21

I'm collaborating with another person on a public-key cryptosystem that we intend to publish. I could only think of two references (one to the 1976 Diffie-Hellman paper, and one to SHA-3 FIPS-202), even though most of the contents are completely original.

I put my content onto Grammarly Premium and it only found 2% of so-called plagiarism - so called because the source of plagiarism is actually a low-quality prank question on a Q&A site that contains a series of digits 9 and asks its sum, that happen to collide with our experiment data.

What can I do to improve our paper?

7
  • 3
    To add to the good answers & comments here, consider that your work may(will?) be read by someone who has a working understanding of the specific field. They may require supporting references to read up and follow your ideas in a more wholesome manner. Apr 4 at 12:54
  • 45
    There's never a reason to run your own work through a plagiarism checker, unless I suppose it's an assignment and you want to check it against the professor's checker for false positives. Apr 4 at 15:42
  • 9
    @AzorAhai-him- But many journals use plagiarism checkers too, so the same reason of checking for false positives could apply for any publication.
    – nanoman
    Apr 5 at 2:17
  • 3
    @nanoman I've never heard of such a practice. Apr 5 at 2:46
  • 2
    @AzorAhai-him- See here.
    – nanoman
    Apr 5 at 2:50
110

Citations aren't just about avoiding plagiarism. They are more about embedding your work into a scientific context. Every paper should explain in an introduction section

  • why this work is relevant, and
  • how this work differs from previous approaches.

It is extremely rare that an approach is so novel that there is no meaningful prior work that can be referenced – many papers just apply an existing concept A to an existing problem B and analyze whether this is useful, or extend an existing approach to have some desirable properties. If you can't reference such prior work, that is usually an indication that you don't understand the state of the art in your field. Such papers are unlikely to contribute to the field, and are very likely to be rejected by journals or conferences.

Most of the citations are usually in the introduction section of a paper, which is where related work is discussed. In contrast, it is somewhat normal that there are fewer references in the section where you describe your novel approach. You would still reference existing building blocks and methods of your approach. E.g. an experimental paper in CS might reference relevant software libraries that were used.

I am not an experienced author and know how hard it is to break into a field. It's difficult to get an overview of the state of the art in your field. However, this is a crucial part of writing a good paper – otherwise, it is quite likely that you're either doing something that has already been done, or that you're flaunting the methodological conventions of the field. For example, there might be existing testing methods or benchmark problems that every serious paper in that field should use. (If those existing methods were unsuitable, a paper about that could be useful as well!)

62

It depends what you're actually saying in the paper. If all you're saying is "We invented cryptographic algorithm X.", then you may not need any citations at all, but that would be a really uninteresting paper. A more interesting paper might say "Cryptography is necessary in the modern world because A, B, and C. We invented cryptographic algorithm X. On benchmark test M, cryptographic algorithm X performs better than cryptographic algorithms Y or Z; whereas on benchmark test N, cryptographic algorithm X performs better than cryptographic algorithm Y but worse than cryptographic algorithm Z. However, cryptographic algorithm Z is patent-encumbered." In that case, you'll need to cite the people who discovered the implications of A, B, and C, the inventors of algorithms Y and Z, the inventors of tests M and N, and the patent on algorithm Z.

9
  • 3
    Excellent answer with a practical example
    – Gimelist
    Apr 5 at 5:47
  • 12
    Writing "Cryptography is necessary in the modern world because A, B, and C" in a paper to be published in a cryptography journal is highly redundant. Which of the journal readers don't know this already? There are way too many papers in CS that do this, and it is soooo tiresome. Apr 5 at 5:57
  • 19
    @CrisLuengo It might be better if they wrote "Cryptography with properties I, J and K is necessary ..."
    – muru
    Apr 5 at 8:30
  • 12
    @ChrisLuengo I, for one, aspire for my papers to reach a wider audience than just specialists in my own field. To achieve that, I have to tell readers on the first page why they should care enough about a paper in my field to continue to the second page. Apr 5 at 8:58
  • 1
    @CrisLuengo From a historical perspective, the A, B and C citation is also relevant. You never know where the paper will be discovered by someone some hundred years later and the mention that ties it to the rest of the human knowledge at the time is pretty useful, as it allows the researchers to put things in context. Hell, it is useful today as it allows automated systems to reference the papers automatically to each other just by fulltext search.
    – mishan
    Apr 7 at 14:14
30

Most of the contents are completely original.

How do you know with only two references?

Show your readers and reviewers that your work differs from what has been done before. Show them that you know what has been done.

That's what references are (also) for.

7
  • 19
    @Buffy It's a rhetorical question asking OP to demonstrate that they are at the cutting edge of their research problem and not reinventing the wheel. It's very unlikely, however, that there are only two publications relevant to the problem.
    – henning
    Apr 4 at 13:57
  • 11
    @Buffy in other words, how can you argue your work is original, if not relative to some other work?
    – henning
    Apr 4 at 14:42
  • 2
    As a rhetorical question, I have no issue. Some work is innovative on its own merits and will be immediately recognized as such by a pro. I haven't seen the paper in question, and don't want to judge it. Leave that for the referees who actually get to read it.
    – Buffy
    Apr 4 at 14:49
  • 14
    @Buffy fair enough. But even if the work is novel, strategically it's better to demonstrate that it is. Not every reviewer is equally diligent.
    – henning
    Apr 4 at 15:24
  • 7
    No reviewer wants to waste their time verifying originality from scratch, and it's unreasonable to ask them to. It's up to authors to make the argument, which reviewers then verify.
    – Reid
    Apr 5 at 22:27
19

While you say that your work is completely original, it is unlikely that nothing you have written relates to previous work. If this really is the case, then the paper is certainly missing vital information.

A cryptography paper is not just a description of an algorithm. There are many important things that need to be discussed:

  • What is the problem that you are trying to solve? Why is it important to other people?
  • What background should the reader be aware of in order to understand your contribution? What theoretical tools will you be using to analyze your cryptosystem?
  • What is the algorithm?
  • Why is it secure? Generally, this requires a security reduction to some well-known computationally-hard problem.
  • How does it perform?
  • How does it compare with and relate to previous work?

Almost all (if not all) of these will require reference to other work. If you don't analyze your cryptosystem in terms that the cryptographic community are familiar with, it will be difficult to convince them that your algorithm is secure. Even if you manage to do so, without comparison to the competition you won't be able to convince anyone that it is better than the current state of the art.

To give a more concrete example, you should analyze your cryptosystem to see whether it provides a certain kind of security under some assumptions. What kind of security, and what assumptions? You will find well-understood options for these in all good cryptographic textbooks, which you can then use in your subsequent analysis. If you don't provide a convincing analysis in terms of well-understood definitions, then even if your paper is correct, then it will be difficult or impossible to validate this.

For your particular situation, though, I think it is important to set reasonable expectations if you submit your paper now. The security community has spent decades trying to convince people not to use cryptography in ways that have not been blessed by the mainstream cryptographic community. If you fail to build on their body of work, which your current lack of references suggests, then you will not get anywhere, for the simple reason that even modern cryptographic systems (let's say, those from the last 40 to 50 years which were clearly designed in the computer era) have failed in so many different ways, that it is simply not believable that you have built something secure without taking into account all the lessons that have been learnt during this time.

2
  • 1
    Excellent list! I would like to add one point. After "Why problem are you trying to solve?", I would ask "Why doesn't existing (major) public key systems solve this problem?" Apr 6 at 9:46
  • @StigHemmer: I wouldn't try to answer that; most likely, they do - however, alternative methods for solving the problems are always of interest. What I would do is try to add references for literature by people who have tried to solve the hard problem (given how much mathematicians have done in the past, I find it unlikely that literally no one has considered it before)
    – poncho
    Apr 6 at 15:00
12

If nothing else, if your paper gives the impression (by skimpy biblio) that you are not acquainted with the mountain of work on public-key crypto in the last 40-50 years, people will likely not even want to look at it. There are toooo many cranks who think they have a terrific idea, but are ignorant of the actual literature.

So, in this year, it would be very, very wise to demonstrate your knowledge of prior work, and state-of-the-art, by a substantial bibliography, and to place your work in that context.

I'd have to think that, if you believe that your work is somehow independent of that body of work, you are almost surely mistaken... and misguided... and will have people not take you seriously.

4

If the work is truly yours, and if you understand the state of the art and how we got to the present in the field and the work doesn't, in fact, plagiarize, then you should submit it. Reviewers will give you feedback on it.

But, a warning. Plagiarism won't necessarily be caught by an AI system, since it is about ideas, rather than their explicit expression. So, a paper could be completely plagiarized if it presents the ideas of others as if it were those of the authors, while using none of the phrasing and few of the words of the work plagiarized.

If you know the state of the art and who is the originator of the various ideas then it is easy to know whether what you write is original. And it is possible to unknowingly express the ideas of others as your own if you don't understand what came before. But that is just sloppy scholarship. And reviewers will catch it quickly.

But, assuming the best case, there is no reason not to submit your paper. The reviewers will give any necessary advice for improvement.


While you need to have read, and understood, the literature relevant to your work, not every paper needs to recapitulate the history. If you are writing a thesis, then you need to include much of that, but for a professional audience it isn't necessary. If you are writing for novices, then you need to say more, of course, but people in the field are, themselves, knowledgeable already about the history and the current state.

1
  • 1
    Most reviewers will be suspicious of papers with only 2 references exactly because of this reason - it is beyond the ability of individual reviewer to know EVERYTHING relevant to the paper. It is author's job to prove that their work is novel, no the reviewer's work to disprove this. Ergo, that's why paper with 2 references will be most likely rejected outright by any serious journal. Serving as a reviewer it is not my job to educate the authors, it is their supervisor's job.
    – xmp125a
    Apr 8 at 20:14

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.