4 ms·

I know this might be a silly question but what exactly do you do in an academic research position? How do you choose a topic to look at? What do you study on th

I know this might be a silly question but what exactly do you do in an academic research position? How do you choose a topic to look at? What do you study on this topic and what is the output supposed to look like? For someone who might eventually want to do a PhD in the space of software, I don't exactly know what you'd do. I normally just take a problem, and write software to fix it. How is the process changed when I want a degree from it? I'm assuming I can't just bring in a flash drive and say "Yea I did some reading and did some stuff, here is the implementation of my new X for the Linux kernel. It speeds things up by a factor of BIG_NUMBER."

light_hue_1 10y ago
> How is the process changed when I want a degree from it? When you understand the answer to this question you're well on your way to a PhD. The PhD is not about implementing something. Although an implementation is sometimes/often a part of it. The real problem is to read a lot, get a grasp on what people understand about a topic, think about it, and say something new and interesting. That might mean coming up with some new algorithm, or a cool observation that solves a problem, or a theorem, or an experiment on humans that shows something interesting, etc. Finding good questions, learning how to ask them in the right way, and how to squeeze out what's actually doable and important from them, that's the point of the PhD. When you can do that from beginning to end, you're ready to graduate. Historically, this was what a PhD thesis was. Your first end-to-end piece of research. As for the first few questions. That's why you have a PhD adviser. Someone who can help you pick your first few questions. Guide you on the path to getting answers, finding new questions, getting a handle on what's known, etc. My biggest advice is find a topic you're interested in, read a few papers about it (google scholar is your friend), find a few people that work on that topic, read 1 paper from each, and when you apply for grad school in their department say something meaningful about what they've done (don't suck up, we know and it doesn't impress anyone) and what you would do instead/in addition. When people pick grad students they like to know that the student they're picking is smart, motivated, and knows enough to say something interesting about their work. It sounds daunting but it's doable. Particularly if you do summer research while in undergrad or do a good masters thesis. Good luck!
- gravypod 10y ago
  The way you put it, it sounds like you can do what ever you'd like so long as you come up with something a little extra. Are there sample "this is what a thesis for a PhD should look like" papers? Where is the standardization?
  - chrisseaton 10y ago
    It needs to be more than 'a little extra'. It needs to advance the state of the art. It needs to have a new idea in it. It can't just be a better implementation of something we already know how to do. There is no standardisation. The level you are assessed against is set by your examiner. Don't focus on the thesis or the software - those are just ways to express your novel idea, which is what is being assessed. But yes you can do whatever you want, as long as you have a good idea and can show it works.
    - BellsOnSunday 10y ago
      When it comes to being examined there are also the prerequisites to advancing the state of the art, which include demonstrating you've learned how to do research and have deep critical knowledge of the field you're working in (so you know what needs to be done to move it forward).
  - light_hue_1 10y ago
    The contribution has to be pretty substantial. And it can't be just one piece of work. The standards around these parts are something like 4+ papers (around 1 paper per year for 5-6 years) preferably including one or two solid journal papers all in top-tier venues. Of course not everyone meets this standard but basically everyone gets close. That's a lot work. It's your full time effort for those 5-6 years, 6+ days per week, 10+ hours per day. What might not be clear above is that once you can do research on your own you're close to done, but to get there you have to do a lot of work. Sure. Best way to get an idea is to just look up previous theses. They're basically all available online. "phd thesis cs" on google is a good place to start and just filter out the non-top-tier departments. Although these days many people just take their publications, reformat them, and that's their thesis (this is more common in the US and much less common in Europe). This saves a lot of time and makes people a lot more productive.
munin 10y ago
> I know this might be a silly question but what exactly do you do in an academic research position? Are you a professor? Then you do whatever you want. If you're not a professor, you do whatever the professor wants. > How do you choose a topic to look at? The intersection of what interests you, what you can actually execute on, and what you can publish. > What do you study on this topic and what is the output supposed to look like? You figure things out and tell a story. > I'm assuming I can't just bring in a flash drive and say "Yea I did some reading and did some stuff, here is the implementation of my new X for the Linux kernel. It speeds things up by a factor of BIG_NUMBER." That's a systems paper! You have to write up a bunch of stuff about what your implementation does and why it's cool, why it's correct, why it's safe, perhaps prove those things, and empirically evaluate BIG_NUMBER in a well controlled and designed experiment.
unfamiliar 10y ago
I'm not in CS but just writing software is not good enough. Software PhDs from my understanding would require stuff like developing a new provably-better algorithm, not just picking one off the shelf and applying it to a kernel driver, or developing CS theory in some way such as developing a new type system with useful properties not found in other type systems. In my field you would have to develop new physical modelling, apply it to an industrially relevant experimental case and show that it is indeed an improvement. The tricky part is figuring out a new model that none of the thousands of smart people before you have come up with.

light_hue_1 10y ago

> How is the process changed when I want a degree from it? When you understand the answer to this question you're well on your way to a PhD. The PhD is not about implementing something. Although an implementation is sometimes/often a part of it. The real problem is to read a lot, get a grasp on what people understand about a topic, think about it, and say something new and interesting. That might mean coming up with some new algorithm, or a cool observation that solves a problem, or a theorem, or an experiment on humans that shows something interesting, etc. Finding good questions, learning how to ask them in the right way, and how to squeeze out what's actually doable and important from them, that's the point of the PhD. When you can do that from beginning to end, you're ready to graduate. Historically, this was what a PhD thesis was. Your first end-to-end piece of research. As for the first few questions. That's why you have a PhD adviser. Someone who can help you pick your first few questions. Guide you on the path to getting answers, finding new questions, getting a handle on what's known, etc. My biggest advice is find a topic you're interested in, read a few papers about it (google scholar is your friend), find a few people that work on that topic, read 1 paper from each, and when you apply for grad school in their department say something meaningful about what they've done (don't suck up, we know and it doesn't impress anyone) and what you would do instead/in addition. When people pick grad students they like to know that the student they're picking is smart, motivated, and knows enough to say something interesting about their work. It sounds daunting but it's doable. Particularly if you do summer research while in undergrad or do a good masters thesis. Good luck!

gravypod 10y ago
The way you put it, it sounds like you can do what ever you'd like so long as you come up with something a little extra. Are there sample "this is what a thesis for a PhD should look like" papers? Where is the standardization?
- chrisseaton 10y ago
  It needs to be more than 'a little extra'. It needs to advance the state of the art. It needs to have a new idea in it. It can't just be a better implementation of something we already know how to do. There is no standardisation. The level you are assessed against is set by your examiner. Don't focus on the thesis or the software - those are just ways to express your novel idea, which is what is being assessed. But yes you can do whatever you want, as long as you have a good idea and can show it works.
  - BellsOnSunday 10y ago
    When it comes to being examined there are also the prerequisites to advancing the state of the art, which include demonstrating you've learned how to do research and have deep critical knowledge of the field you're working in (so you know what needs to be done to move it forward).
- light_hue_1 10y ago
  The contribution has to be pretty substantial. And it can't be just one piece of work. The standards around these parts are something like 4+ papers (around 1 paper per year for 5-6 years) preferably including one or two solid journal papers all in top-tier venues. Of course not everyone meets this standard but basically everyone gets close. That's a lot work. It's your full time effort for those 5-6 years, 6+ days per week, 10+ hours per day. What might not be clear above is that once you can do research on your own you're close to done, but to get there you have to do a lot of work. Sure. Best way to get an idea is to just look up previous theses. They're basically all available online. "phd thesis cs" on google is a good place to start and just filter out the non-top-tier departments. Although these days many people just take their publications, reformat them, and that's their thesis (this is more common in the US and much less common in Europe). This saves a lot of time and makes people a lot more productive.

munin 10y ago

> I know this might be a silly question but what exactly do you do in an academic research position? Are you a professor? Then you do whatever you want. If you're not a professor, you do whatever the professor wants. > How do you choose a topic to look at? The intersection of what interests you, what you can actually execute on, and what you can publish. > What do you study on this topic and what is the output supposed to look like? You figure things out and tell a story. > I'm assuming I can't just bring in a flash drive and say "Yea I did some reading and did some stuff, here is the implementation of my new X for the Linux kernel. It speeds things up by a factor of BIG_NUMBER." That's a systems paper! You have to write up a bunch of stuff about what your implementation does and why it's cool, why it's correct, why it's safe, perhaps prove those things, and empirically evaluate BIG_NUMBER in a well controlled and designed experiment.

unfamiliar 10y ago

I'm not in CS but just writing software is not good enough. Software PhDs from my understanding would require stuff like developing a new provably-better algorithm, not just picking one off the shelf and applying it to a kernel driver, or developing CS theory in some way such as developing a new type system with useful properties not found in other type systems. In my field you would have to develop new physical modelling, apply it to an industrially relevant experimental case and show that it is indeed an improvement. The tricky part is figuring out a new model that none of the thousands of smart people before you have come up with.