diff options
Diffstat (limited to 'content/til/how-to-write-great-papers.md')
| -rw-r--r-- | content/til/how-to-write-great-papers.md | 55 |
1 files changed, 55 insertions, 0 deletions
diff --git a/content/til/how-to-write-great-papers.md b/content/til/how-to-write-great-papers.md new file mode 100644 index 0000000..6edd8f2 --- /dev/null +++ b/content/til/how-to-write-great-papers.md @@ -0,0 +1,55 @@ ++++ +title = "How to Write Great Papers" +date = 2026-04-22T20:42:27+02:00 +draft = false +categories = [ "TIL" ] ++++ +I watched a [video](https://www.youtube.com/watch?v=WP-FkUaOcOM) that was recommended by my professor. It contained a lot of interesting tips. + +## Don't wait: Write +It's better to follow: +>Idea -> Write -> Research + +Instead of: +>Idea -> Research -> Write + +Because by starting to write down immediately you can crystallize your idea. It forces you to be clear and focused.\ +Then it's easier to discuss it with other people because you have concrete material to reference. +>*Writing papers is primary mechanism of **doing** research.* + +## Identify your key idea +It's important understand what idea you want to convey. The goal of the paper is to *infect* the reader with this idea.\ +The paper should have just one, clear **idea**. If you have lots of ideas just write lots of papers. +>*Do not be afraid on developing apparently trivial ideas, often it turns out that they are quite interesting.* + +>*Even the greatest ideas are **worthless** if you keep them by yourself.* + +## Tell a story +You should capture the reader attention in the introduction, because most of the time the reader does not dive further if it's not engaged.\ +You should clearly state the problem and your solution in a way that keeps the reader interested in your idea. +>*"This seems a though problem!" "I wish I knew how to solve that." "Mmh, the solution is very clever!"* + +It's better to do it by giving a simple example and not the general case. +>*Act like you are trying to explain it on a whiteboard.* + +## Nail your contributions +Clearly point out what your contributions are: they drive the entire paper. Do not leave you reader wondering what they are reading about.\ +They should be refutable (written in a form that can be verified), so very precise and not vague.\ +Then you provide evidence to support each claim. + +## Related work: later +You should introduce your idea first because it should be the focus of the paper and the reader could get tired of reading other technical descriptions.\ +After explaining your contributions in detail you can expand on other approaches and, even better, make comparisons and point out the benefits and weakness of your idea. + +## Put your readers first +Don't be overly technical as it might exaust your reader or make them feel stupid.\ +Try to instill an intuition first (using a simple example) then dive into the details. The other way around does not work. + +## Listen to your readers +Ask as much people as possible to read your drafts. +Give them specific instruction like: +>"Read until you don't make sense of what is written then stop". + +This way you can point out where the paper is overcomplicated. + +>*Be truly grateful for criticism as well as prise.* |