3

A common problem I encounter every time I do a literature search on a new subject is determining a reading order for papers. It feels akin to jumping into a prolific sci-fi fantasy series, but since there usually don't exist enthusiastic fan bases for a random subset of papers on numerical methods for PDEs, I'm always struggling to figure out how to get started. My default is chronological order (oldest to newest), but this frequently poses problems if the oldest paper is using outdated notation or technology. A newest to oldest reading order has the problem that it might assume background information from previous papers, which leaves me feeling like I have to keep jumping around texts and never really get a coherent picture of any of them.

Does anyone have better ideas for determining an "optimal" reading order for related academic papers? Perhaps a simple metric for ranking importance (but not citation counts, since this would skew towards older papers).

Note that when I say ``read a paper," I'm already employing a multi-stage technique where I read the abstract, conclusion, and introduction, then skim the text, and afterwards decide whether it's worth downloading and reading in detail. So the set of papers I have now, I've determined are worth really reading in detail.

Improve this question
4
  • 2
    Sadly, you probably need to read them all to figure out the order to read them in.
    – Jon Custer
    Oct 16, 2019 at 12:55
  • Usually the sensible alternative to starting at the beginning - if a topic has been going for a long time - is to find a good, recent review. Or a textbook.
    – Flyto
    Oct 16, 2019 at 17:41
  • Ideally, a well-written literature review from a single (recent) research paper should tell you everything you need to decide how to approach their particular research niche. Assuming you fit within the audience addressed by the journal the paper is published in.
    – A Simple Algorithm
    Oct 16, 2019 at 21:25
  • Although the tip to read a review paper is nice, it's ignoring a key point in my question: I've already deduced a set of papers, some of which may have been found through a review paper in the first place. Also, it's possible that there exists no review paper for my topic (or that I might be the one writing it).
    – user108403
    Oct 17, 2019 at 8:42

3 Answers 3

Reset to default
4

I often struggle with the same question. One approach which I find often gives good results is to start by skimming through all the papers (typically in chronological order) and use this first reading to decide on a more appropriate reading order, typically starting with the ones which seem more introductory or survey-like and ending with the more detailed ones.

Improve this answer
1
  • 1
    I was hoping to avoid another round of skimming, but you make a good point here. Thanks for the simple solution!
    – user108403
    Oct 17, 2019 at 8:46
2

Not sure if this applies to your particular field but in mine I used to have the same problem. What I discovered (too late in my PhD) is that textbooks and review papers tend to be very useful and can point you to the right papers so that's something worth trying if you haven't already. I also found it more usefult to start with more recent papers and then follow the chain of references back in time on aspects that I found important or interesting.

Improve this answer
1
  • Yeah, I've tried the approach of recent-to-oldest, but most of the time, I get frustrated not understanding the context-- a bit like watching the end of a show without having seen the earlier ones.
    – user108403
    Oct 17, 2019 at 8:44
1

I think a general answer to such a general question is not possible. It depends on those papers and you don't know how it depends until you have some understanding of each of them and how they fit together - the progression of the main idea(s).

But I also think you have explained a pretty good general strategy: get a general idea of each and then plunge in where it seems needed.

But it is also true that you don't need to completely understand every idea in each of the papers. You started with one of the papers for a purpose, I suspect. Read enough, and understand enough, to meet your purpose.

But also, take a lot of notes as you read. If you read on paper you can easily make marginal notes for some of it. But it is also good to write brief summaries of each paper, keyed to your original purpose.

Improve this answer
2
  • 1
    Just to add: As soon as you want to cite the paper, you definitely should understand it completely (way too many people cite something without even reading it properly...).
    – Dirk
    Oct 16, 2019 at 13:51
  • @Dirk I'd say you need to understand the other paper only to the degree that your claim about it implies you understand it. Though it would certainly be nice if people worked harder to pick the best possible reference, rather than simply one that fits.
    – A Simple Algorithm
    Oct 16, 2019 at 15:20

You must log in to answer this question.