It's amazing that none of the commenters so far seem to acknowledge that competition for prestige is one of the forces that fuels research. Take that away people will be less aggressive about pursuing and investing their time in valuable areas of inquiry.
There are downsides to competition but there are considerable upsides also. The bet we make in our culture is that the upsides outweigh the downsides, and based on history it seems to be true.
Based on history we can observe that we do make scientific progress, but you don't have the data to claim that it is true that the upsides outweigh the downsides. The downsides could very well outweigh the upsides but still allow for progress. What's meaningful is to compare vs an actual prestigeless system and see which is more effective -- does the search for prestige outweigh the corruption?
If the publication includes source code, must the copyright information be scrubbed?
If the publication concerns a software project, can the publication link to where to get the source code, on a site with the authors' names? Or is the program name and/or author names also be anonymized?
there's been work on writing stylometry, namely detecting authors by style of writing. Now academic papers might be a combination of enough authors that the results are not easily classifiable, but still, just stripping author names is not enough.
You can also tell quite a bit based upon the citations. I've been able to guess at who was the original author based on how the paper was written/structured and the citations.
You can also guess who writes reviews. There was this one guy who got pissed off at me at a conference for nonchalantly discussing who I thought the reviewers on our paper were... whatever man, it's my review now.
It would probably be infeasible to judge every single paper purely on its merits. Reputation may be an imperfect proxy for merit, but it's certainly not worthless.
In any event, this argument is there and needs to be taken seriously.
Prestige from whom though? Many of these researchers are chasing money first, knowledge second (as they should). And, they know that the big money will pay them orders of magnitudes more if they're 'Prestigious'.
It's the same reason why my RSS feed is more likely filled with junk articles today than it was 5 years ago. It's a game of quantity as much as quality.
> You don't go into research if money is your motivating factor - there are plenty of easier ways to earn a living.
Seconded. Just made the switch to industry a couple of months ago and I'm making twice as much for half the hours. Academia is looking more and more like slave labour to me these days.
As the author notes, there is no way to make conference presentations anonymous. But the problem is much bigger than that. Even most informal conversations between scientists involve promoting and discussing one's own ideas and work.
How could hiring or awards work? It is simply not possible to make such decisions on meritocratic basis without individual credit for ideas. The only consistent way I could see that working is to make such applications anonymous, but at a certain point anonymity will inevitably break, if only for in-person interviews or once people are actually hired.
More broadly, complete anonymity is impossible, and incomplete attempts towards anonymity would only reinforce existing old-boys networks within science. Authorship would become a matter of gossip and innuendo, which would make it even harder for new-comers to break into science.
You may be right. This may be a hilariously bad idea.
But if you are right then it is a shame that you are right. What is the goal of science? Shall we say truth? But what are the goals of scientists? You can say that it is truth, but perhaps it is fame/prestige/success, call it what you will. And what about ever-present human bias? And what about discrimination based on ethnicity or gender. I'm no bleeding heart, I assure you, but these things are real.
Humans have spent most of our evolutionary lives not being scientists. The suggestion to render part of the scientific process anonymous to mitigate the our human failings, though idealistic, is an _admirable_ idea. That this admirable idea may also be a hilariously bad idea is sad really.
You know what? It doesn't have to be all or nothing. Another commenter here wonders what it would be like if we could flip a switch and make the process of scientific discover both anonymous and wiki-based. But why should it be all or nothing? How about a scientific enclave (or more than one!) that adhered to these principles? Well it turns out that at least in mathematics there are precedents! Bourbaki[0] was the pseudonym of a collective if I'm not mistaken and nCatLab[1] is a wiki-lab for collaborative revisable work in math, physics, and philosophy.
I don't entirely agree with your arguments that anonymity would be a good thing. Or even that it's sad that we can't have it. Although I do agree that it's helpful in some situations.
You say anonymity would prevent people pursuing science purely for fame. However first of all, science is a hilariously difficult area to become famous in. And secondly, wanting to be recognised by your peers is not in and of itself bad, if it causes someone to pursue science then it might even be considered admirable.
I think discrimination against authors is a lot more worrying, but the situation is better than it once was since it is now no longer possible to prevent someone from publishing their work. If nobody would read it if they knew who the author was, or if it's conclusion is so outrageous that it might damage the author's reputation, it's also possible to use a pseudonym.
I vehemently disagree that it's a sad reality that we can't have full anonymity. A scientific process that is completely anonymous is either unnecessary or if it's necessary then that would be very very sad.
"science is a hilariously difficult area to become famous in"
There's fame, like Einstein had, and there's fame like how everyone in your specific field knows who you are. For a comparison, there's "president of the United States" but there's also "Mayor of Albuquerque".
The former is very hard to get. The latter, more common, and more achievable.
Here's an example of how anonymity would hurt. A lot of pharmaceutical research is done in industry. While some of the research is kept proprietary, especially early on, other parts are published. It's a lot of work to publish. It may include talking with company lawyers so that confidential information isn't revealed. Why might someone go through this extra effort?
One factor, of course, is the altruistic goal of advancing science. Another factor is the more egocentric desire to finally be able to tell others - who are often friends - the cool things you've been working on. A third is the knowledge that layoffs come, and a paper is a form of advertisement for future employment.
The latter two disappear if all publications were anonymous.
Scientists compete for scarce resources: funding and great students. How do we allocate money and students to the people who will use them most effectively if we can't measure their effectiveness?
It's not just prestige. Some scientists really are better and more productive than others. They should get the resources first.
The point of science isn't so much to product truth but to determine which truth we actually care about. So instead "science" is really a community of scientists, and in order for this to exist you have to have all the gatekeepers, in-groups and out-group, canonization, etc. Bourbaki wasn't just a group of random anonymous internet users, it was an inner circle of mathematicians who considered themselves at the center of the discipline, which is completely opposite.
So you are proposing essentially communism meets scientific journals... Great on paper but terrible in practice.
Communism created Zero incentive for being innovative.
You're proposition creates Zero incentive for being innovative.
It sounds like you are on some high horse superiority complex with your noble ideas.
The fact is that the researchers that are receiving disproportionately large funding once paid their dues and received little funding.
It's Capitalism 101. Plenty of companies have better products but lose because they messed up in other areas.
Plenty of researchers have better research but lose because they fail to educate others, fail to take it to the next level, fail to produce anything that has meaning for society.
With anonymous publications special interest groups can drop an unlimited number of publications without exposing their network of shills. So while the proposal might help with some things (which I doubt), it makes others worse.
If the 'shills' are wrong, then the peer review should take care of them.
The whole problem that this proposal is trying to eliminate is a combined problem of discrimination (for various reasons) and cargo-cult/groupthink; calling someone a 'shill' is an example of at least the first of these issues.
If the peer review process cannot successfully refute claims in a paper, and has to resort to name-calling, that is a sign that the reviewers are either incompetent or lazy (, and that there may be fundamental flaws in the peer review system).
> and that there may be fundamental flaws in the peer review system
There are.
One problem is that there are too many papers to be reviewed, another is that the papers generally do not include complete datasets. Also, where large volumes of data are involved it is common to perform computational transformations on that data to extract something; the source code and build process are not always described in such a way as to make the executable reproducible.
And those are just some of the problems with peer review, others include more human problems such as insufficient remuneration for the reviewer, reviewers who have too much at stake to give an unbiased review (whether too positive or too negative), too few reviewers with the necessary expertise, etc.
Even with competent and committed reviewers the system is flawed; unfortunately there doesn't seem to be any alternative.
I agree with your description of other problems with the current system, and think that anonymizing authorship will serve to force them to be solved. Many popular journals currently look at the author's position and previous works to filter out the unpopular and/or unprestigious authors; anonymization would eliminate the bias against young, innovative, and/or unconventional scientists, and force the journals to improve their process.
An anonymous science conclave could require that the data is complete & calculations/code presented unambiguously... unless that can only really be detected upon analysis during a serious peer review, rather than some surface level filtering.
I think you meant s/peer review/replication study/g. Peer review aims to check that papers are publishable, not refute or affirm them by replicating their results; that's a separate process.
I agree shills will eventually get sorted out though. The truth always wins in the end. Just not as fast as just peer review.
Replication should be done more often, and anonymization may serve to catalyze such an increase. Peer review is supposed to examine a paper for flaws, and do so fairly for all authors. This determination of which papers are well thought-out and documented is then supposed to decide which are publishable.
I think you are also tacitly describing a pessimistic view (which I share) of the reality of peer review, that it does not succeed in these lofty aims, and that publishability is influenced by a number of extraneous factors.
> If the 'shills' are wrong, then the peer review should take care of them.
That's not what peer review does. Peer review isn't investigative journalism, nor is it a criminal trial, nor is it designed to figure out if the results of the study are correct or incorrect.
Peer review is used as a way of determining whether the methodologies or underlying logic used in a study are flawed (in a provable way), and what the severity of the flaw(s) is/are.
>"Peer review isn't... designed to figure out if the results of the study are correct or incorrect."
There is no way of verifying the correctness of most studies and papers; they are simply examined for flaws. I do not see any reason why papers by 'good' authors should be examined less critically than those by 'shills'.
That's the point. If you publish anonymously then you can publish methodologies and analysis that is up to "gold standard" while completely falsifying/manipulating data.
The current system mostly relies on reputation and professional ethics/pride to filter out that kind of behavior even before the publication stage. The cost to replicate experiments is typically high - even in our current system we cannot replicate to the degree that we want. One presumes that in an anonymous system, the problem would get even worse.
It's a question of resources. Once a shill has been identified editors can reject them at the editorial stage without even sending it to the peer reviewers.
If funding sources are also not disclosed in the paper, then the peer review process will be unable to determine bias.
You seem to be describing a 'fruit of the poisonous tree' view of funding, and saying that studies funded by certain authors are right to be discounted.[1] For one thing, anonymous publication could serve to discourage partisan funding of studies, as the donor would no longer be able to verify the authorship of any given paper.
The use of 'credibility' as a criteria for publication also creates large biases against novice scientists, or scientists from unusual backgrounds (scientific and otherwise). In addition, 'credible' scientists should never be given any benefit of the doubt; if someone depends on this, they are not doing good work.
In our previous work, Anonymous et al. (2002, Nature), Anonymous et al. (2003, Science), Anonymous et al. (2009, Cell), we established the validity of our approach and we outline the details of our methodology necessary to repeat these kinds of experiments. Here we use this approach to address a novel question, ...
So use numerical identifiers, hashes, or PGP keys, in conjunction with a moratorium on outing yourself as the author associated with a certain key before a given length of time has passed, during which the work is vetted.
For example, if 2000 scientists (that may or may not work together) use "Anonymous et al." name then how anything inside an article will make someone choose a particular scientist's papers from a list of papers in a journal?
Obviously, if you're reading an article; you may recognize the work (science is very specialized: if you understand the article; you'll probably recognize the author (the group) even if all references/names are removed anyway -- but you need to read the article).
The suggestion doesn't remove all bias but it can remove some. There is no silver bullet that would make people rational.
The authors in their paper propose a different approach to citing articles, which is to cite the article name instead of the author. This criticism, while humorous, is off-base from what's actually being proposed.
This is not an example that the authors gave, but you might write something like "On the Origin of Species (1859)" instead.
This joke also misses the point in that a follow-up anonymous article would not, to match the author's intent, reference the authorship of previous work. The whole point is for each article to stand on its own, so referencing yourself pseudonymously as the author of a previous article does not achieve the goal.
I don't want to be "that guy" who takes the humor too seriously, but I worry from the comments so far that people will conflate the humor and the actual proposal. The overall aim and proposal of the article is to find a way for papers to be judged individually on their own merits, rather than on the merits of their author, the author's institution, the author's nationality, or the paper it was published in. This is a worthwhile goal even if we cannot see a serious means by which to achieve it.
To be fair to the jokes, while reading the article I was unsure if it was satire. I'm still not completely sure. I believe the authors are sincere in their belief in anonymity as promoting merit, but I remain unsure whether they think their proposal is realistic, and whether or not they are making some kind of meta-point about the state of scientific review by observing how far away the current review process is from one that would be (as best as we can make it) unbiased. That is to say, upon reading the paper, the idea of implementing their anonymity proposal seems ridiculous - and this realization may be partly the point of the paper, to get the reader to realize how many different points in the editorial process depend on the author's identity rather than the material's merit. It does not strike me as satire, because the work takes itself seriously, but it may seem like satire because of how large the gap is from where we are to a meritorious review system.
"Ha! It would be ridiculous for work to be reviewed, published, and promoted purely on its own merit, rather than based on the legacy of the author."
On the other hand, it's hard to say how well a fully anonymous system would work. Human biases often serve a useful cognitive function in allowing some filtering operation to be performed more efficiently, albeit with less accuracy. A random lunatic who claims to have invented a perpetual motion machine is likely to be wrong. It's possible that devoting equal review time to all claims and all papers would lower the sum value of scientific output. If the volume of scientific material is too great to comprehensively review, from any given observer's standpoint, then a system that relies on identity as a heuristic and is slightly less meritorious may be strictly better than an anonymous system that is slightly more meritorious, for example. Society might have a different goal for the output of the scientific process, which is to maximize the discovery/promotion/publication of useful scientific output - and that is a different goal than judging individual work fairly.
The paper isn't suggesting 'Anonymous et al. (2002, Nature)' style links. In fact, it addressed that concern directly:
> It is suggested that, if using parenthetical referencing, we are not referring to the author(s) any more but to a few title-keywords of the article and do not refer to the journals name. Those title-keywords can replace the authors’ names completely. They can be chosen by the author(s). Taking the APA citation style as an example, citing Nosek and Bar-Anan (2012) could look like this: "...how to change scientific communication (e.g. Opening Scientific Communication, 2012)". The full reference should then be placed in the reference section together with other bibliographies starting with the letter O. The latter can be applied to the Vancouver citation system, too.
I happen to think it's not a useful solution, because 1) 'chosen by the author' is a free-text field open to all sorts of ways to hack the system, 2) names are easier to transliterate than titles are to translate, and 3) citations are often not made because of the main topic of the paper but can include also minor points which are later found to be more important. Using the author's name is a more neutral link in that case.
(For example, I've seen a paper cited because there's a one line comment about the software used to make an image in the publication. That software became popular, but there was no publication about the software itself, so people cited that first paper instead.)
In a twist of irony, this paper itself is not anonymous.
Jokes aside, while many suggestions and rationale for them make sense, I wonder how it is going to be implemented in practice. Many scientific areas are narrow enough for people "in the know" to figure out who the author is.
I use authors' names all the time to see what else they published, if I need some more information on the topic.
It may be useful to enforce anonymity for the submission process but once accepted, publish it with names.
Poignant? I half-suspect you're trolling. It's ironic because one might assume that a paper that argues for anonymity to itself be anonymous, but it's not. Irony: when things are the opposite of what's expected.
It is really useful to know the authors because if you come across a researcher who has been looking at the same types of problems as you, then you read all their previous work to gain much better context of the domain. Plus then you look at all the coauthors and look at their work, and you can find some really interesting vein of knowledge that way.
That seems more a criticism of search engines than of anonymous publishing. What you need is a search engine that can find similar work. Being written by the author is just a heuristic in a search for something different, even if probably helpful.
"A sufficiently advanced search engine" should be the new "sufficiently advanced compiler."
I'll give an example of how difficult that is. I am researching the historical development of graph canonicalization for chemical information. I have bits and pieces from published papers, but did those people work together or independently?
The papers include institutions, so I know two of the people (Gluck and Morgan) worked together at DuPont before Morgan went to the Chemical Abstract Service. The latter then cites a presentation Mooers gave at an American Mathematical Society meeting in 1958.
But what was the relationship or key insight from Mooers?
> I was party to a conversation in which Harry Morgan [CAS] and Calvin Mooers discussed the issue of a canonical connection table. Mooers, in the light of the counter-example raised against the work of Dave Gluck, suggested a permutational approach to the design of an algorithm, which would overcome the objections. Harry Morgan (1965) then devised the method that bears his name and that has been a keystone in chemoinformatics since that time.
That reveals a closer connection than seen in the papers. But what was this counter-example?
> The presentation of counterexamples, such as the arbitrarily numbered planar graph in Figure 5 devised by Dr. Lehman of the Walter Reed Army Institute of Research, have necessitated revisions to canonical ordering techniques such as those used by DuPont and the Chemical Abstracts Service.
And there is it - the counter-example. But notice the use of institutional names, not published papers?
How is a search engine supposed to make those connections?
I can totally understand the argument that people should be referring to unique names, instead of a choice of names, to make this sort of thing easier for mechanized search. But that's not how people work. Why are they using institutional names? Because those components, like canonicalization, are part of a much larger system, which is more coupled to the institution than a single person.
> How is a search engine supposed to make those connections?
I can't tell you how exactly, but if current word embeddings can do stuff like "Paris - France + England = London", I don't see why future systems shouldn't be able to resolve the relationships between institutions, people and publications you mention, in a similar fashion. I am not saying that it will be easy, your example requires much more "one-shot learning", but I could in principle see how an automated system might do it.
If we're going to group papers by author, we should probably assign some kind of unique, convenient identifier to such individual groupings, to make it easier for researchers to communicate between each other regarding groupings of interest to them.
What I would like is to be able to explore the full citation graph, both interactively and programmatically. I think it would make identifying the key literature in a new topic much faster.
> For many years, scientists and others have worried about reporting biases such that negative or null results from initiated clinical trials may be less likely to be published than positive results, thus skewing the literature and our understanding of how well interventions work.[2] This worry has been international and written about for over 50 years.[3] One of the proposals to address this potential bias was a comprehensive register of initiated clinical trials that would inform the public which trials had been started.[4] Ethical issues were those that seemed to interest the public most, as trialists (including those with potential commercial gain) benefited from those who enrolled in trials, but were not required to “give back,” telling the public what they had learned.
> Those who were particularly concerned by the double standard were systematic reviewers, those who summarize what is known from clinical trials. If the literature is skewed, then the results of a systematic review are also likely to be skewed, possibly favoring the test intervention when in fact the accumulated data do not show this, if all data were made public.
This is public information, and (quoting https://en.wikipedia.org/wiki/ClinicalTrials.gov ) "A 2013 study analyzing 8907 interventional trials registered in ClinicalTrials.gov found that 23.2% of trials had abstract-linked result articles and 7.3% of trials had registry-linked articles. "
To be truly anonymous would mean hiding this link, which would be a backwards step.
I need to be blunt on this. This is just not gonna happen. As nice as it may sound. It's like wishing for world peace.
Time and energy is better spent choosing specific issues to solve the issues the paper brought up. A universal band-aid of "just get rid of the author names" won't fly.
Talk to any academic. You'll find a few sympathetic ears, but for the most part it's the way it's been for 300+ years.
There are some interesting points here, but I don't think the author adequately addresses research funding transparency concerns. Many funding sources require researchers to include a grant number in any publication. This would effectively deanonymize at least the Principal Investigator and their work related to the grant, since that is public information.
On top of that, many fields are small communities, and everyone knows what everyone else is working on and which methods they use.
Anonymous publication is in all probability impractical, but the concept also has serious flaws. Identifying authorship is important not only to give credit (or blame) where it's due, but also for historical reasons, to know how research developed among particular threads in a given disciplines.
Naming authors is necessary to verify results, e.g., when authors admit to faking results (not a rarity), other articles by those authors are suspect as they should be. Publishing anonymously makes it hard or even impossible to know the legitimacy of reports especially as time since publication gets longer.
While anonymous publication is highly unlikely to happen, IMO there are a few things that would materially improve the process. I think peer review and perhaps editing itself should be done anonymously. That way much of the "politics" is eliminated because anonymous review would have to accept/reject a manuscript only on its merit. In line with principles of full disclosure, an accepted paper would be published under its authors' names.
It might also help to reduce political pressures associated with publishing. One change to consider is encouraging reports of research findings or results, but discouraging premature "conclusions", i.e., when effectively authors' speculations. In many fields data is lacking or inadequate to support conclusions, it's hard to say what data mean until there's enough such that clear patterns emerge. Constraining the urge to be the first to reach a conclusion could reduce unnecessary competitiveness in research communities.
Contributing to the community is important in science, but contribution has a value that needs to be recognized or rewarded if we are to see it continue. Publishing a scientific report is an accomplishment to be proud of, certainly not selfish to enjoy being known for having done something worthy.
Even anonymous papers aren't fully anonymous. Not only are any mentions of previous work a dead giveaway, but knowing the topic lets you narrow down to the research groups are currently working on the issue. Not to mention writing style. If your field is small enough you might be able to identify the writer just by his mannerisms.
If science could get those to work, the idea would pay for itself...
More seriously - science isn't just about papers. There are also conferences, conference proceedings, books, institutes, funding programs, research labs, industry collaborations, and awards.
How do you find good researchers to hire and get them through the tenure process?
The trouble here is that opportunistic anonymity is not enough. You need guaranteed anonymity, that even if someone tries to leak information about the connection between a researcher and their work (intentionally or unintentionally), it won't work. Otherwise social pressure will be in favor of those leaks: e.g. faculty who do hiring will take their knowledge of what researchers are really working on in account when they look at researchers to collaborate with.
And I don't see a tractable way to achieve guaranteed anonymity.
My academic wife's impression of the idea is that while eliminating bad heuristics can be good, there are also good heuristics that come with name recognition.
In particular, it's often difficult in social science to know if the research methods were really done correctly. More prestigious authors tend to be more rigorous than unknown authors.
So the point of eliminating as many bad heuristics as possible (regardless of how feasible it actually is) might throw out a size-able chunk of the baby with the bathwater.
Double blind reviewing is mentioned, but is glossed over as "not ready yet". This is what we should be focused on improving. The big problem seems to be that editors have more oversight than the reviewers and can see who the authors are.
Most of the top journals and conferences already enforce double blind reviewing, so in principle nobody knows who anyone is. The submission requirements ask that you do not reference yourself directly in your work so that the reviewers can't infer anything.
On the other hand it's often obvious who a submission is from if you know the people in the field, so that goes out of the window. This also doesn't stop political rejection, where the reviewer has an interest in the research and claims "Oh this has already been done".
I also have no doubt that certain big names in academia would kick up a huge fuss if their work was rejected.
> However, the speaker should choose a sufficiently different title from her publication.
What's the point of that? So I give a talk, I have to come up with a different name for my work and then when I tell people "You can read about it in my paper here" I've given the game away. Or do I have to make everyone play hide and seek trying to guess which paper is mine?
This (anonymity) is impossible in my opinion. Research projects can be so focused/narrow that anyone who is following the conference proceedings/papers could narrow down at least some of contributing authors in some papers fairly easily.
Biases arise if people are allowed to cite other papers even when the publications are made anonymous. The importance of a paper can be estimated from their citations. For example, an important work cites another ground-breaking work. Besides, novel research idea and topic is rare and serves as a signature to the authors and their affiliations. Because researchers often verbally share their research results with others (through the internet), every people will recognize them and their works finally.
My feeling is that as long as the whole open system makes scientific progress, and suppresses malicious or false scientific results, it is fine.
So let me get this straight. In order for attribution (credit to researchers) to work correctly, we should require them to publish anonymously?
I don't think there is a free lunch here. Either you are using publication record to determine who should get paid or tenure, and then you will have problems with bias, or you are not doing that, and then the authorship of the papers doesn't really matter too much.
Otherwise, we're putting a lot of trust in journals as the gatekeepers of the true identities of their content creators. That seems like a fine opportunity for corruption.
I agree, and I also think science should be done on something closer to a wiki than a PDF.
Unfortunately, there is no switch to flip that can make this happen. Making the change would require a cultural shift across a majority of scientific disciplines, and the cultural leaders are the ones who are being accused of benefiting from the status quo.
Of course, but this is people's lives and careers. If you've spent 60 hours a week for the last 40 years on something you aren't about to hand it over to some randomer on a wiki.
That is precisely the cultural shift I am talking about. If you are doing science, the thing you spend 60 hours a week doing should be "trying to understand the world better." Once you have achieved some new understanding, you should want to share that understanding with others.
What you're talking about are things like pride and profit, stubbornness and renown. My point was that the current science culture is currently as much about fame and pride as it is about understanding. In order to have any sort of anonymized, wiki-based science, we would need to ask people who have become famous and respected by doing things one way to start over in a new way.
There are downsides to competition but there are considerable upsides also. The bet we make in our culture is that the upsides outweigh the downsides, and based on history it seems to be true.