support them economically. Throw another reason into the “why we’re getting dumber” bucket.

[0]: https://m.youtube.com/watch?v=uxUXDfvBZ7o

The comparison of pay between what PhD students get in the US vs other countries is plain silly. It’s apples to oranges, the cost of living can be an order of magnitude different, which is not mentioned even once, favoring the absolute number comparison as a good thing.

Beyond this the article talks about low stipends, while an important topic of conversation - nothing new.

I dont see a problem with that. Too many people in research - market needs correction.

I would love to have a PhD, and the associated prestige. I like academics and would probably enjoy the classes.

I looked into it a few years back, I was absolutely stunned at the amount of institutional butt-kissery that was associated with it. That, and the fact that most of the PhD programmers I know are not any different from the average has convinced me to stay put.

> enjoy the classes

You might be teaching some classes (or grading, most likely) - but you won't be taking much of them. PhDs are about doing research.

Having said that there are certainly a lot of PhDs who are not above average.

> enjoy the classes

But you don't really do 'classes' as a PhD student. At that point you should be the world expert in your subject so who would be teaching you :)

In my field, it seems most PhD students spend the majority of at least their first year taking classes. But it's true that the overall course load is significantly less than bachelor's degree. Of course you don't come into a PhD program already being an expert. Even if you are, there's certainly much to learn about related fields.

> But you don't really do 'classes' as a PhD student.

In the US almost all programmes begin with a two year long Ph.D. student period at the end of which, assuming you pass your qualifying exams you can drop out with a Master’s. Once those are done you absolutely must focus on research though if you haven’t started on it already you’re behind. After the Master’s you’re a Ph.D. candidate. Somewhere in this period, if you’re doing it right you’re very likely to become the world’s leading expert on your particular, extremely narrow field. Then, or while becoming that expert you write up your thesis, defend it and get the degree.

In the UK it’s still possible to go directly from Bachelor’s to Ph.D. candidate and be done in three years but the trend is away from this, at least in the exact and social sciences.

> At that point you should be the world expert in your subject

Well, maybe by the end of your PhD, but certainly not at the start, when coming out of the MSc.

A Ph.D. is not about the classes, it’s about the research. The classes are to prepare you to do research. If you’re interested in the classes do a Master’s. Georgia Tech has a great one in CS for under $10K[1] and the University of London has many in other fields for not much more[2].

[1] https://www.omscs.gatech.edu/

[2] https://london.ac.uk/ways-study/distance-learning

Yeah indeed. In a very course heavy program you might take 12 courses. At two per semester that’s half of your 40 hour week for 3 of 5 years. That means you’re spending 30% max on coursework. Even less if you include your summers as working weeks. Say like 15-20% then.

> I was absolutely stunned at the amount of institutional butt-kissery that was associated with it.

What does this mean

> most of the PhD programmers I know are not any different from the average

Very few computer science PhDs are set up to make you better at programming. In fact, the unstructured and unreviewed quality of much research code might make you worse at programming. The goal of a PhD is to get better at research and communicating your research. If you’re not interested in research, do not do a PhD.

Reply on news.ycombinator.com