Who knows!! By the time you graduate, ML may be out of fashion OR there may be too many ML PhDs by then.

I got my PhD in 2008 and my room-mate was doing a PhD in object recognition. Fei-Fei Li (who now is a bigshot at Stanford now) was a professor at UIUC in the same group. In the pre deep-learning days, object recognition was tedious, boring and un-cool. Semantic web etc was hot then!! And look at things today.

4 years is generous. I know very few people in the states who have gotten their PhD in only 4 years.

It's the best-case scenario. I figured comparing 4 years to 9 months was bad enough, but it could easily turn into 9 years. Or more.

That's because in the US, PhD is actually the combination of masters degree (typically 2 years) and PhD (typically 3-4 years) in other countries.

At least in sciences, US universities don't typically even offer a MSc program; people can get MSc by enrolling into PhD, which is typically the only available graduate program, and quitting it after completing the coursework.

That isn’t really true either. We don’t receive masters degrees unless we do a gentleman’s withdraw, then it’s only an M Phil. If you come into a PhD program with an MSC, you can still take 8 years to finish, ya, you get to skip some course requirements, but that isn’t where the majority of the time in a PhD goes.

MSC’s aren’t offered as much anymore because they had the same problem as PhDs (it could take 4 years to finish one) without the prestige of a PhD at the end. They don’t even guarantee much of a jump start on a PhD. These programs have mostly been replaced by professional non-thesis or thesis-light options, along with “5th year” programs at universities that tack on a few courses and a straightforward thesis to an undergrad program (MIT is famous for these, many universities have copied that).

European universities often have the latter 5th year programs as well where they are considered even more crucial because many of them have only 3 year undergraduates.

What exactly isn't true about it? In all institutions that I know all, students are given MSc status as soon as they complete the requirements, not M Phil (never even heard of it).

Some students leave the school after that point, but it doesn't invalidate their MSc, nor their MSc gets converted to M Phil or whatever. It's not a common thing, but I've seen students drop with MSc mostly because of issues with their supervisors (tenured faculty without much incentive to do research at a good pace).

Edit: I looked it up, and apparently, you're making a blanket statement based on a niche practice in a handful of US universities: https://en.wikipedia.org/wiki/Master_of_Philosophy#United_St...

"Although most American universities do not award the MPhil, a few award it under certain circumstances."

And it takes 2 years for most students to meet the MSc requirements (basically they need to complete the coursework), not 4 years, barring some special cases. Again, I'm talking about the common practice, not the handful of exceptional universities in the US you're referring to.

Many universities just toss you an MS degree along the way for free in your PhD, but coming in with one has no real effect on your PhD length besides waving some clad requirements. At the end of the day US PhD programs just assume you will do 5 or more years of research, for those looking to be professors this is a boon as many US PhDs skip the postdoc step often required for EU PhDs which are typically 3ish years.

In many US schools, an undergrad entering a PhD program will need to satisfy the Masters requirements along the way. The only way you don't end up with an MS degree is if you forget to file the paperwork.

It depends on the discipline. In the geosciences, MSc degrees are the typical 'working degree'. If you want a job, you basically need one, therefore lots of schools offer them. Most people spend 2-3 years on an MSc and then hopefully get a paper or two out before starting work or going on for a PhD (another 4-8 years). You also have to have a pretty solid thesis to get the MSc so you can't just bow out of a PhD program.

Some of the more academically-focused (private) schools don't have much of an MSc program but they're atypical, and they don't dominate research like they do in other fields.

Oh yes, all my experience is for computer science.

> has little to no value in industry.

So what ? It has great value to history and humankind.

The average case of a PhD (even in machine learning) is that a handful of people will read your work, and you'll get cited a few times. That's it.

The analogy I've always used is that researchers are like miners in a gold rush. Most workers take their pickaxe, labour at the rock face, and come away with little more than sore muscles. A lucky few will strike gold (sometimes by looking in the right place, sometimes by working hard, and sometimes by being lucky). You need all those hundreds of people labouring away to find the gold, but the efforts of a single worker matter less than you might think.

This is exactly how I decided to go.

If your goal is to maximize your lifetime income, academia is clearly not for you, but neither is the Peace Corps, the clergy, civil service, or anything like that. However, if you can afford to spend a few years trying to make the world a better, more interesting place then grad school isn’t a terrible option; you can always do something else afterward.

I find that to be a funny statement. Would you actually argue that every single phd has "great value to history and humankind"?

No, of course not. I would argue that:

- Some do, and we're really bad at figuring out which ones the are, especially ahead of time.

- It's a collective effort. Your thesis and my thesis might not be particularly groundbreaking or influential individually, but each one is a tiny step, more or less in the right direction, towards understanding something. It's mostly a myth that "geniuses" make huge leaps forward, while everyone else muddles around. Most ideas are presaged in the literature for a while before someone puts them together and runs with it.

I find it sad when people have no other metric of value other than $$$. "I can't sell this, what is this?"

It's not about money.

The large majority of phds collect fewer than 20 citations. Most papers are ready by only a handful of people. "Screaming into the void" is a pretty accurate representation of many phds (including my own).

Working in industry and making money also contributes to humankind. Whatever the company is doing must be of some value in order to make money. So someone who may not have the right motivations to getting a PhD is probably better suited to move into industry sooner rather than later and be the most productive contributor to humanity as possible.

I agree with you on the point that money is overemphasized, but does a PhD really mean you're more likely to add great value to history and humankind?

Yes, that is basically the job description; the point of a PhD is to contribute new knowledge.

PhD's doesn't contribute new knowledge, they contribute knowledge that can't be proven to be old.

It might sound similar, but it is not. Let me explain the difference:

A person performs a study creating lots of data. If he wanted to contribute to human knowledge he would publish the data with no comments, as he is probably not the best person in the world to analyze this data. If on the other hand he was a PhD he would not publish the data, instead he would publish some pet theories related to the data so that he can build his brand, and most of all he will absolutely not publish the data since it could possibly be used to prove that his pet theories are not relevant or maybe even wrong which would be disastrous for his brand.

Now there are of course PhD's who do the right thing but it doesn't benefit their career at all.

I would argue that, while the pursuit of new knowledge is surely admirable, not all of it can be assumed to meet the high bar of being of "great value to history and humankind".

But in fact most often old topics are revisited with slightly different approaches.

I’ve always been curious about why people write “$$$” instead of “money”. maybe you could elaborate

I can't speak for the person you replied to, but sometimes this is done as a way of disparaging money and the desire to have some, perhaps in the same vein as writing "Micro$oft".

I don't know, I guess it's a bit more cartoony.. The dollar sign is tightly connected to concepts like profit/capitalism etc

Maybe humankind should reward it more, then.

I think you're on to something here.

many of us do. or are you saying it should be mandated by the people who have all the weapons?

Well if it has no value to industry, chances are it has no real value to history/humankind other than "oh that's kind of interesting"

Value isn't always apparent immediately. Quantum theory seemed largely useless, but now it's useful in computing.

> ... but now it's useful in computing.

I do hope you're not referring to the google leak. That result is merely unreproducible with classical computers. It isn't currently useful for anything except PR.

Unless you mean something very specific by "Quantum theory" I don't understand how you can even make that statement about such a broad topic. Quantum mechanics are extremely useful.