How to Find Research Problems | Hacker News Reader

How to Find Research Problems

cs.jhu.edu

22 pointsplinkplonk14 comments

Interesting choice of analogy (research as "bone finding", which is research itself).

The thing is, researchers rarely set out to find a T.Rex -- at least, not in a way that their career success is predicated upon finding one. Good scientists make a career of the "small bones", because they know that they'll stumble upon the "big bones" only once in a while. Truly great researchers will assemble a whole skeleton over the course of a long career.

A few, exceptionally lucky people will discover a whole mammoth on the first day of their expedition; counting on this is like counting on winning the lottery. Bad researchers set out to find something big, and refuse to "waste time" with the "little bones" along the way.

"Interesting choice of analogy (research as "bone finding", which is research itself)"

I think the "bone finding" is less important (for the purposes of the analogy) than the distinction between actively looking for small vs big results.

I know there are many PhDs on this board so I'll ask something I have always been curious about.

Is it possible for a hacker to teach himself to do research?Is "research" fundamentally different from hacking? ( in terms of learning how to do it; most good hackers are self taught)

Or is going though a structured process or some kind of apprenticeship to a good researcher a necessity?

From my limited experience, I'd say most hacking is using the results from research to get shit done - Not that hacking isn't great, its just not generally research. However, it'd be pretty hard to "hack" a new filesystem without a research component. But designing a website doesn't usually involve much research (learning a framework isn't research in the sense i'm trying to describe). Designing a new web framework could be research because you need to test different ways of doing things and people might use your results.

Where normal design and normal engineering turn into research isn't the easiest thing to define. Man, this comment isn't working out as well as I'd hoped.

I don't think that "hacking" is fundamentally different than research, in the sense that they're both forms of problem solving. The difference is one of purpose and mind-set: a good researcher is more interested in proof-of-new-concept than in implementation and polish.

"Is going though a structured process or some kind of apprenticeship to a good researcher a necessity?"

Theoretically? No. Ideas are ideas. Anyone can have good ones.

Practically speaking? Yes. If nothing else, you need the time and training to become familiar with your field, the terminology and protocols in common use, and the boundaries that define interesting problems. Furthermore, most cutting-edge research is expensive; you need the apprenticeship before anyone will take you seriously enough to give you the money needed to do interesting work.

you need the apprenticeship before anyone will take you seriously enough to give you the money needed to do interesting work.

Hmmm... perhaps not unlike venture capital a few years ago?

I wonder if we'll someday see some sort of YCombinator version of the NSF.

Furthermore, most cutting-edge research is expensive; you need the apprenticeship before anyone will take you seriously enough to give you the money needed to do interesting work.

This is a key observation. One of my advisors put it thus:

A PhD is just a union card. What you do with it is up to you.

You can use it to establish credibility in certain lines of work; you can use it to command a higher salary in industry; you can simply apply the things you've learned in some totally unrelated field. A surprising number of people choose the latter.

Yeah, the union-card line is a sad bit of truth -- I can't imagine a union card that is harder to receive, which carries fewer benefits.

I would be applying for government grants right now if I had said union-card.

Don't sweat it...with funding rates in the single digits, you're not missing much. ;-)

I heard a presentation once from a consultant who people hired to help get government grants. Does the funding rate go up a whole lot if you hire someone to rewrite your proposal?

Does the funding rate go up a whole lot if you hire someone to rewrite your proposal?

Depends on their name, the journals and review boards they sit on, and their current interests vice yours. The clever thing to do if you have good juice is simply to get their name on it.

I work with some people who rarely get turned down, and others (in the past) who often have. It's more fun to work with the former, if you have the choice, but sometimes the topics end up being a little less interesting.

Grant writing is such an insane time sink...

I don't doubt that most competent self-taught hackers could learn to do CS research if they were interested. A passionate interest in the learning new things, without waiting for someone else to teach you, is fundamental to teaching yourself to hack, and that same attitude is the main component of doing research. imho the biggest difference between hacking and doing 'research' is the mental shift from building something to creating new knowledge, whether it is a way of solving a problem, or something more abstract.

The most valuable things I've gotten out of the formal process has been the built-in mentorship provided by a good supervisor, the community of fellow students, and the opportunity to work on something with no immediate economic value. If you have the time/money, and don't mind harassing strangers via email you can get some of that without going to school for it. Still, if that mentorship, community, and money wasn't valuable, no one would be applying to ycombinator.

>Is it possible for a hacker to teach himself to do research?

I don't know a single good researcher who doesn't fit that description.

However, I do recommend some flavor of apprenticeship (1). A good mentor will suggest useful bones to search for, and act as a tour guide to stuff that already exists.

A comment like "did you know that in the 80's, $BIGNAME tried to do $X with $Y (not $Z), take a look" can save you LOTS of time.

(1) The useful part of the PhD is the apprenticeship to your advisor. If you can get the apprenticeship without the PhD, take it.

I don't like the bone finding analogy much, actually. Heres why: in most sciences, big bones are made from small bones.

(I'm sure there is some flavor of this in paleontology as well. But I don't like the analogy because to people like me who know very little about bone hunting, it sounds very wrong.)

Let me give an example. I'm hunting a big bone right now: clearer MRI images from low resolution data.

To do that, I need the following small bones:

1. Understanding why low resolution data sucks, and finding unused information. (This is something along the lines of literature review.)

2. Small bone: How to optimally extract that data?

3. Small bone: Result of step 2 gives me an unordered list of points. All I know is they are samples from some set of smooth closed curves, and I also know the tangents to the curve at each point. How do I get the curves?

3a. Smaller bone: some points are wrong. Can I figure out which, and ignore them?

3b. How much of this can I prove?

4. Small bone: If I knew these curves, how do I use them to get a picture that a radiologist will like?

(Step 3 gets extra emphasis, since that's what I'm doing right now. ).

Every scientist is a small bone hunter. The only question is whether they are looking for small bones to build one particular big bone (like me), or whether they are search for small bones in some region where they like the scenery.

Reply on news.ycombinator.com