51–60 of 159 posts
See also "Why does the C++ standard ship every three years" (as opposed to ship when the desired features are ready):
https://news.ycombinator.com/item?id=20428703 (2019-07-13, 220 comments)
Earlier quoted context omitted.
> You have to take a topic you find interesting and read all possible related work in it This is definitely the wrong way of going about a research project, and I have rarely seen anyone approach research projects this way. You should read two or at most three papers and build upon them. You only do a deep review of the research literature later in the project, once you have some results and you have started writing…
The usual justification is that if you don't do at least a breadth-first literature review, you can get burned by missing a paper that already does substantially what you do in your work. I've heard of extreme case where it happens a week before someone goes to defend their dissertation!
Earlier quoted context omitted.
Day 1: We aim to demonstrate the effectiveness of an existing industrial catalyst in a novel application that has not seen commercial usage, potentially lowering cost of production of precursors for essential medications Day 400: Having thoroughly described a universal theory of everything, we set out to build an experimental apparatus in orbit at a Lagrange point capable of detecting a universal particle which acts…
Damn, that's an incredible amount of progress in just 400 days
Earlier quoted context omitted.
The usual justification is that if you don't do at least a breadth-first literature review, you can get burned by missing a paper that already does substantially what you do in your work. I've heard of extreme case where it happens a week before someone goes to defend their dissertation!
Excuse my naivety, but isn't it good if the same results get proofed in slightly different ways? This is effectively a replication, but instead of just the appliance of the experiments, you also replicate the thought process by having a slightly different approach.
Earlier quoted context omitted.
The usual justification is that if you don't do at least a breadth-first literature review, you can get burned by missing a paper that already does substantially what you do in your work. I've heard of extreme case where it happens a week before someone goes to defend their dissertation!
Excuse my naivety, but isn't it good if the same results get proofed in slightly different ways? This is effectively a replication, but instead of just the appliance of the experiments, you also replicate the thought process by having a slightly different approach.
Replicating existing results doesn't meet that criteria so unknowingly repeating someone's work is an existential crisis for PhD students. It can mean that you worked for 4-6 years on something that the committee then can't/won't grant a doctorate for and effectively forcing you to start over.
Theoretically, your advisor is supposed to help prevent this as well by guiding you in good directions, but not all advisors are created equal.
Earlier quoted context omitted.
Excuse my naivety, but isn't it good if the same results get proofed in slightly different ways? This is effectively a replication, but instead of just the appliance of the experiments, you also replicate the thought process by having a slightly different approach.
For the humanity? Yes, it's generally good. For that particular researcher's career? Not really. Who wants to pay for research into something that's already known?
Incidentally, this describes what I believe to be the great difficulty of PhD research. You have to take a topic you find interesting and read all possible related work in it, which tends to result in significant scope creep as you realize just how much there is that already does you want to do. Having exhausted your initial energy and excitement for the project, you have to force yourself the remaining 20-30% of he…
You’ll almost never see a PhD thesis that has anything particularly interesting, novel or directly applicable to the sciences.
Prototype a minority of the time. Research a majority of the time. At some point the ratio flips as research fades out and producing increases.
Project where the sole user is you in your kitchen? Sure, hack it together.
Project where you actually want other people to use the product? A research phase matters and helps here.
Consider what the goal is and the amount of effort to invest typically becomes more evident.
I'm _exactly_ in this situation right now with a side project. It's in a field that I have little experience with (Information Retrieval). So there is obviously prior art that I could learn from or even integrate with. This article motivates me further to learn things by focusing on building my own and peek into prior art as I go, when I'm stuck or need ideas. Recently a Clojure documentary came out and the approach…
That was also on my mind thanks to the documentary. Then I followed up with "Easy made Simple" and "Hammock Driven Development", and it makes me want to learn Clojure.
Clojure documentary on CultRepo channel: https://www.youtube.com/watch?v=Y24vK_QDLFg
Simple Made Easy: https://www.youtube.com/watch?v=SxdOUGdseq4
Hammock Driven Development: https://www.youtube.com/watch?v=f84n5oFoZBc