The illustrated guide to a Ph.D. (2010) | Hacker News Reader

The illustrated guide to a Ph.D. (2010)

matt.might.net

66 pointsmohitchawla46 comments

> 4. Ability to work with poorly defined goals.

This is the most difficult part of a PhD for me.

Note: This quote is from the article on which this thread was pointing to before it has been changed by mods. https://news.ycombinator.com/item?id=13382223

It is actually the part of the PhD that led me to terminate with my masters. Three years later I'm still back and forth on whether I did the right thing.

I bailed on a PhD with a terminal master's in 2004, worked as an engineer at a factory for a few years, and completed a PhD in a different field in 2016. In hindsight, both decisions were right, and having finished the PhD in one field, I don't think I would have been happy with a PhD in the other.

I'm interested in your story, can you tell a bit of it?

I'm trying to figure out if it's the right choice for me to pursue a PhD.

What was your research experience like?

I've seen a few sources mention the advisor conflict of interest (no incentive to help you finish and graduate because they will lose you after) that you mentioned earlier. Do you think this always a is a significant problem, or only if one is unaware of it?

My first time in grad school (E.E.), I was really young and naive. I'd barely had any real-world experience of any kind, and I was very good at taking classes. So I signed up for a Ph.D. program and had a lot of fun doing the coursework. But then when it came time to do actual grad student things, like original research and paper writing, I was totally unprepared and my advisor showed me the door.

The second time around, I think I still went into grad school for the wrong reasons, but they were different wrong reasons. After dealing with a lot of people who were woefully underprepared to work in industry, I thought getting a Ph.D. would put me in a position to fix the next generation. I underestimated how little the academy wants to do that. I opted not to continue in academia once I got my Ph.D.

Regarding my specific research experience, I'm not sure where to begin. You work like a dog, your advisor probably doesn't agree with your conclusions, might not agree with your premises either, and he wants you to go back and read an ocean of literature. Repeat for 4-10 years.

Advisor conflict of interest varies from advisor to advisor. Some of them like to produce lots of successful students, others like free labor. Either way, bear in mind that the advisor has his own agenda and is not just advising you out of the goodness of his heart.

Final piece of advice: read some "quit lit" to get a perspective on the academic job market. Read some of the critical responses to "quit lit" by academics who haven't quit. (http://www.theatlantic.com/notes/2015/09/no-one-cares-that-y..., https://chroniclevitae.com/news/216-why-so-many-academics-qu..., https://historiann.com/2015/10/07/tired-of-academic-quit-lit...). Bear in mind that if you're in an engineering field the job market isn't quite as bleak as in the sciences, or worse yet, the humanities. But it's no picnic either.

I think it's the best part. The definition is left for the student to learn.

"Ability to work with poorly defined goals."

Poorly defined goals should not be unfamiliar or scary to anyone who has done at least one IT project for money. Actually, I haven't encountered many well defined goals in my professional life after graduating high school many years ago.

Exactly this. I've never encountered any well-defined problem outside of a textbook. In practice, I've always had to discover the goals / problems and then the underlying processes that define them, and only then can you begin to manipulate the processes to optimize your goals. This whole process is iterative, self-influencing, and dynamic. I honestly can't think of one situation where anything was well defined and thats in over 7 years of software engineering exp, albiet startups, agencies, and freelancing, perhaps at Big Corp #1 its different, experience tells me to doubt it though...

Having lived on both sides of the fence, I'd say work and PhD are different levels of poorly defined. Poorly defined goals a manager imposes on you are one thing. At least you don't have to make them up. Poorly defined goals you impose on yourself (based on funding and advisor constraints) are another thing. You can make the self-direction work for you, or you can drown in your own creativity.

I think the difficulty the parent poster mentions is really meant to be read as:

"Ability to work with poorly defined goals AND almost no one knows about this subject you are discovering".

So of course you could start a new Web App for a customer and have poorly defined goals (good luck with that!), but the problems you might encounter have solutions already fleshed out by millions of other Web App devs (Stack Overflow). Something tells me undergoing a PhD is not quite the same, as the 'resources' in this context are a few orders of magnitude scarcer, and the subject matter a few orders of magnitude more complex ...

(I am not a PhD, please forgive me if this is wrong)

Yes it is a little bit of that.

In industry goals are maybe poorly defined, but at least you have directions, and people that can answer the questions you may ask. If the contract is about creating a website, at least you know what you are going to achieve, even if you may have to fight with the customer to understand what he wants/needs on this website and what problem he is trying to solve.

Using the image of pushing the boundary: Doing a PhD is like wandering within some heavy fog. You want to push the boundary, but first you have to find the boundary. You don't know in which direction is the closest boundary as you can not see further than 10 meters from you. Maybe also the closest boundary is on top of a 8000 meters high mountain that you don't have the capacities to climb. So at the end you spend you time trying to figure out in which direction to go. Some PhD students may have supervisors that are guiding them through the fog, as they already explored the area and know it, but some other are on their own, trying to figure out a boundary to push that is accessible for them.

Situations of all PhD students are probably different, but this is my feeling about it.

I can relate to the parent comment and there are differences between doing an IT project for money and PhD.

I think, you rather meant poorly defined requirements for an IT project. The goals of the project do not matter for you, you get the requirements from a paid project and the one responsible of the project has to think about the goals. While during PhD you have to define your own goals.

Also, as on the picture in the post, you have to push from the circle, you have to find it where you want to push and you have to define how much should you push. While doing an IT project, you are usually moving inside the circle, i.e. you are usually combining working concepts.

Sure, but for the average 22 year old who has (up until then) gone through life moving from one exactly delineated set of expectations to the next, complete with quantification of the quality of the result (grading) and 'will this be on the test' expectation management, the PhD unboundedness is as alien as it is frightening.

The key thing I learned was that you're done when you decide you're done. From what I've seen, if you sit down and write a dissertation, it's a rare committee that will stand in your way. This even includes a friend whose advisor refused to read his dissertation -- the rest of his committee approved it and the advisor was pressured into compliance. (In this case the advisor wasn't adversarial, just always "too busy" and didn't want to lose my friend's almost-free labor.)

In my own experience, I had a year that was almost completely wasted. I almost gave up, but instead I decided to stop trying to do work that other people wanted. I wrote the two chapters I wanted to write, because it was the only way they were going to get written, and in the end the committee was fine with them.

That is the part I enjoyed while getting my PhD. I spent 5 years learning and solving scientific questions. The lack of a concrete goal or predefined problem made it interesting because you get to figure it out along the way.

Not mentioned in the article is the strange workings of scientific publishing. In my 20s I found writing open source software much more interesting than writing scientific papers describing the software. Whereas to me at that time it seemed everyone else was describing the buggy collection of Matlab code in articles without publishing any of the code (or datasets used), making it impossible to reproduce their results.

The topic of my unfinished PhD was finding structure in musical pieces with AI.

I'd just like to say that a master's degree makes you a better leader in the same sense that reading a book does. Knowledge is great but knowledge alone doesn't do diddly-squat for becoming an effective leader. That takes soft skills. I know, I have a master's myself. It taught me nothing about what a tech lead or manager needed to understand.

This is old and has been out there for ages. Regardless it's still one of my favorite simple communications of a concept.

Is the dichotomy at the end of the Might's article content accurate? It seems too black-and-white to me, but I didn't go through a CS curriculum. Do MS/CS students not have to do original research or read research papers?

There is such a thing as a masters by research. I guess the article is talking about a taught masters.

I have a MS in computer science that I got at UC Berkeley. I had to take a lot of classes and write a thesis. Primarily, the thesis involved the design and implementation of something interesting (at the time, at least), and then I wrote about what it was, how it worked, and so on.

I also got a PhD in science education at UCB, and that required writing a research proposal, carrying out a six-month classroom intervention, and a lot of other work on both sides of the research (design and framing the work, analyzing the data). The amount of work was very different. The number of classes was not too different - I had about the same number of credits.

All that said, my experience in academia (and out) has taught me that a degree is only one indicator among many. I've had doors open to me because of my degree, advisors' connections, and where my degree was granted, some doors I did not particularly deserve, but the degree itself only shows that I can do a certain kind of work when I want to. I've met plenty of other people who can do the same kind of work who don't have the same kind of (or any) degree.

So, it feels like one of those 'neither necessary nor sufficient' sorts of things.

In my department (mid-tier university, I suppose), yes, an MS degree can be earned simply by taking courses (11 courses and a seminar on research). Alternatively, one can complete a project (9 courses, research seminar, and project defense) or a thesis (8 courses, research seminar, and thesis defense). However, even if you choose the course option, you will still have to read plenty of papers in your courses.

Depends where you go. I did my MS by research, no coursework at all. The research wasn't particularly novel, more a working implementation of literature in a slightly different context, but if it was more than that it would have been a PhD, right?

I still find it odd that some people do coursework during their PhD. Seriously, put on your big girl panties and teach yourself. That's what your degree is all about, discovering things without someone holding your hand.

As a rule, you don't have to achieve any original research for a master's degree, doing research on the current knowledge and summarizing and/or applying it is considered sufficient.

Good students often manage to get some original research done during their master's years, but many (most?) don't.

Ah, the mighty diagram. It's been a while.

It reflects the mainstream and very unfortunate view of academia inside academia. (Disclosure: I'm a PhD student)

So what's your view?

Somebody had explained this to me in words long ago. Interesting to finally see the visual representation!

I pointed out that this guy failed to properly attribute the work, and he blocked me on Medium.

What a piece of work.

Really like how minimal but not ugly that blog is

"A PhD degree equips you to do original research and potentially lead R&D teams."

No you don't need a PhD to lead R&D teams...you don't even need a degree to do that. And no you don't even need a PhD to push human knowledge. Too few of those who did push it, did it through a PhD. In fact its just "possible" that a PhD will push the boundary. Usually it does not.

He didn't say that you can't lead an R&D team without a PhD. He said it equips you to do so. There's no claim about it being the exclusive path to leading an R&D team.

PhD theses are supposed to push the boundary of knowledge. I'd posit that if it doesn't it shouldn't be accepted and the student be sent back to his or her research.

That being said, sometimes pushing the boundary of knowledge is exploring new processes to do the same old thing. We shouldn't downplay that; those are genuine contributions.

The content of this article is lifted entirely from Matt Might's famous post; I'm not sure what value the author intended to add by rehosting it.

http://matt.might.net/articles/phd-school-in-pictures/

It looks like someone changed the post's link to Matt Might's now.

Yes, a mod changed it from https://thebittheories.com/masters-versus-ph-d-2b26ef33f883#... a little while ago.

Props to mods and/or OP for changing the link from a reposter's blog to the original source.

In order to give at least a portion of the comments in this thread context, the original post linked to a recent blog post which wholesale copied Matt's article with de minimis attribution.

It seems a little nefarious to blanket copy Matt Might's article, lead it with some emphasized and bolded statements that give the impression that the article is the work of the poster and distract from the plain text attribution.

It looks like a bunch of hand-waving that results in a person skimming the article believing that the poster is the author.

Yeah, the author of this repost doesn't even bother to fully follow all the license guidelines from Matt Might. Ironically this would be considered plagiarism in any academic setting.

Great ...

so just Matt Might's article.

great. :|

For completeness, here is the link to the original: http://matt.might.net/articles/phd-school-in-pictures/

Seriously? Did the mods also fuck with that link? I don't care if the other post was a ripoff I still would like to read it.

I don't know if the link above has been changed by mods or was a mistake, but here is the link to the original article:

https://thebittheories.com/masters-versus-ph-d-2b26ef33f883

The original content is licensed under a CC license [1], and now (it probably has been updated) the reposter is respecting the terms of it:

    1. Please attribute the original work to me (Matt Might) and link back to this page in your reproduction: http://matt.might.net/articles/phd-school-in-pictures/ as The Illustrated Guide to a Ph.D.

    2.When you attribute, please also link my name, Matt Might, to http://matt.might.net/

    3.And, don't forget the "Keep pushing," at the bottom!

So I don't see any reason why the original link can't be shared in thread's comments, especially because people should be allowed to see by themselves the original article in order to forge their own opinion on discussions happening here.

[1] http://matt.might.net/articles/phd-school-in-pictures/#licen...

> and now (it probably has been updated) the reposter is respecting the terms of it

The reposter still doesn't link back to the actual article as requested, though that's just a "please" clause. I guess naming (and even linking) the author and giving the title suffices to satisfy the attribution clause of the license.

> The reposter still doesn't link back to the actual article as requested, though that's just a "please" clause. I guess naming (and even linking) the author and giving the title suffices to satisfy the attribution clause of the license.

Indeed, I thought there was a link to the blog post, but there is not any. The situation is different than before though, as the real author is now at least credited.

By "original" I meant where the content originally came from, not the "original ripoff link". So no, the mods did not change that, I did link to Matt Might's site.

Reply on news.ycombinator.com