Here is a possibly unpopular opinion. I have a Phd (got it when I was young). In some areas of computing (e.g. OS, Systems) work gets obsolete fairly quickly. I suspect my PhD topic was useless about 4-10 years after I finished the thesis. Of course, I got a few hundred citations, and this suggests that perhaps more enduring work was done based on mine, but in any case, I don't buy the dent in the universe argument for some fast changing fields (e.g. Deep Learning today). My wife was reading the life story of Louis Pasteur to our 6 year old, and it was remarkable what a dent his contribution made. Most people I knew who finished Phds did not do much of note (and this is fairly elite tier). One person has an algorithm named after them but it is also a bit obsolete (e.g. happened to a lot of classical CV and NLU algorithms post DL).
I was talking to a theoretical Physics PhD at a conference recently, and even he was quite jaded (he switched to CS post PhD). It was obvious that this person was wicked smart and had quant knowledge. So if you want a PhD for transferable skills, go for it! A masters might be faster though.
Do what makes you happy at the end.. it is sort of nice I can drop in the fact I have a PhD and it was probably good for me on the whole .. but if you are thinking of getting a PhD in CS, consider attending top conferences in your field before committing (e.g. SOSP for OS/Systems, CVPR for vision, CHI for HCI,...). I just had no clue what I was getting into and I actively wish someone had given me this sort of info before I got into it. A PhD has a huge opportunity cost.
Whenever anyone asks about pursuing a PhD, I usually tell them it only makes sense if one of more of these hold true: (1) You want to pursue a career in academia or research where the credentialing will matter; (2) You want access to equipment or material that is otherwise difficult to obtain; (3) You want the freedom & flexibility to pursue random interests for another couple years; or (4) You have a deep affinity for a specific topic at the frontiers of human knowledge.
The PhD definitely isn't required if your goal is to maximize learning & earning, apply research to commercialization, or to make a huge dent on the world -- the latter is just exceedingly rare statistically speaking.
Unless you’re a super genius or are super lucky, you’re likely not going to make a dent in any field especially at a PhD level. Not in these days at least.
But the product of a PhD, according to me and many mentors I respect, is you. It’s your mind and experience (not even necessarily the knowledge). A good phd fundamentally transforms who you are as a person. I’d wager a PhD is probably the most reliable way to effect real change in your mind as an adult. I probably realistically lost millions I could have earned in my twenties because I spent 9 years on my higher education (which has nothing to do with what I do now). But I’d not have done most anything different about it!
Caveat being that it needs to be a good PhD with a good mentor - which is harder to find than a good job or even a good partner lol.
> I’d wager a PhD is probably the most reliable way to effect real change in your mind as an adult
Not sure how you define change, but I'm pretty sure one experience with psychedelics is going to effect more change in about 4 hours than a PhD in 4 years. Not the same type of change, granted.
I'm three years into a PhD and during that time have tried some very potent psychedelics (psilocybin mushrooms, LSD, DMT and mescaline). Fearless experimentation - nothing more than science, right? :)
The change I've experienced throughout my PhD (that is, becoming a confident researcher with at least some depth to his ideas) has probably been more intense and almost certainly more long-lasting than my psychedelic experiences. Nurturing an uncomfortable familiarity with the immense limits of my knowledge, as well as regularly pushing myself to exhaustion in order to overcome those limits - inch by milimiter by angstrom - has completely transformed me as a person.
If anything psychedelics had the effect of making me believe that I had become enlightened to various true-natures-of-everything whilst providing me very little concrete understanding as to why this was the case. Much of the personal change that followed my experiences were less due to the substances and more due to my own desire for those experiences to mean something in the long run. Psychs are great and I'd recommend it wholeheartedly for those who have the courage (note: that's not to say that the classical psychedelics are dangerous - they're only dangerous for people who are scared of them). It is my opinion that one should not start believing that these things anything more than chemicals that make you feel a certain way, or help you along the path to doing so.
At least according to Feynman many of his successful colleagues said psychedelics helped them with their imagination. But I’ve not heard of anyone doing great science (or great anything really) because they went on a trip. It looks more like the Lu just get happier about their life but nothing else.
> But I’ve not heard of anyone doing great science (or great anything really) because they went on a trip.
Kary Mullis claims to have come up with the idea for PCR while on an LSD trip, though he also said a lot of nonsense so I'm not sure you should be too much weight on that.
But he also mentioned in one of his popular books about his own experiments in this regard that he stopped playing with these things because he considered the health of his brain too valuable for that.
> A good phd fundamentally transforms who you are as a person.
Huh? I've heard this a lot and I just don't see it. What is supposed to happen to you? I got one in math, published my 3 papers like they wanted, and that was it. I don't think I changed me one bit, other than not enjoying education/academia anymore.
I’d argue that you didn’t do a good PhD then? Even it’s original intent was to be an understudy and learn how to do research and create new knowledge. This changes slightly between fields but my take away is that a good PhD teaches you to find, and respect your boundaries, while constantly trying to push it. It teaches you to plan years ahead and work towards goals that are simultaneously abstract but large as well, with concrete practical steps. It should make “learning” second nature to you, where you don’t need courses or classes any more: you learn whatever you need to accomplish the project at hand.
If you’re able to do all of those things even before a PhD then I suppose you’re born gifted.
Also it should never be about just the fucking papers. Either you get them or you don’t who cares. Not me at least.
I get what you are saying, I just thought all of those things were either present or absent in a person way before going to university even, let alone getting a PhD. If you are not a self learner, I don't see how you can even excel in university in the first place, at least not in a heavily technical subject. No professor is really going to explain how any of that stuff works, if anything they are going to keep it vague and try to weed you out. You have to be able to read texts and make sense of it on your own, maybe with the occasional question to your prof for a detail or two, but if you cannot do 98% of the learning, you are toast. At least in math which I did, there is no way a person can lean on faculty for their knowledge, it really is up to you.
You’re either talking from a place where you’re taking every skill for granted or have no clue at all. I didn’t say anyone will spoon feed how to learn, you just have to figure it out yourself. But the PhD is precisely the time to do that, with occasional nudging from your mentor. I didn’t even have a great mentor and he still spent hours advising me on what I need to improve. If you didn’t get anything like that from your mentors I feel sorry for you.
Unless you’re working next to Terrance Tao, I.e., at the top of whatever game you’re supposed to be in, you clearly haven’t pushed yourself to be better. Of course that’s a choice to begin with if you want to even do it. But given that you do, a PhD should have been the time when you got better at everything no matter where you started. If you already started with amazing skills then you should have become world class at it by now. If you did a good PhD that is. In your case, if the 3 papers you published were “pedestrian” by your standards, you and your mentors should both have pushed for more ambitious stuff.
> In some areas of computing (e.g. OS, Systems) work gets obsolete fairly quickly
That's a very surprising statement. The number of advances in systems in the 21st century is.. embarrassingly low. It feels like a stale dead field. People keep pumping money into what is basically a dead body at this point.
> consider attending top conferences in your field before committing (e.g. SOSP for OS/Systems, CVPR for vision, CHI for HCI,...). I
I don't think that this is good advice. Certainly I would never suggest it to my own students. It's hard to see what they would gain.
So you go to CVPR or NeurIPS, you see hundreds of posters per session and talk after talk. Maybe you stick around for workshops to interact with people. So what? This gives you zero information about what you will be doing as a PhD student or how useful your PhD will be.
The way to figure out if you want to do a PhD is simple: do research first. Spend a summer in a research lab related to some topic you like and figure out if you like it.
> I was talking to a theoretical Physics PhD at a conference recently, and even he was quite jaded (he switched to CS post PhD).
Speaking of dead fields... I know many postdocs who come from physics into ML because many areas of theoretical physics just don't seem to be going anywhere anymore. Not a good sign!
> Speaking of dead fields... I know many postdocs who come from physics into ML because many areas of theoretical physics just don't seem to be going anywhere anymore. Not a good sign!
I do not necessarily agree with the notion that many fields are dead --- they could just be "resting" until some paradigm-shifting idea comes along, at which point they may spring back to life --- but I do agree with your observation that many people with PhDs have to move out of science or research. There simply are too few jobs available in research, so many people with PhDs eventually transition to non-scientific or non-research roles as their opportunities to continue doing research diminish. It's just economics. I do have to agree that it is frustrating how slow progress can be while conducting research, but I think people leave science more often due to the lack of job opportunities than the lack of progress in their research.
As an undergrad, I asked if I should take a control theory class, and was told it was a dead field.
It has become pretty important, since: all the quad copter controllers run it, and it is in network flow management protocols. I would have used it, unlike most of what I did study.
I think your idea about how this dent is supposed to work is a misconception.
If you made some nontrivial insights that were only temporarily useful but other people used it to make their own insights without having to go through the manyears of doing your work, and then the process repeats, there is something permanent there that would not have happened without your work.
Of course someone else might have done your work in your stead, but the same is true for literally all the other scientific greats we can think off, it's extremely unlikely that we would just end up at a standstill with nobody figuring out what's going on if certain people never existed.
Counterpoint: I wrote my Master Thesis on Value Networks in 2004. While I got a perfect grade on it it was largely unnoticed. Now with Social+ and Crypto getting together, it becomes really useful for my work almost two decades later. So I am saying, give it time.
You are missing the point. I had a lovely 20 years where my work didn't connect to my Master Thesis much. Pick an area where you want to work in because you like the area.
Which GP did not claim it was. Also: besides being pedantic this comment mostly seems to be to put the writer down in some way, so applying the principle of charity here, what did you want to achieve with your comment?
There are outstanding master thesis, which introduced quite revolutionary concepts or algorithms. E.g., John Canny developed the canny edge detector within his master thesis. There are of course other examples...
One outstanding example is not much in the way of a proof.
A masters thesis is a few month's work from an untrained student going through an idea that's already being explored by the school's research group. A PhD typically involves far more deep independent work that's deemed relevant by at least one research institution.
I don't know what you need to be huffing to consider a master student at the final stages "untrained". A master is basically the highest form of schooling there is.
... except for a PhD. A master's is typically an admission requirement to a PhD - at least, around here.
In my experience, a master student's final project typically constitute their first foray into / encounter with academic research. So yes, untrained in that sense.
Around here, a training and education plan is required. Some institutions have a requirement on ECTS or another form of "study points" a PhD student must have gathered before defence.
The fact that you are paid doesn't preclude it being a traineeship.
> PhD is not schooling. It's a research job. You get paid to do a PhD.
No not really. A PhD student at most receives a scholarship, and currently at least in Europe has to undergo a curricular part comprised of a dozen or so courses. How does that not count as schooling?
If I'm not mistaken the only exception to the schooling requirement is if a student is able to publish a few research papers on a topic that are coherent and related to a concrete research topic that could support their thesis.
This is not the way it is in the Netherlands. You get a full salary and don't have any courses. You after all already did all relevant courses during your masters.
>>I don't know what you need to be huffing to consider a master student at the final stages "untrained".
I don't know what could possibly lead you to believe that someone just starting off his master's thesis has any experience whatsoever in research, and has any contact with any academic topic beyond the standard curriculum that every single one has to go through to get a degree.
No, not experienced. But a master degree generally includes research as part of the curriculum. There is a world of difference between trained and experienced.
While technically correct how does that relate to the dialogue above? A PhD is even deeper work in a niche topic than a masters, right however that doesn't contradict that the niche know how may become useful later.
Louis Pasteur is one of my most favorite scientists. That being said, he won't have been able to do it all on his own. Variolation was practised in Africa, India and China long back[1].
The record shows that [...] Edward Jenner heard a dairymaid say, “I shall never have smallpox for I have had cowpox.[...]” It fact, it was a common belief that dairymaids were in some way protected from smallpox[1].
That knowledge (regardless of the accuracy of the specific story of the dairymaid) helped Jenner to develop the smallpox vaccine, and that in turn helped Pasteur in coming up with his array of vaccines and other innovations.
So, I would argue that every person who contributes to research is making a dent in the universe. Your published paper may have been the impetus for someone else to make a "visible dent". Publishng papers of failures would have been ideal, but until that happens, every person who contributes to research is contributing to the progress of humanity in the long run.
>Most people I knew who finished Phds did not do much of note (and this is fairly elite tier).
One could argue that research in general is a probabilistic, distributed gradient descent search and you are fulfilling a certain voluntary social duty even if you are not successful.
Unfortunately that doesn't apply to many pursuits because of various perverse incentives in modern academia...
I was talking to a theoretical Physics PhD at a conference recently, and even he was quite jaded (he switched to CS post PhD). It was obvious that this person was wicked smart and had quant knowledge. So if you want a PhD for transferable skills, go for it! A masters might be faster though.
Do what makes you happy at the end.. it is sort of nice I can drop in the fact I have a PhD and it was probably good for me on the whole .. but if you are thinking of getting a PhD in CS, consider attending top conferences in your field before committing (e.g. SOSP for OS/Systems, CVPR for vision, CHI for HCI,...). I just had no clue what I was getting into and I actively wish someone had given me this sort of info before I got into it. A PhD has a huge opportunity cost.