I might offer a corollary as an practice to adopt (in the absence of the above possibility) -- learn to enable your research to have visible incremental gains on a known path every day, rather than hoping (hoping!) for some amazing breakthrough at the end.
Amazing breakthroughs have high risk, and make it highly likely you'll have a crisis when it doesn't happen.
And what I mean is that, even if you don't know what the answer is going to be at the end of your research, you must think about, or have an idea about, the format of what that amazing answer is going to be. Write the outline of your thesis and "ghost out" what the major charts will be. Write the intro sentences of each chapter -- what are they? (and I don't just mean the boring review of the field part, but your findings part)
You should know what major finding, plot, or table your research is going to output. What are the columns and rows of that table, or axes of that plot? How many data points are required? How many of them can already be guessed? Where is the surprise going to be? What is the conclusion going to be?
For most graduate research, the finding is not an amazing field-changing big bang. Few grad students are that fortunate. Or at least most of the bulk of the work will be of that sort. You should be able to predict what the answer is going to look like from past work, and the error bars.
Draw out the answer you're aiming for, now. If you can't even articuate what the answer will look like, you may be in for a bad time, so work on fixing that. It will also push you and your advisor to be specific about what the output of your thesis will be.
> You should be able to predict what the answer is going to look like from past work, and the error bars.
If you basically know the answer in advance, chances are you are not doing very interesting research.
The worst work I have done has been of this sort. The best has had me completely change my view of (mathematical) reality multiple times during the process.
In emperical fields as well I would assume the purpose of doing experiments is that you don't already know what the result will be. I accept that this may be a idealised or naive...
I'm not talking about the exact result, right? Just that someone engaging in research should know what the format of the plot/table/output should look like, how much work is needed to populate that table, and what kind of conclusions will come out of it.
If you (the general "you") are in industry breakthrough territory, you're an advanced student and my advice isn't for you. Otherwise, I think it's a good practice.
Fair. I think considering what possible conclusions could come is certainly important. I worried about your original post suggesting actually writing the conclusion in advance.
One of my professor's recommendation was to write the actual paper that you want to write, introduction, related works, methods, discussion, conclusion, leaving the figures, tables, etc. blank, before you actually start doing the experiments.
Many advisors don't push their students to do this, and they should. Or they haven't had enough practice to know to do this. Or they're not taking their responsibility seriously enough.
An effective PhD advisor/thesis isn't wandering the woods to find something. It's a guided coaching exercise, with an outcome in mind. If your advisor doesn't know this, maybe time to find a new advisor...
> An effective PhD advisor/thesis isn't wandering the woods to find something. It's a guided coaching exercise, with an outcome in mind. If your advisor doesn't know this, maybe time to find a new advisor...
You don't know me, but I really needed to hear this. I left grad school essentially because of a dearth of coaching. Thank you for framing this so succinctly.
There are multiple slightly different versions of the seminar (another is linked below that one), but unfortunately they all came out after I'd finished my PhD, and I didn't hear that advice from anyone else.
The process I arrived at after losing time on failed projects was basically "fail fast": find the simplest quickest way to demonstrate that your idea won't work, and do that. Then find the next simplest, and so on, until either it works or you've proven it doesn't and moved on.
Introduction, related works, methods, OK, but writing the conclusion before doing experiments sounds like an integrity issue (assuming you re-write the conclusion after the experiments -- else I'd label it fraud). What is the idea behind it?
As someone who was encouraged to do something similar, it was akin to creating a template manuscript for the project at hand: sketch out the experiments you plan to do, the likely results, and illustrative figures. This is meant as an exercise very early on in the project discussion phase, before any experiments have been performed, and while you're still figuring out the papers to read.
No. Obviously you rewrite the whole paper. Academics rewrite their papers multiple times before submitting them. The whole thing is a draft of a draft. The idea is to make your ideas very very concrete by writing them down as a paper.
Really annoys me when the word fraud is thrown around so freely, so I'd appreciate it if you don't.
It is not mentioned and was not obvious to me. Still, I'd postpone writing the conclusion to after you actually have the data. To do otherwise would be steering the experiments to get the expected results. I also reckon this depends on your field, how long the experiments take and your confidence in the results.
> Really annoys me when the word fraud is thrown around so freely, so I'd appreciate it if you don't.
I agree, 'academic dishonesty' is a more appropriate term here.
You typically have a finite set of overall results you expect to get, usually something like positive or negative. It helps to write out what you would conclude when you get either results. Think of this like a process that helps make your ideas clearer. And makes writing your final paper a lot easier.
This is amazing advice I hadn’t heard before, it rings very true.
I remember after my first paper was finished and submitted, and wow, it seemed easy to bang out papers after that, all you needed was four to five figures :)
My Doctoral Advisor had similar advice, though he suggested to add in aspirational figures as well. It can be helpful to help guide your initial readings and your experiments.
For most graduate research, the finding is not an amazing field-changing big bang. Few grad students are that fortunate. You should be able to predict what the answer is going to look like from past work, and the error bars.
Does it ever works that way for field changing big bang ideas? Every human achievement was built upon the back of previous achievement and works, either by you or someone else.
It's easy to forget that grad school is the start of a career rather than the summit, so the work just has to be publishable, not earth-shattering.
Most nobel/turing/fields level work doesn't happen while the researcher is still in graduate school, because a) it usually benefits from additional experience, knowledge, and resources and b) it can sometimes take years to complete.
It also might not be obvious to a dissertation committee (or anyone else) how field-changing certain work actually is until many years have passed, and they might dismiss it early on. Consider Tim Berners-Lee, who was actually a Physicist but received the Turing award because the web (and HTTP/HTML) turned out to have much greater impact than predecessors like FTP, NNTP, PLATO, HyperCard, Director, Xanadu, Intermedia, Gopher, AOL, etc..
Another reason that more mundane research tends to work better is that groundbreaking work is often hard to publish (and get support/funding for), particularly if it isn't completely solid and seems to contradict or supersede established theory or practice.
However, if you are choosing between multiple projects that all seem doable within the allotted time frame, picking the one with higher positive impact on the field (and/or elsewhere) may be a good approach.
Amazing breakthroughs have high risk, and make it highly likely you'll have a crisis when it doesn't happen.
And what I mean is that, even if you don't know what the answer is going to be at the end of your research, you must think about, or have an idea about, the format of what that amazing answer is going to be. Write the outline of your thesis and "ghost out" what the major charts will be. Write the intro sentences of each chapter -- what are they? (and I don't just mean the boring review of the field part, but your findings part)
You should know what major finding, plot, or table your research is going to output. What are the columns and rows of that table, or axes of that plot? How many data points are required? How many of them can already be guessed? Where is the surprise going to be? What is the conclusion going to be?
For most graduate research, the finding is not an amazing field-changing big bang. Few grad students are that fortunate. Or at least most of the bulk of the work will be of that sort. You should be able to predict what the answer is going to look like from past work, and the error bars.
Draw out the answer you're aiming for, now. If you can't even articuate what the answer will look like, you may be in for a bad time, so work on fixing that. It will also push you and your advisor to be specific about what the output of your thesis will be.
Your future self will thank you.