I. The question, and some distinctions
Should science communication be collaborative? There are two ways to understand this question, and so also two very different reactions to it. One reaction is that science writing already is very collaborative. Scientific articles are typically co-written by labs or by other collections of colleagues, because most experiments cannot be done by just one person; scientific discoveries are now typically made by several or many people cooperating. So, of course science communication should be collaborative.
The other reaction understands me to be talking about collaboration in the wiki sense, or what I call radical collaboration. And to that question there are typically mixed reactions. On the one hand, what Wikipedia has done is very exciting, and if scientists can tap into the same sort of collaboration, perhaps great things will result. On the other hand, scientists and scholars in general are very suspicious of the notion that anybody can edit our words. Many scholars scoff at Wikipedia’s motto—”you can edit this page”—as incontrovertible evidence that it cannot be very reliable.
The question I am interested in is actually the latter one: should science communication be radically collaborative? So let me define this piece of jargon. Collaboration is radical if it goes beyond two or more people merely working together. In addition, the collaborators are self-selecting; they determine what they are going to do, and are not assigned their roles. Finally, there is equal ownership or equal rights over the resulting work, or in other words, there is no “lead author.”
So, should science communication be radically collaborative? I cannot give you any simple answer to this question, but I do want to say that radical collaboration is part of our future, and will probably result in some amazing new scientific resources. I’ll be asking how big a part of our future it should be—as well as what we should not expect radical collaboration to do.
But first, it will be useful to draw a distinction between two kinds of scientific communication: original and derivative. Original communication is aimed at advancing knowledge in the field with never-before-published findings, discoveries, first-hand accounts, survey data, theories, arguments, proofs, and so forth. Typically, such communication takes the form of papers in peer-reviewed journals and online pre-print services, as well as conference presentations, posters, and some other things. By contrast, derivative communication merely sums up what is already known, and takes the form of news and encyclopedia articles, textbooks, and popular science books and magazines.
I don’t pretend that the distinction between original and derivative communication, if one examines it carefully, is easy to make. One reason that it is difficult is that, whenever one reports scientific and other scholarly findings, analysis almost inevitably occurs; and sometimes, an analysis can be as interesting, challenging, and pathbreaking as the findings reported on. So I imagine that such interesting analysis can be a borderline case between original and derivative communication.
There is another reason the distinction is difficult. Frequently, we want to criticize certain published papers, which purport to present original findings, as being almost wholly derivative—they do not really advance the field at all. I am told that this happens much more than it should, in scientific publishing. So I admit that sometimes, purportedly original communication is actually derivative.
In fact, I will admit something more: it is far from clear what constitutes an advance in any given field. If someone merely deduces something from previously published experimental findings, is that an advance? Sometimes, sometimes not. If someone does an experiment that is only trivially different from any of many already-published experiments, and obtains similar results, is that an advance? Not necessarily, it seems to me. If someone merely applies an established paradigm to a domain of knowledge for the first time in a published article, is that an advance? Perhaps; but perhaps not, if the application was simply obvious.
So there are, I realize, several reasons to be critical of the distinction between original and derivative communication. That admitted, I do think there are many perfectly clear cases of both original and derivative communication; in fact, I think most scientists and scholars would not have trouble classifying most communication in their fields as either original or derivative. When Watson and Crick originally described the double helix, that was definitely original. When Wikipedia, or a biology textbook, describes the double helix, that is definitely derivative. And where we are uncertain, on philosophical grounds, about whether some finding really is original, at least we can tell whether the author is treating it as original.
I draw this distinction because I think that we might actually wish to give different answers to the question, “Should science communication be collaborative?” based on what type of science communication we’re talking about. In particular, I think it is very plausible that derivative science communication, like encyclopedia articles and science news reporting, are much more amenable to collaboration than original science communication. I think, moreover, that in explaining this we will uncover some very interesting insights, or at least questions, about collaboration and perhaps even about science communication itself.
II. Derivative science communication
Let me begin with derivative science communication—again, things like encyclopedias, science news reporting, and textbooks.
Over the last few years, I have conversed with dozens of scholars and scientists about how to set up wikis or other collaborative knowledge communities. There is a fascinating pattern to these conversations. They go like this. The scientist, impressed by the vast quantities of information in Wikipedia, tells me: “It is amazing what can be accomplished when many people come together, from around the world, to sum up what is known. What would happen if we tried this in our field? The resulting resource could be a central, authoritative clearing-house of information for everyone in the field, as well as for the general public. So, what is the best way to set up ‘a Wikipedia’ in our field?”
This is an interesting question, but it is not the question that they end up answering. Instead, the scientist goes off and consults with his colleagues, and then I hear this: “We have a couple of concerns. First, we are concerned about lack of credit in the Wikipedia system. The careers of scientists depend on names being on their publications. So we want to make sure that authors are properly named and identified on articles. Second, we are a little nervous about the idea that just anybody can edit anybody’s articles. We understand that it’s important to be collaborative, but we think it is reasonable to nominate a lead author or lead reviewer for each article, and restrict participation to experts. So, what do you think of that?”
I think that the scientist and his colleagues are confused in a fascinating way. I try to be diplomatic when I say this, of course. But the scientist seems not to realize two facts:
- If you name authors, you award lead authorship or editorship for articles, and you carefully restrict who may participate, then you are not building a collaborative community in anything like the radical sense. You are merely using a wiki to replicate an older sort of collaboration, common in scientific writing.
- It is precisely the newer, more radical sort of collaboration that explains Wikipedia’s success. Wikipedia is successful in large part precisely because everyone feels empowered to edit any article. If you disempower people, they won’t show up.
As a result, there is no reason to think that the scientist’s group will enjoy success anything like Wikipedia’s, because they have actually rejected the Wikipedia model.
I am not saying that using wiki software to replicate old-fashioned systems won’t work at all. In fact, in 2005, I helped set up such a system myself, called the Encyclopedia of Earth, and it seems to be working reasonably well so far—but, as far as I know, not much actual collaboration goes on, and a large part of the few thousand articles that they have were imported from other sources. Another scientist-run encyclopedia, Scholarpedia, has a somewhat similar set of policies, and has produced even fewer articles. To be sure, the quality of the articles produced by these projects is good. But it seems to me that the articles have little chance of ever fulfilling the original, high hopes of the project designers. Many of them won’t be incredibly detailed, balanced, authoritative, and a pleasure to read, which is what one might hope to get from a large group of experts coming together to work on a piece of text. Nor do such projects have any chance of achieving the depth of coverage that Wikipedia has. In short, as far as I can tell, the most that projects like the Encyclopedia of Earth and Scholarpedia can hope to achieve is to produce a free version of old-fashioned sorts of encyclopedias. I do not mean to say that there is something wrong with that. I merely claim that they will not enjoy the advantages and potential that a radically collaborative project has, the advantages and potential that made them imitate the Wikipedia model in the first place.
This, then, raises a question. Do those scientists, who have rejected the Wikipedia model, have a legitimate complaint about it? Or have they made a mistake in rejecting it? I think they are partly right in rejecting the Wikipedia model, but also partly mistaken. Let me clarify, first by explaining what they have gotten right.
Essentially, the scientists I’ve advised are quite right to reject the wide-open Wikipedia model, according to which anyone can alter any article regardless even of whether the person has logged into the system or is using his or her real name. Wikipedia’s rock-solid commitment to anonymous contribution explains many of its problems, in my opinion. It explains why Wikipedia has so much vandalism and people editing abusively and in bad faith; it also explains why the Wikipedians have never been able to enforce some of their own basic principles, such as neutrality and politeness. Scientists and scholars generally are very well justified in rejecting Wikipedia’s anonymity policy. I have argued for this thesis elsewhere, and can’t spend the time to explain arguments now.
So that’s why my scientist colleagues were right to reject the Wikipedia model. But they are also mistaken to believe that articles must be signed by their authors, that they must have lead authors, and that participation should be restricted to experts. They believe they must adopt these policies because, otherwise, the result will be unreliable or of poor quality. They appear to think that, since all trustworthy encyclopedias in the past had signed articles, lead authors, and participation restricted to experts, there is no way to design an encyclopedia project that changes these features.
Now, I don’t have time in this paper to argue for this point in detail, but I simply want to point to the example of the Citizendium, which is a wiki encyclopedia project I started a year and a half ago. We do not sign articles; we do not have lead authors; and we open participation up to anyone who can make a positive contribution to the project. But we do make a role for experts. Despite the fact that we reject so much of the traditional model of content production, the quality of our articles is remarkably good, especially for such a young project. The articles that have been approved by our expert editors, in particular, are extremely readable, as well as being authoritative. My point, then, is that it is possible to have a radically collaborative system that produces high-quality, credible content. So if my scientist colleagues rejected radical collaboration because they thought the results would necessarily be of substandard quality, they were simply mistaken, as our experience with the Citizendium shows. Moreover, I should point out that we are far more productive than Scholarpedia or the Encyclopedia of Earth; we have over 7,000 articles and are growing daily.
I can imagine a reply to this, however. One might concede that the Citizendium‘s articles are, or will be, of reasonably good quality. But will they be better than articles written by small groups of experts? Not necessarily, of course. Still, I would like to give you some general reasons to think that they could be better. More precisely, I want to answer this question: is there something about radical collaboration per se that improves the quality of articles? I think so.
Given enough time, an article that is written with a large and diverse set of authors—particularly if it is under the gentle guidance of experts—can be expected to be lengthier, broader in its coverage, and fairer in its presentation of issues, than an article written by a single or a few hand-chosen authors. It will be longer, because many collaborators will compete with each other to expand the article. It will be broader in its coverage, because the collaborators often can fill up gaps in exposition that others leave. It will be fairer in its presentation of issues, because self-selecting collaborators in a very open project will tend to have a diversity of views, and they must compromise in order to work together at all.
In short, radical collaboration naturally pushes articles in the direction of being longer, more detailed, and fairer. When the collaboration is gently guided—not led and controlled—by experts, and when the collaborators respect the experts and are willing to defer to from time to time and when necessary, the resulting articles can be outstanding. A number of the Citizendium‘s approved articles are outstanding for these very reasons. We have quite a few outstanding unapproved articles as well.
So far, I have spoken only about one kind of derivative communication: encyclopedias. But there are other kinds, as I said: journalism, textbooks, and popular science writing, for instance. I could discuss each of these, but again I lack the time. Instead, I want to make a general point about all of them.
Often, in expository writing and even more in fiction writing, we derive value from the text precisely because it is personal, because it presents a single, unique point of view that we find compelling. We find the writing interesting because we find an individual mind interesting. Why are we fascinated by the minds of Stephen Hawking, Richard Feyman, Stephen Jay Gould, or Steven Pinker? (And for that matter, why are so many famous scientists named Stephen?) Well, it seems that, in works by these authors, the addition of another author might subtract from the value of their text. Why is that? Why is it that we find individual minds interesting? It is not because their thoughts are more accurate or more exhaustive. Rather, a text with a single author, especially one who is expressing his personality, is a window into another mind, and so it represents how we, each of us individually, might also want to think. Only an individual seems to be able to serve as a credible model of how to think about the world; and, for whatever reason, we do take other thinkers as models. Collective productions can convey useful information, of course, but they necessarily do not express the views of any one person. They are largely useless as complex, full-bodied, human models after which we can pattern our own thinking.
But almost all encyclopedia articles, most news articles, and some textbooks are used just to get information, not to serve as an entrée into an interesting perspective on the world. Idiosyncracy and personality are annoying when we merely want information. When it’s bare information we want, we don’t care about persons—only facts. The point, then, is that radical collaboration is suitable for gathering impersonal information. That, we might say, is its proper function.
III. Original science communication
Up to this point, I’ve been talking about whether derivative science communication should be collaborative; the answer, in short, is yes and no. So now let me talk about whether original science communication should be collaborative. But first, I think we need to examine whether, and in what sense, original science communication, such as papers that express new research findings, can be radically collaborative. Maybe a better question is this: to what extent can original science writing be collaborative? We already know that scientific research can be collaborative in the old-fashioned sense, because it is so often is, in fact. What is the feasibility of making it more collaborative?
Applying certain aspects of the Wikipedia model to original science communication—and even the Citizendium model—strikes me as simply impossible. For instance, if research papers were not signed, but instead were attributed to a nameless collective, the traditional motive of scholarship—personal glory, the honor of one’s peers and of history—would disappear. In short, I very much doubt scientists would participate at all in a researech collective without definite personal credit. We may not need prominent personal credit to create derivative works collaboratively, but original works are another matter entirely. Indeed, the economics of the two kinds of communication are different, because our motives are different. Many scholars and scientists will not write an encyclopedia article, news article, textbook, or a popular science book without some compensation. But the same people routinely publish much more difficult research papers and monographs with no monetary compensation. The glory and honor of discovery is the motivation for such work. Wiki work is just not that glorious, or at least, not in the same way.
Another aspect of radical collaboration is open authorship, that is, the authors select themselves. This again seems impossible, or very difficult at best, for original science communication. For one thing, original communication expresses original thoughts, and such thoughts necessarily tend to be controversial and difficult. To open up authorship of original work very wide would, hence, permit the participation of persons who disagree with the conclusions or who don’t even understand them. But if participation is limited to like-minded scholars who understand the research, the collaboration can no longer be called “radical.” It’s just a variant on old-fashioned collaboration.
In fact, beyond issues of feasibility or difficulty, I detect an incoherence in the very idea that original research might be radically collaborative. The act of publishing a research paper does more than merely convey some findings; it also stakes a claim, that is, it has the force or effect of attaching some definite name or names to the findings. To make original science communication radically collaborative would be to nullify the act of taking credit. If we were to list as co-authors people who are not responsible for the research, the author list would not longer be honoring those people actually responsible for the finding. It would just be a list of people who happened to work on the paper that summed up the research, even if some of the people listed had none of the thoughts or conclusions contained in the paper.
One might say that open collaboration on communication of original research would help to elaborate the full range of arguments and analysis releated to the research. But that already happens, I suppose, in the give-and-take of scientific and scholarly conversation that happens before and after a paper is published. Indeed, it has often been observed that science and scholarship generally are massively collaborative in the sense that researchers build on each others’ work; it was Newton who pointed this out when he said that he saw farther only because he stood on the shoulders of giants. I have no doubt that new Internet methods can and already do facilitate this very old sort of scientific collaboration. But I see no need, in addition, to permit others, who had nothing to do with some research, to participate in the writing itself of original research findings.
That said, there is at least one way that original science communication might be amenable to radical collaboration: I mean what has been called “open research” and “open science.” As I understand it, this involves inviting others to participate actively in a study—not merely collaborating on the writing, but actually doing the research for, designing, and performing experiments, surveys, and so forth. This is something I know very little about, and I will not embarrass myself by pretending to know more than I do. An example of such research, perhaps, was the lightning-fast investigation in multiple labs that identified the avian flu virus. Such research can be somewhat open and self-selecting. So perhaps that is one sense, and a very interesting sense, in which original science communication can be radically collaborative. I’m afraid I can’t presume to say anything else about that, though.
So, to sum up, should scientific communication be collaborative? I’ve made it clear, I hope, that it depends on the type of communication. Derivative communication that merely aims to express impersonal information can, and in some cases perhaps should, become radically collaborative; the Citizendium system shows how. But when a specific personality, or point of view, forms an important part of the value of the communication, collaboration is denaturing and devaluing. And original scientific communication should be collaborative only to the extent that the research it reports has been collaborative.
In the interests of keeping this paper short and provocative, I have not answered many important questions. Perhaps the most important unanswered question is: what constitutes a contribution to knowledge? Also, I said that some derivative communication should not be collaborative, because its value depends on its coming from an individual mind; I said that the productions of individual minds sometimes have some special value because they “model” how to think about the world. What do I mean by that, and what is valuable about it? I also asserted that scientists would not participate in research programs without the expectation of credit. That seems obvious, but perhaps I should have explained why not; that is really a core issue. Finally, I only barely glanced at the prospects of open research, or open science. What is such research, really? Is it radically collaborative in anything like the wiki sense, or is it merely the practice of making our research available to others for free, and talking a lot?
Without having given clearer answers to these fundamental questions, I can’t say I have adequately discussed whether science communication should be collaborative. Clearly, this is a big question, with many ramifications. But I do hope I have at least introduced a few of the salient issues and given you something interesting to think and talk about.