9

The process of paper writing is intended to communicate our proposed model to a peer. The peer is assumed to have solid knowledge of the particular topic.

If one comes up with a state-of-the-art model on some task in a domain, then the contribution is evident to the peer.

Then what is the purpose of writing the "related work" section?

For example, if I come up with an algorithm that can solve the Traveling salesperson problem in less time than all the previous algorithms, then the contribution is evident to any reviewer.

4
  • 30
    in less time than all the previous algorithms --- I would think that a justification for this claim would include a brief survey of the most recently published algorithms, and reference to a survey paper/monograph on the topic for an overview of less recent work, although maybe this would not be given as a specifically titled "related work" section as opposed to what one would ordinarily say (at least in math papers) in the first few introductory paragraphs. Oct 16 at 15:03
  • 32
    Very relevant: xkcd.com/2501
    – Stef
    Oct 16 at 19:11
  • 3
    I think your question is "what is the need for including citations and related work in a paper". This is pretty different from asking "what is the need for a related work section in a paper", so I suggest changing the title to match the question.
    – usul
    Oct 17 at 2:06
  • 1
    A scientific paper is not written for the peer review board, and the primary purpose of its content is not to impress the reviewers. Your primary audience is the scientific community (primarily within the same field, but not necessarily), and these people might be interested in looking into referenced and related work for all kinds of reasons. And your secondary audience are the gifted amateurs, and the general public. It's called a publication for a reason. You are publishing an account of your research to the world. Oct 19 at 9:14
53

The purpose of the related section is to establish context, within which your work is to exist, and against which it represents an advance.

That context is highly individual-specific, and varies greatly even across researchers working in the same field. It is unlikely that any two researchers will have exactly the same knowledge, context and approach to a topic, because they are likely to have engaged with it differently and had different journeys while exploring it. This could have different perceptions of the same idea. Therefore, the stated assumption - " The peer is assumed to have solid knowledge of the particular topic. " - is not very meaningful, because knowledge does not have neat boundaries.

Further, articles are written for posterity: a reader who uses the work after several years will need to know the context within which the work was carried out. State-of-the-art is constantly changing, and it is necessary to describe it for that very reason.

@Greenstick has correctly pointed out that this part of a paper can also facilitate the writing of literature reviews of different kinds (topical, systematic) subsequent to the publication of this paper. This would be a service to the community, at the very least.

3
  • 3
    Another aspect is that by citing similar work in the related section, papers can help facilitate the process of writing a literature review or similar type of article. Further, if you believe that knowledge should be generally accessible to the public (and posterity), providing those references can be very helpful for people new to the field or who have a more casual interest (journalists, hobbyists, amateur researchers, etc.)
    – Greenstick
    Oct 17 at 16:45
  • "experentially"?
    – Bergi
    Oct 17 at 20:18
  • @Greenstick, Bergi- thanks for the helpful additions, incorporated them. Oct 17 at 20:53
32

The process of paper writing is intended to communicate our [results] to a peer.

No, not to a peer. To a wider audience than your peers. Which is why you are expected to provide some context.

Also, a "related work" section helps both your peers and other readers clearly identify the novelty in your results: "Smith & al. formulated a model of phenomenon A, in this paper we present a model of the related phenomenon B"; "Smith & al. formulated a model with accuracy alpha, in this paper we provide a refined model with accuracy alpha/2"; etc.

2
  • I'm not sure if the word peer is the problem. To my ears, my fellow colleagues in my department, say, are my peers, but they do know the same things I know, even the ones in my field.
    – Kimball
    Oct 16 at 22:57
  • 4
    This. If you were to communicate a result or proposal to a specific peer whose level knowledge on the topic you know, there's indeed no reason to include a related work section - but then you'd just send a personal message to them. Not publish a paper in a journal.
    – Bergi
    Oct 17 at 20:23
20

If you published a paper with an algorithm substantially improving performance on the traveling salesperson problem, I'd be really interested in that paper myself.

I know next to nothing about the state of the field there, though. Related work would be very important for me.

These days there are so many papers published that it's difficult to keep up with all of them. Even in a narrow field, people generally sit on different edges of that field and bring in distinct background knowledge. Solutions to one problem may be motivated by solutions to other problems that are not so intimately related.

In a typical paper I read I doubt I've read more than 10% of the other papers it cites.

5

From my point of view it has to do with availability. Science is supposed to be a accessible to each and everyone, not only to an elite group that already knows all the background and details of your work.

While the main audience of your paper is people working on the same or related subjects, a more broader audience should principally not be excluded from accessing the contents of your paper. For that purpose, it is typical to provide some basic introduction to the methods you are using and the purpose of your work as well as a thorough review of the previous literature. The goal for this is not to explain things to your peers (that they probably already know) but to provide some degree of self-containedness to your work.

Of course, in practice it takes a lot of work for a potential reader to go through all of the previous literature and understand the relevance of your work. But keep in mind that every paper could be someone's first entry to a new topic and I think it's worthwile to provide them with the necessary ingredients to value your contribution.

1
  • Well, not each and every one, but let's say - a somewhat larger elite group of people...
    – einpoklum
    Oct 16 at 19:51
5

While other answers already noted that the related work section is targeted at an audience much broader than peers in your field of study, I also want to note that the related work section could be essential for peers with "average knowledge in your field" as well.

At times, it happens that a specific problem in field A is best solved via techniques from field B. However, when applying techniques from other fields, perhaps with some modification, it can be difficult to determine whether this application is still a novel result. The related work section can provide the context in order to determine this.

Another purpose of the related work section is an indirect argument for the relevance of your results: if you are studying the same problem as all those other published papers, then a priori it seems your paper would also be relevant if it adds something significant that is not already in those papers. (Of course, all of this depends on context. It is possible some other recent result renders the entire field, including your paper, obsolete)

Finally, a thorough discussion of related work can be valuable on its own. It can give an brief overview of a (sub)field, like a mini-survey. Sometimes, there are papers which I primarily read for this discussion, if there are no good survey articles or PhD theses.

1

One reason not yet mentioned is "crackpot detection". :-) Some areas attract attention from "motivated nonexperts". One item in the "crackpot detection checklist" is "does the paper cite relevant related work?". If it does not, that may be because the author (tends to be solitary) does not know related work, prior art so to speak. Then, how did they get to their result(s)? It is suspicious.

Also, a well written related work section does not just mention other works, but rather show how they relate and compare, how they are different. Precisely for an expert who knows such other work, this may be the key element in the paper. A "delta" from previous work, if you will.

Thirdly, your paper may be so good that someone else may read it, such as a student working on their thesis, someone working on an application, or some expert in a very different field which at the time was not related but know a connection was found and both fields are now related. And so on.

0

Let me interpret the question as "why do many papers cite irrelevant or barely relevant works"? The answer is: because citations are the currency of bibliometry-based career systems.

There is no good reason for each paper to do a complete review of the literature in the field, but there are perverse incentives to do so.

4
  • What's in it for the author? Why should they artificially boost someone else's citation count?
    – henning
    Oct 17 at 18:20
  • They might attract the attention of the authors they cite. They might get cited in return out of courtesy, even when no 'citation cartel' is in place. Oct 17 at 18:29
  • 2
    @henning: From what I've learned in Academia SE the past few years, reciprocity agreements (both explicit and implicit) are fairly common in some fields and/or places. It also might be due to increased suggestions from reviewers, seeking (possibly not even consciously) to increase the citation count of themselves and/or those they work with, over 2+ decades ago when citation counts were not used much (because the data was nearly impossible to come by). Oct 17 at 18:32
  • "What's in it for the author"? The reviewer won't let them publish unless they pay their dues to those that came before. Oct 19 at 17:19

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.