This is something I used to think about too but how do you do research without learning about prior work and how do you learn about prior work without an advisor to guide you?
Follow the paper trail. Find a research topic you are interested in and find an interesting paper on Google Scholar.
Then look up some of the references and referencees or stuff from the same author. Keep doing this one tactic and you'll easily find years worth of reading to do.
Besides finding the abstracts you'll need access to the research. This used to be an issue but luckily now we have sci-hub.tw and paperdownloader.cf
All disciplines have journals, and many are either public access or available on sci-hub. If all else fails you can get a limited JSTOR subscription for only a moderately outrageous sum.
Mostly you do reaearch by being interested in something. If you need someone to tell you what to be interested in, a PhD may not be for you.
A good supervisor will tell you to be interested in things you may not have considered. But empirically, most supervisors will steer you in the direction of their own interests.
These may or may not match your own interests. The mismatch us at least as likely to be a bad thing as a good one.
In my own experience, I decided to work independently instead of starting a PhD. There are only a couple of directly relevant journals, and I literally skimmed every issue, reading and taking notes on the papers that counted as prior art.
Those papers often quoted other papers outside the immediate domain, so I followed them up - and that’s how you start.
I have a pretty good idea of the directions I’d be steered in if I was being supervised, and near certainty that those are not the directions I want to explore.
> Mostly you do research by being interested in something. If you need someone to tell you what to be interested in, a PhD may not be for you.
Maybe I'm misinterpreting but this comment seems a bit condescending.
The thing is, it's not really research unless it's something new, and you can't know if something is new unless you know what already exists.
> All disciplines have journals, and many are either public access or available on sci-hub. If all else fails you can get a limited JSTOR subscription for only a moderately outrageous sum.
Sure you can have access to journals but how do you even know what you should search for? I suppose this is sufficient if what you want to do research in is something that is currently mainstream?
> Sure you can have access to journals but how do you even know what you should search for? I suppose this is sufficient if what you want to do research in is something that is currently mainstream?
I "research" stuff all the time, do a google search for some term and find an interesting paper then go through the references that seem interesting. Works better than you'd imagine, I was looking up CESK machines and down through the rabbit hole I eventually found the original paper with the non-obvious name of The Calculi of λ-v-CS Conversion: A Syntactic Theory of Control and State in Imperative Higher-Order Programming Languages. Honestly, those math-heavy CS papers tend to make my head hurt though.
Don't think there's a whole lot of people who would claim lambda calculus is mainstream but I can spend hours upon hours finding stuff to read about and most of (or all) the papers are easily downloaded off the author's website.
