A Norm for Self-Citation (guest post by Colin Klein)
“How to self-cite without giving away your identity? I’ve seen two ways of doing it over the years. One is great, and one is really frustrating. We should all stop doing the frustrating one.”
The following is a guest post* by Colin Klein, associate professor of philosophy at Australian National University.
[Jan van Eyck, “Portrait of Arnolfini and his Wife” (detail)]
A Norm for Self-Citation
by Colin Klein
Philosophers like to cite themselves. Reviewing standards in philosophy are extremely fussy about preserving anonymity of authors. This sets up a conflict: how to self-cite without giving away your identity? I’ve seen two ways of doing it over the years. One is great, and one is really frustrating. We should all stop doing the frustrating one.
Suppose my draft says “As I argue in Klein (2015), pains are imperatives”. To anonymize this, I could do two things:
1) “As Klein argues in his (2015), pains are imperatives”
2) “As I argue in my [Author paper], pains are imperatives”
The job of a reviewer is to determine whether a paper is fit for publication. Part of that job is determining whether the citations adequately support claims that are made. Option (1) lets me check the citations. That’s the good option. Option (2) doesn’t. Option (2) thus fails to give the minimal information a reviewer needs to do their job. That’s the bad option.
This is not hypothetical. I’ve seen some amazing things asserted using the mechanism of (2). Load-bearing claims. Preposterous claims. Without taking the author’s word on it (hint: I don’t), such papers are dead in the water.
You can do (1) poorly, of course. If I wrote “As Klein argues in his brilliant, under-appreciated 2015 masterpiece…” then you might have a clue to who I am. But you shouldn’t write like that anyway. If you’re writing in a small subfield, you might worry that that (1) gives clues to my identity. (How many people cite my 2015 uncritically? I suspect just me.) But neither (1) or (2) will help there: that’s a deeper problem about trying to preserve anonymity in small fields.
Indeed, that leads to a related problem with (2). If you know the topic, you’re going to wonder “Gee, this person writes about imperatives and pain, they should at least slag on Klein as part of due diligence. Then you go to the bibliography and there’s no Klein. You might think that the author has overlooked an important part of the literature. Or you might readily—even inadvertently—infer that you’re dealing with a Klein manuscript. So (2) actually makes it easy to inadvertently reveal your identity. Again, I speak from experience here.
So here’s a proposal for a norm:
i. Are you self-citing because the thing you’re citing actually adds to the philosophical discourse? If so, treat it like any other citation, and remove information from the body of the draft’s text that suggests you are citing yourself.
ii. Are you self-citing to establish that you are one of the authors who believes the thing, thereby staking out philosophical turf? You can probably get by with just putting in a third-person citation. But if you worry, then don’t put a citation at all. Add it in the final stages if it gets accepted. The reason we care about anonymity in reviewing is that we ought to be able to evaluate the quality of a paper without knowing who wrote it and whose philosophical career will be advanced by its publication. So staking your claim can come after review.
iii. Are you worried because you’re responding to somebody who is attacking you and it’s really hard to write in the third person about yourself without giving away clues to your identity? Honestly, I think editors need to step up here and admit that this happens and there’s no sensible way in which these kinds of papers can preserve anonymity. But in any case, there’s nothing you can do, so you might as well do (1), because (2) is going to make your paper unreadable.
iv. Are you self-citing for some other reason? Don’t. There are no other good reasons.
Discussion welcome.
There is at least one additional good reason: you are self-citing to show that you aren’t trying to claim that an idea is original to the paper, but rather originates elsewhere (independently of whether it ‘adds to the philosophical discourse’). That is, you are self-citing to avoid self-plagiarism. Maybe that’s covered by the numbers listed, but I think it’s distinct. In any case, it’s my main motivation.
I also want to push back on this point. I have a manuscript I’m working on that elaborates a concept I coined in a previous publication. There are some other publications that I have found since that discuss that neologism, and it would be remiss to not self-cite. Further it’s hard to cite anonymously, as it would be a bit absurd to assume that I have gone so in-depth about some third party’s previously half-sketched theory that I then take up and extend into a larger, more dedicated article.
How about a norm regarding self-promotion in the philosophical blogosphere? (“(Klein 2015)…(Klein 2015)…(Klein 2015)…”)
Would there be a philosophical blogosphere left? I have at least tried to ensure that my post also lacks a proper bibliography.
I might agree with Klein on where to come down on this issue, but his post seems to ignore a significant problem with option (1): it may make it seem that you are just piggybacking on someone else’s project in some small way, rather than extending your own project.
Perhaps that shouldn’t make a difference to whether the manuscript under review is publishable, but it often does make a difference to whether the manuscript is perceived as worthy of being published.
Of course, the downside to going with option (2) in this case is that if the referee is in the know enough to be aware of and interested in the project you are building on, then you have compromised anonymous review.
But it seems a reason to both argue for (1) and to at the same time plead with referees not to reject papers simply because they are “just” a (perhaps interesting) extension of an extant project.
I’m not sure this is all that satisfactory, however, and it seems to make the choice between (1) and (2) more fraught than suggested here.
Perhaps that shouldn’t make a difference to whether the manuscript under review is publishable, but it often does make a difference to whether the manuscript is perceived as worthy of being published.
I cannot see what basis there might be for saying that, other things being equal, a paper that extends your own work has more merit (as far as publication goes) than a paper that extends someone else’s work. However, I take your point that referees might nonetheless unreasonably assess these cases differently. However, this problem can be directly addressed by the journals. All journals should explicitly prohibit self-citations in the style of (2) for the various reasons that have been stated. Once they do this they can also include in the instructions they give to their referees a line about being interested in publishing work that extends other published work and seeing no difference between an author extending their own work or someone else’s work.
I second what Alexander suggests, but I would like to add something important. The worry raised by Alexander is not so pressing when you are a famous professor, but *way more pressing* when you are an *early-career philosopher*. There are two main reasons why an early-career philosopher may prefer (2) to (1), and be right to do so:
– *PERCEIVED VALUE*. This connects with what Alexander says. But note the difference when one is early career. If, as a referee, you see an established philosopher cited in the text (say, Williamson), you’re gonna say: fair enough, you wanna build on Williamson as everyone else, that sounds right, plausible enough. But now suppose that the author is an unknown adjunct from some minor Eastern-European university, i.e. me, Philosoloser. The referee’s gonna say: wait, you wanna build on Philosoloser? Why do you wanna do that? Nobody does so. There are more prominent voices in the field that you could refer to instead. All of a sudden, you need to justify your reliance on Philosoloser. From experience, this is guaranteed to result in a *really* hard pushback from referees: obviously they don’t object to *citing* your minor work, but rather object on *expanding* on it, and they want you to *prove* that what Philosoloser did in his 2018 is actually solid (as if this wasn’t exactly what Philosoloser does in his 2018). The idea seems to be that the lesser known the work of a philosopher is, the more you need to do to justify your reliance on it. In short: there will be a *strong* prestige bias against you building or expanding on the work of Philosoloser.
– *ACTUAL ANONYMITY*: Williamson gets about 1000 citations a year: that’s a thousand papers that come out citing him every year. Suppose Williamson submits a paper using (1). The paper he submitted may just be one of these 1000 papers that cite him. He has no risk of undermining anonymity by using (1). But every year, only about 5 papers come out that cite Philosoloser; of these 5, at least 1 or 2 are from Philosoloser himself. Now, unless the referee is naive or downright incompetent, the referee will realise that there is at least a 50/50 chance that the mysterious author hoping to expand on Philosoloser is Philosoloser himself. This undermines anonymity; if we are honest, in most cases, it will make anonymity simply impossible. And again, this will favour the established professor over the early-career one.
In short: I believe that there is a strong argument from prestige bias against using (1) rather than (2). I don’t think it’s a conclusive argument, but I think that matters are way more complicated than the OP presents them to be. We don’t live in a rosey world where everyone has the same chances of getting published. We live in a world in which prestige is a big factor in determining which papers get accepted (surely also with double blind review, and somehow also with triple). It is at most naive to think that (1) is better than (2) without considering that there is a considerable risk that this further reinforces the privilege of established professors from prestigious universities.
These are very nice additional points by Philosoloser.
Let me just add to them that many of us already dislike the journal culture in which it is hard to do or even lay the foundations for a big interesting project in one paper (because one must carefully defend every incredibly small step away from the well trodden path).
If we go with (1) as the norm–and bear in mind these points by Philosoloser about prestige and citation practices and what projects are perceived to have merit–that will mean that only big name fancy people (the kind whose in-text citation and project-furthering merits no eyebrow raising) can take on such big interesting projects over a series of anonymously reviewed published papers.
I’m not sure changing the culture around this will be accomplished just by giving referees instructions that include “a line about being interested in publishing work that extends other published work and seeing no difference between an author extending their own work or someone else’s work.”
It’s one thing to extend Williamson’s work–whether you are Williamson or someone else. It’s another thing to extend Unknown Philosopher’s work, or Currently Unemployed Sometimes Adjuncting’s work, or Still A Grad Student’s work. This will often be seen as just less significant or important. (And this needn’t just be prestige bias; it can be hard to know how important some project is when it is just getting off the ground, and when the person who is most associated with it is completely unknown to the referee.) But it might also either (a) seem inappropriately terrain encroaching if the extension is too intertwined with Unknown Philosopher’s project; or (b) simply compromise anonymity, resulting in the referee simply assuming that this paper is also by Unknown Philosopher.
Again, I’m not sure these considerations are decisive. But I don’t think it’s as easy a case for (1) as Klein suggests, and these strike me as important to note costs of going with (1).
I also agree with Kenny that if somebody cites 4 or 5 of their articles in the (2) way, that is often a red flag of its own. Lots of self-citation almost always looks bad, and is usually unnecessary.
Philosoloser: I take your point, and I think this is the sort of thing that ought to prompt broader reflection as a discipline about why we value anonymity in reviewing in the first place and why we think papers are worth publishing. (Until then we should do (1); it’s more in line with the *stated* aims of reviewing.)
As it stands, the stated reasons for anonymous reviewing are supposed to be that we can evaluate a paper without knowing who wrote it. As you rightly point out, most people don’t think that the philosophical goodness of a paper is fully intrinsic to it. A paper also has to engage with and advance present debates, and the degree to which *that* is interesting sorta depends on who is involved in the debate. The third option above for self-citing is kind of the limit case.
I’ve been mulling on this because I do a lot of co-authoring with scientists these days, and their journals quite often don’t have anonymous review, and everyone thinks that’s sort of how it should be. A replication of a study carries a lot more weight if it’s made by a distinct lab than the original one. Surprising results from a lab with a track record for high-quality work are more interesting than if they come from an unknown one. That creates an obvious prestige bias (and authors grumble about that all the time), but it also means that trying to establish the merits of the paper — both intrinsic and extrinsic — is more transparent.
These are good points Philosoloser, but it’s not clear to me that *PERCEIVED VALUE* gives an early-career author much reason to prefer (2) to (1). The referee you describe thinks that the only work worth extending is that of significant scholars whose work has already been recognized as important. When a “non-significant” scholar uses style (1) such a referee quickly identifies the cited work as non-significant. However, using style (2) seem to often lead to the same outcome. The referee presumably has reasonable knowledge of the paper’s topic and the surrounding literature. When she sees “This paper extends the work of [reference withheld]” she knows that if that reference is the work of a significant scholar she will probably pretty quickly guess whose work it is because she will already be familiar with that work. If, by the end of the paper, she still has no idea what work was being referred to then that is very strong evidence that it is not significant work. Thus she may be just as inclined to reject the paper on these grounds when style (2) is used.
Indeed, sometimes (1) does better than (2) with this problem. If the work the early-career person is extending was published in a decent journal then the referee, when seeing the full reference (but not knowing that it is the authors work) might think “this scholar is not very significant, but this work was published in a decent journal so maybe extending it has some value”. By contrast, if the reference is withheld and referee concludes that it is not significant work by a significant scholar then, without a citation that at least ties it to a decent journal, she may be more inclined to dismiss it. Indeed, when she has strong evidence that the reference is to a non-significant scholar, but doesn’t know the journal, her default assumption may be that its probably a poor journal.
Now perhaps you point was supposed to be that (2), but not (1), allows the referee to identify the citation as a self-citation, leading her to cut the author some slack. In other words, extending the work of someone else who is a non-significant scholar is judged more harshly than a non-significant scholar extending their own work. But I cannot see why this would be true. A referee who is biased against the work non-significant scholars would seem likely to be more dismissive in the case when they have strong evidence that both the author of the paper they are reviewing and the author of the work that paper is extending, are not significant.
Regarding the concern of *ACTUAL ANONYMITY*, I agree that this is a problem but I think that the threat of having the referee identify who you are is just as bad or worse when style (2) is used. Furthermore, an author who uses style (1) and is worried that a referee will guess her identity because she is citing her own rarely cited work can take steps to prevent such guesses. For example, as an aside, she can express in her article disagreement with one or more of the minor points she makes in her previous published work that she is citing (e.g., “I will be using the model of Philosoloser (2019) which I think is helpful here although, even though I disagree with some of the minor details of Philosoloser account.”). A referee reading such a line will typically assume that it is not a self-citation.
There is another unethical reason that I think motivates some people to do (2). If you do (1) then we don’t have any hints about your status in the relevant sub-discipline or the profession more generally. You might cite five of your papers but we have no idea that they are yours. Indeed, we may have no idea whether you are junior or senior, established or unknown. However, if you do (2) and make many self-citations, then you make make yourself look like an important scholar who has already proven themselves to have valuable things to say on this topic. Even if your identity remains anonymous, this impression about your status may bias reviewers, leading them to read your work more charitably and be hesitant to dismiss your argument with the first half-formed objection that pops into their head. But this is not fair as those who do (1), or don’t have any relevant publications to cite, should not have tougher standards applied to them.
For this reason, and the ones that Colin cites, Journals should actually explicitly prohibit self-citations in the style of (2) and send submissions back to the author that fail to self-cite in the correct way.
I find that when I see five instances of (2) while I’m refereeing, I get that this person sounds like *they think* they are an important scholar. But I see it at least as often in crank papers that should have been desk rejected as I do in papers where the author really is important, so it actually comes across negatively.
This is also my experience, Kenny, and one of the things that kicked off the rant in the first place.
Yes, there is a significant risk for the person who uses this technique to signal their “status” that it will backfire. However, shrewd operators know this and are much more subtle in how they do the signaling. They are careful not to excessively self-cite and not to overhype the importance of the self-citations they include. Nonetheless, they strategically include self-citations at the right place and describe them in just the right words to suggest they are someone of status. Indeed, these techniques can get very devilish. A shifty scholar might have a publication that says something similar to a much better known paper by a significant scholar. They might then self-cite their paper but describe it in a way that might lead referees to wrongly guess that they are citing the better known paper by the significant scholar and thus to wrongly identify the author as this scholar.
My general point is that (2) allows the author to explicitly give the referee information about their scholarly activities (I have previously published papers that make points related to this) and there is always the potential for unethical operators to game this information channel to try and swing things in their favor.
JTD, I also find this — a one-off instance of (2) might pass by without harm, but the really annoying versions of this are the ones where people are clearly trying to give the impression that they are Big Fancy Scholars and if only you knew you’d be shocked at how important the paper you hold in your hands is… etc etc. But I have the same experience as Kenny, in re this backfiring.
I have had at least one journal ask me to resubmit after changing all my self citations from style one (which is my standard practice) to style two (which I also find a subpar citation style). I’m posting this because I can imagine a referee who is frustrated with a submission using style one perhaps giving a reject verdict if they were already on the fence about the paper. That referee should not necessarily blame the author, and perhaps should apply something like a principle of charity when it comes to style two citations .
Christopher: fair (and sorry to hear it). This should also be read as directed as much at journals and journal editors and reviewers as authors.
Same here. I submitted my paper in style 1, then got a note a couple weeks later demanding that I change self-citations to style 2, because any occurrence of my name is forbidden.
Because the topic is one that, as far as I know, only I have contributed to (apart from one contribution 36 or so years ago), that was basically a recipe for immediate outing. (Except, of course, that probably almost nobody has read my other paper!)
For what it’s worth, the BJPS policy has been in line with (1) for many years now: http://www.thebsps.org/auxhyp/howtoanonymous/
Good to know! BJPS is a class act.
I do a mix of style #1 and, when that’s less appropriate, adding [citation omitted for review] when it’s more clear that I’m building off of my own material and never had any problems from reviewers or editors. If the paper is accepted I add the citation in later. Been doing it for 15 years at this point.
I agree with Colin that (1) should be the default, but preserving anonymity should be one’s main goal and as Caligula’s Goat points out in some cases (1) makes it difficult to preserve anonymity; this is especially true if you’re building on work that is still forthcoming (yes, people read each other’s paper before they’re published but how often people build on another person’s work before it is even out). I was convinced by this after reviewing a paper that mixed the methods (it used (1) for a well-known piece and (2) for a forthcoming piece). Had the author used (1) or (2) exclusively, I would have certainly known who it was. But the mix worked well.
Prior to reading this, I had thought that (1) would be the right way to do it, but it struck me that if I used (1) it might make it look like there was independent confirmation for my ideas, rather than me simply citing myself as confirmation for my ideas, intuitions, and theories. For example, if the review doesn’t know that I am Smith, then ‘Smith (2020) holds X’ might look like other people how the same view as I do when it’s only just me. This seems misleadingly problematic.
My current norm is just to have no citation at all until acceptance. That can make the paper look like it has assumptions that need to be defended, but it is usually the best way to preserve anonymity. Anonymity is crucial. Reputation is a very serious source of bias.
More an interesting anecdote than an argument for one or another. I co-wrote a paper a while back where one of the points was to correct a rather important technical (mathematical) mistake that both my co-author and I had made in earlier, distinct solo works. We explained the mistake and then attributed the mistake to the relevant papers via citation style (1). One of the referees gave a report (otherwise positive) demanding that we add an entire section going through the arguments of the earlier papers in question in detail, since otherwise no one would believe that the authors we were accusing would make such a blatant mistake, given their expertise in the area. Fortunately, the editor was sensible and suggested we could ignore that particular suggestion 😉
For what it’s worth, I’ve had a reviewer ask me to self-cite. (At the conclusion of the paper, I mention applications of the central idea to other philosophical puzzles – the reviewer said that the paragraph should be eliminated unless there are other papers I’ve written which I can cite. I have written other papers which I could cite, but I thought that any mention of the explicit papers at that point would destroy anonymity. So I cited them as ‘citation omitted for blind review’.)