Advice to a Beginning Graduate Student - Manuel Blum (CMU) | Hacker News Reader

Advice to a Beginning Graduate Student - Manuel Blum (CMU)

cs.cmu.edu

56 pointschromophore32 comments

There's a lot of filler in there, but one advice that I found interesting was to write right back what you're reading (I take it to refer to technical material and not exposition - e.g. proofs and definitions). Anyone has any experience with trying that?

(I've tried this with math books before, but it always felt too tedious to keep up. But I suspect I may have tried to hard, going for a nicely formatted shortened exposition of the entire material; I didn't try to simply jot proofs and formulas down as I was reading them).

Yeah this is something I do frequently. I discovered it while studying for my qualifying exam. When I first took the exam, I failed using the same study methods I used in undergrad. I realized the scope of information I had to know was much too broad to be able cram or understand through past homework problems. So as I went through my study materials I wrote down the material after reading in designated sections, keeping the summaries to just one page. Then during further review I used these pages of notes. The second time I took the exam I got 100%.

When I would study for exams, I found the process of writing notes would be almost more useful than the process of reviewing them. The process of writing notes made me boil the concepts down into more discrete forms, and the physical action of writing helped with the memorization. A few days before an exam, I would often find myself re-reading my old notes and condensing them down into briefer notes, eliminating full explanations of concepts in favor of keywords or mental links to the full note sets.

The net result was that when I was in my exams, I would find myself mentally looking at my notes, and finding a formula or something would just be a matter of 'reading' them.

When I went to grammar school in England, many of the other students seemed to do this.

I used to write summaries for all study materials as I was reading. It helps get a better broad view, I think, and it also encourages writing down any related thoughts that come to your head.

I have found that it is best to summarize in a question-answer format, much like a FAQ. Having questions separated from the answers is useful later when revising and the questions are more open-ended than a straight summary.

There's a website that a friend and I set up to store exactly such question lists (with answers): http://www.problemata.com . It's rough, but already usable.

Your question-answer format reminds me of Cornell Notes.

I found this article extremely insightful and decided to prepare it for the print so that I could read it more carefully and add some notes on the margins.

While I was doing it, I realized how awfully true was the fact that science people _can't_ write.

No disrespect to Mr. Blum, but really, what is going on with the writing in this article? New line every several words, unconsistent emphasis, various writing styles, quick abbreviations in some places, but not in others, inconsistent division in paragraphs, several misspellings.

When I was a kid, my father (an EE guy) used to tell me the importance of writing while reading. I mostly ignored that in my career and now a very smart person(Manuel Blum) has validated it. I hope there aren't many things (as told by my father) that I would want other smart people to validate ;(

Did not realize this: "the median of n integers can be found with just O(n) comparisons."

Must figure this out!

There are a couple linear-time selection algorithms, some randomized and some not. An example of the latter is described here ( http://www.ics.uci.edu/~eppstein/161/960130.html ). The constant factor is non-trivial here (<= 24n comparisons for this one) but for large datasets it will eventually be faster than sorting.

"Scrolls must be read like the Torah from one end to the other." - Associating Islam with being outdated can be good for your demographic I guess.

Er... what? Are you perhaps confusing the Torah (Jewish scripture) with the Koran (Islamic scripture)?

I used to love to read these essays on how to be a good and productive grad student... until I came to grad school. Same with phdcomics.

LOL "TriinT" I know what you mean. :)

I think I'll keep reading PHD comics, just because it's really funny. I don't relate it to grad-school as such. Just the whole exaggerated sense it is written in.

phdcomics is funny, but not if you're phd student =)

They're still funny even years after I finished my PhD :)

But were they funny when you were doing it? I mean they're funny, but they're so negative I don't like to read them. At one point another PhD student printed out and put up one of them (it was the one which had a time plot about how you get more and more demotivated over time), and I actually took it down.

I read them all the time during grad school, as a form of procrastination.

One of my fellow grad students related a tale in which he was his advisor was telling him to work on something and he wanted to know whether his advisor was going to help. His advisor's reply: "Let me consult Ph.D. Comics." (Said advisor has a sense of humor.)

This is actually quite good advice. It's not overly preachy and it's not overly poetic.

Advice to a beginning graduate student: do a start up instead ;)

Why?

why not?

Well, perhaps because some might be more interested in say digging in Egypt to undercover archaeological sites, or maybe they interested in being a doctor, or perhaps a lawyer, or maybe they would rather do research on how they can make machines think, or a million of other things which have not much to do with starting your own business, to start with anyway.

So will you now tell me why they should start a business instead?

Because it will take you years and years of effort just to get the advanced degree that will enable you to try many of these lines of work, [1] at which point you may discover that you don't especially like them.

I can tell you about this experience at great length if you really want. ;)

(Dabbling in medicine or law is particularly expensive, because... it's expensive. You have to go into massive debt, on top of your potentially massive college debt, before you figure out that medicine or law can be a really miserable line of work. At least I didn't do that -- my years of career trial-and-error were cash-flow positive, though I didn't exactly live like a prince on my grad-student and postdoc salaries...)

Meanwhile, you don't have to do years and years of academic hoop-jumping before learning whether or not you'd enjoy starting a business, or working at a startup. (Or working at a big company, for that matter.) You can learn a lot just by trying it. You don't even have to wait for graduation.

---

[1] Beware: You may think that you can learn whether you'd enjoy (e.g.) archaeology by volunteering to help on a dig, or by being a grad student in archaeology. And that is a good idea. But enjoying life as a grad student or intern is not the same as enjoying a career in a given field. In many fields, professors spend most of their time raising money, managing people, playing political games, and writing papers and proposals, not digging holes, performing calculations, or building neat stuff in the lab.

I think you're leaving out a very important aspect of graduate school, which is simply the ability to experiment and find out what you truly love to do. You can be pigeonholed into a business you don't particularly like, just like you can with a graduate project. But I feel like graduate school offers you much more flexibility and time to figure your shit out, at which point by the end you atleast have a solid credential to back up your work ethic.

I am a new grad student after spending a few years in industry. My advice for people who don't know what they want to do is to go straight to work. Don't go to grad school. It is much nicer to figure out what you want to do while earning $60k+ than $18k.

I don't feel like money is the issue though. When you're working for a company your job is to produce something for them--something that moves the company forward. In grad school, your work is much more selfish. You have ample time to think about problems in ways that only you can solve.

Remember, though, that if you enjoy life as a graduate student, there's nothing forcing you to become a tenure track professor and deal with all the management/politics.

And if you enjoy being a grad student, then there's little disincentive (other than opportunity cost) not to do that, learn a lot, and come out with a PhD that will make it easier and better paid to find similar work out there.

(n.b. I am assuming that one is getting a PhD in the sciences, or any field where there is reasonable amounts of funding available and one need not take loans to continue being a student.)

That's why I want to be financially independent, diff than a startup, tho.

Doing a startup might get me financial independence, but only at the expense of possibly years and years of grinding effort, given the normal stats.

However, if my standard of living is low enough, and I pick my location well, I should be able to become financially independent in not too many years at a normal job.

Additionally, if you put a few years into the military, you could even get a free grad degree.

Because its more work than getting a master's and also much higher risk.

This is something I thought about a lot. I'm currently a grad student at a great school and even though I sometimes get frustrated with classes, I've never regretted it. There is a lot more to grad school than just taking classes and doing research.

For me, the main attraction and benefit of being in grad school was being immersed in an environment filled with smart people wanting to "do more". My school's heavy emphasis on entrepreneurship has also energized me about creating a startup. The caveat is that one must find a way to attend grad school for free (which actually isn't hard at all).

Reply on news.ycombinator.com