The Value of Wikipedia Contributions in Social Sciences

Summary

Improving society's wisdom, especially in comparison with its technological power, seems fairly robustly positive. Spreading knowledge of important findings—particularly in the social sciences, philosophy, and so on—is valuable, and one of the best channels for doing this is Wikipedia. While I wouldn't say that contributing to Wikipedia as a first-order priority is necessarily altruistically optimal except for very important topics, I would encourage people to add relevant material when they happen to come across it or if they already have deep knowledge of it such that adding it has low cost. If you plan to write a factual piece, consider whether you could add the material to Wikipedia instead of or in addition to reinventing the wheel. It's not clear what the sign is of Wikipedia articles that accelerate technology growth, so I tentatively recommend pushing on fields of knowledge that enhance reflectiveness and cooperation relatively more than technology.

Note: I created a list of ideas for Wikipedia articles to create, improve, or refine in case you want to jump-start your imagination. That said, I find it's generally easier to start with a paper I'm reading or a topic I want to learn about and then find what Wikipedia article can most appropriately include that information.

Introduction

When I was a child in the early 1990s, I once told my mom something like the following: "People should create a big book of all the world's knowledge, and then everyone can add to it the things they uniquely know about." My mom wrote this down in her journal of interesting things that her children said.

In 2005 I began hearing about Wikipedia from friends in school, who were remarking on the fact that anyone could edit it. Soon thereafter I began reading Wikipedia articles extensively, and they massively expanded my view of the world. A sizeable fraction of everything I know has come from Wikipedia. The rest of the world seems similarly enthusiastic, given that Wikipedia articles tend to show up at the top of search results, even when an official page for the topic also exists.

While I had been a consumer of Wikipedia for a long time, I only began editing articles in 2009, after hearing of some friends who did the same. I wrote a blog post encouraging readers to save interesting material they found to Wikipedia. In the present essay, I underscore that recommendation, specifically for contributions to social sciences, humanities, and other fields that help advance society's wisdom faster than it's technological power.

The value of popularization

When you read an academic paper, what section do you find most useful? The results? The discussion? The figures? Obviously this open-ended question has many answers depending on the circumstances, but often I find the most densely insightful section of a paper is the literature review. This is because I can get not just one particular study's (possibly noisy) findings but instead an overview of the whole field, helping me understand where the bulk of opinion lies and what kinds of general problems the field grapples with. Obviously if I were a specialist in the field, the literature review would merely repeat things I already knew, but for everyone else, literature reviews can be like gold.

Broadening horizons

Many commentators remark about the flood of information in which we find ourselves today. Because there's so much to explore about every topic under the sun, it's possible even in the age of the Internet to get stuck in a tiny subset of the space of intellectual insights. I'm skeptical of claims that the Internet encourages this balkanization on balance; there's a lot more informational balkanization that goes on unseen in poor regions of the globe that don't have Internet access. Still, it's important to break out of the narrow communities and niches that we might get stuck in, to take a more global perspective on what there is to explore and which topics are most altruistically important. Popularizations and literature reviews are essential for making this possible, by giving non-experts a glimpse into new domains that would otherwise require a lot of time to understand.

Of course, it's easy to criticize popularizations as being overly simplistic, and readers should not presume that just because they understand the intuitive summary that there aren't important details and qualifications remaining hidden. But the only alternative to popularization is leaving the insights to a few experts and missing out on the gains from expanding the minds of everyone else.

Insufficient incentives for good popularization

Unfortunately, there aren't enough incentives for academics to popularize their work. Tenure is based on formal publications, and sometimes popular books and articles can be seen in a negative light. Consider Carl Sagan:

Inevitably, Sagan's public success has drawn crossfire. "His talent for popularization is quite unusual in science," observes JPL director Bruce Murray, "so it carries some bitter fruit among elitists." [...] The more common view is that Sagan's critics are simply jealous. "Many scientists are envious of Carl," says planetary geologist Larry Soderblom. "Most of them come across in the media like a pot of old dishwater. Quite honestly, I think his time is better spent popularizing, because he's so skilled and we need it." ("His Cosmos a Huge Success, Carl Sagan Turns Back to Science and Saturn's Rings")

Similarly, on an Ask MetaFilter thread, one academic says:

Writing for the public will not get you tenure, or even count much towards it, in any field that I have any contact with, and unless you are one a few very rare people, the time spent on this would probably substantially hurt your case.

When you're optimizing for impact on the world rather than career success, you may end up with different priorities than those of academics.

Of course, there are also journalists, book authors, and other popularizers of intellectual findings. They are indeed valuable. However:

1. There's an externality problem. The social value of improved wisdom is much higher than the personal consumption value of reading interesting articles and books. Hence, we should expect too little popularization relative to the social optimum.
2. Since popularizers need to be paid by consumers, there's incentive to go for sensational, flashy stories at the expense of more careful, systematic discussions that portray the whole history of a field and its divergent viewpoints. One day we see a health news story about X reducing life expectancy; tomorrow we see another story about X increasing life expectancy. Better would be a summary of a meta-analysis combining hundreds of those individually noisy findings.

Intuitive argument for the value of popularizing over direct research

Research is valuable mainly insofar as the insights produced have an impact in society. (There are other side benefits as well, such as providing life experience to the researcher.) Sometimes most of the impact will come from influencing a few dozen colleagues who intimately understand the narrow field under discussion. This is especially likely for highly technical findings with less broader importance. However, many academic fields—especially social sciences, humanities, and pure sciences like physics—have significant implications for social policy and people's outlooks on the world, and these findings need to be more widely known.

Say it takes 10 person-months to produce a high-quality journal article in some academic field. It takes the reader at most a few hours to absorb the insights from that article; the benefit-to-cost ratio for the reader is much higher (ignoring the fact that the reader may want to do some direct hands-on work in detail to have a better picture of what science actually looks like). Now suppose that reader summarizes the core idea of the article in a few sentences on a related Wikipedia page and includes a citation. The benefit-to-cost ratio for Wikipedia readers is even larger: It takes just a few seconds for them to absorb those sentences, but now they have at least a taste of the important juices the article contained.

In the same time that it takes a scientist to conduct the studies for one paper, a popularizer could write summaries of tens to hundreds of such studies to share with others. If spreading a popularization of an existing study is even 0.5% to 5% as valuable as doing a new study from scratch, the popularizer is making a bigger contribution.

I personally see immense numbers of studies buried in old journals or academic websites containing fascinating insights that have not seen the light of day, because no one has written about them for a more lay audience. Empirically, there are plenty of low-hanging fruit to snatch up.

Standing on shoulders of giants

Many people love to get their hands dirty. Running experiments, creating new models, and so on are cool and fun. Indeed, they are, and there's value in trying some of them if only to get practice for what it's like. But before (or at least, in addition to) running off to develop a grand new theory or collecting new data, it's good to explore the literature that already exists on your topic. Luke Muehlhauser articulated the importance of this in "The Neglected Virtue of Scholarship." Popularization makes it easier for more people to stand on the shoulders of more giants.

Why Wikipedia?

Popularization can take many forms: Blog posts, articles on your website, YouTube vidoes, sharing links on Facebook and Twitter, and so on. However, I recommend adding the information to Wikipedia as a first resort, unless

• what you want to share isn't noteworthy or authoritative enough to be on Wikipedia,
• you want to make very long comments about the material rather than distilling its key insights, or
• you want to express an opinion about the material.

Even in the latter two cases, you can consider sharing the basics on Wikipedia in addition to writing about it separately. After all, if the material is already in your head, it should be low-cost to Wikipedify it.

Following are some reasons to prefer Wikipedia.

Readership

As of 2015, all the articles on "Essays on Reducing Suffering" combined get about 370 page visits per day. Assuming there are ~130 pages on this site, that's about ~3 views per page per day. (The distribution is far from evenly distributed, though.) In contrast, most of the pages I've created on Wikipedia get several times more views per page per day. For example, so far in June 2015, "Insect euthanasia" has been viewed ~12 times per day, "Welfare of farmed insects" has been viewed ~8 times per day, and "Fish slaughter" has been viewed ~4 times per day.

Of course, page views are obviously not the only consideration. My website has additional value insofar as

• the ideas are new and hence more counterfactually important than making available existing knowledge,
• readers of my website may be more affected by these essays, and
• my website allows me to express my opinions, which is more valuable by what I care about than readers perusing more factual content.

I sometimes see what I write on this website as enormously more important than what other people are doing, but in many cases, that's an illusion. Good research is good research regardless of who did it, and from a more distanced perspective, I realize that the information I add to Wikipedia can be enormously valuable for shaping how other people see the world, just like this website hopefully is.

Credibility

Not only does Wikipedia have more readers, but the content people read on Wikipedia will be taken more seriously than that expressed by a lone Internet blogger. If you want to share factual information rather than an opinion, Wikipedia will not only bring it to more people but will do so more persuasively.

Peer review

One reason for Wikipedia's credibility is its peer-review system. Making a contribution to Wikipedia is a signal that what you write can withstand inspection. Many editors on Wikipedia are very smart.

Concision and context

Blog posts popularizing academic findings are great, and we need more of them. However, they're scattered across the Internet and not easy to find. They also tend to review one or a few articles without necessarily putting them in complete context. (Usually when they do provide context, it's in the form of linking to Wikipedia articles!) Moreover, the popularizations may be long, not well summarized, and are not necessarily the best starting point for entry-level readers.

Of course, you could make your popularizations more Wikipedia-like, by being more thorough and more contextually interlinked. But in that case, why not just write it on Wikipedia itself? And if the content already exists on Wikipedia, don't reinvent the wheel. In general, the world would be better off with a single, best introductory article on a topic than several, not-as-good articles explaining the same things. (Of course, there's some allowance for creating a few different introductions depending on the background and age of the reader.)

In general, Wikipedia is the go-to place for the world's knowledge, and you're not going to beat it at what it does elsewhere. If what you want to write is suitable for Wikipedia, it should go there.

Fast turn-around

Publishing for a journal or other professional venue means waiting at least weeks, and often many months, for publication. With Wikipedia, you can publish immediately (or, with a new article, as soon as the content is substantive enough not to be deleted), and you get feedback right away.

I also personally find immediate publication more psychologically rewarding, because I can get a little "high" whenever I make a successful edit, rather than thinking of the project as a long slog where the payoff comes only in the distant future.

Epistemic discipline

Wikipedia's standards of high-quality evidence and representing multiple viewpoints are not just a burden to live with; they're in fact epistemic virtues that you should be aiming for on your own. Of course, our personal opinions on an issue will be more than just a mix of expert opinions, but we should know the main expert opinions and give them serious consideration. If your pet theory isn't suitable for inclusion in a Wikipedia article, how sure are you that it's correct? (Of course, there are many cutting-edge topics that do not yet have suitable presentations but should be explored. If we only relied on existing, authoritative sources, we'd have no new knowledge.)

A model of cooperation

If I had to envision the information-gathering stage of Eliezer Yudkowsky's dream of coherent extrapolated volition (CEV) in a concrete form, it might looks something like Wikipedia: People coming together, sharing insights, resolving disagreements, and so on. Of course, CEV has a significant moral component that's not present on Wikipedia, but the idea is not dissimilar, and improving wisdom is a big part of CEV as well. So to some extent, Wikipedia is like an achievable, first step toward a part of CEV. (Other features of the Internet and other social institutions like democracy are also crude but feasible approximations of a CEV-like approach.)

In general, Wikipedia is sometimes cited by pundits as a sign of hope for humanity—that individuals can voluntarily come together and build something amazing without debilitating levels of conflict. Wikipedia has extensive philosophy around consensus, dispute resolution, and so on.

Social praise

I suspect many people would feel that if you wrote 5 Wikipedia articles, that would be more socially valuable than 10 similar articles on your own website. This is because Wikipedia has greater readership, but also because it's an investment that allows your work to last, continue to be updated, and morph into something larger than yourself.

Writing your own material is good too, but there lingers a sense of ownership and status-seeking when writing something with your name on it. Because Wikipedia articles are author-less, contributing to them is a more pure signal of altruism. Of course, you can show off your Wikipedia contributions, as, for example, Pablo Stafforini, Vipul Naik, and my page demonstrate.

Low-hanging fruit for improvements

You might expect that most of the basic and really important material would already be on Wikipedia, so that additional contributions would just represent some bonus points. In fact, there are some major holes in Wikipedia's content, as Lance Bush discusses in "Low-hanging fruit: improving wikipedia entries." For instance, in Nov. 2013, I discovered that the famous Harvard psychologist Joshua Greene had no Wikipedia entry, even though he has done some groundbreaking work in the neuroscience of morality and is a well known author. Likewise, until recently, there were no Wikipedia pages for effective altruism, earning to give, or room for more funding.

We should not use Wikipedia as a self-promotional tool. Doing so would be one easy way to make enemies and hurt our cause. However, what we can do is notice the (sometimes significant) lacunae in the material that is available on Wikipedia and selectively add those topics that we think would have the highest social value if they were more widely known. This brings me to the topic of my next section.

Differential Wikipedia progress

In "Differential Intellectual Progress as a Positive-Sum Project," I highlight the importance of advancing society's wisdom relatively faster than its technological power. Before we develop ever more dangerous tools (in the fields of neuroscience, artificial intelligence, and so on), we should seek to improve social institutions, international cooperation, tolerance, and philosophical sophistication so that we can more circumspectly shape the future and achieve more of what everyone wants in expectation.

Given this premise, the value of contributions to Wikipedia in aggregate is not immediately obvious. While Wikipedia is an enormous boon for societal wisdom, it also advances technology at a faster pace.

My suggestion is that we focus our Wikipedia contributions on topics that tend more to improve wisdom: social sciences, philosophy, other humanities, and some natural sciences (e.g., cosmology) that are extremely illuminating but not directly applicable to engineering.

Of course, the net sign of engineering itself is not obvious. Maybe faster growth even in that field is better because it reduces the duration for which humanity is exposed to risks of bad outcomes due to suboptimal situations that technology could remedy. I don't know. My prevailing intuition is that technology will come in due course, and what we need to do first is make sure we can handle it in positive-sum ways for many value systems before it arrives.

Finally, note that popularization of technology is a different matter from advancing the state of the art in the field. Wikipedia's science and tech articles do speed up technological development by helping students and researchers learn faster, as well as possibly by encouraging more people to go into those fields. How much contribution this makes to the pace of progress in these fields I'm not sure. On the other hand, popularizations of technology are a big component of what we need to enhance debates in society about how to respond and what structures we can build to make sure technology is used for mutual benefit.

On balance, if I could costlessly contribute to all of Wikipedia's work, I would probably do so; the expected benefits seem to outweigh the costs. That said, there's high variance in this assessment, and it seems more clearly positive to selectively contribute to social sciences and related disciplines.

Currently I shy from contributing to ecology articles on Wikipedia because the information seems more likely to be used by conservationists than by people concerned with reducing wild-animal suffering. That said, if I had confidence that a given environmentalist issue aligned with reduction of wild-animal suffering, I would contribute to articles in that area.

Is contributing to Wikipedia cost-effective?

Currently I'm doing a lot of basic reading in many fields relevant to suffering reduction, and this process involves surveying the academic literature rather than producing new discoveries per se. If, as I'm reading an article, I find that its insights are worth sharing, I can pull up the most relevant Wikipedia page and add a few sentences to it. This is fairly cheap for me because I'm already reading the article anyway. I also find that

• writing to share with others is more motivating and fun than just reading in private,
• writing about an article forces me to pay attention to its core ideas and really try to understand it, and
• once I've written about the article, I have my notes about it on Wikipedia whenever I might want to return to them.

Of course, your mileage may vary on these factors.

Once you've overcome the startup costs of learning how to contribute to Wikipedia, it seems the main long-term cost is the time it takes to add citations, because this is mindless and isn't something I would do reading on my own. Overall, this probably isn't a big deal.

So, if you're reading interesting things, or if you already have extensive expertise in a topic and can cite relevant sources without difficulty, I think the benefit-to-cost ratio for Wikipedia contributions is high enough that you should do it, except maybe in some more technologically relevant fields, or if the topic is not very important intrinsically.

If you're not already reading things that are worth sharing, it's more doubtful whether doing so is optimal just for purposes of sharing on Wikipedia. Relative to some very unique skills you might have, Wikipedia contributions are somewhat replaceable, although the degree of replaceability can vary. Some of the contributions you might make—e.g., to pages on the mathematics of cooperation theory—might be sufficiently technical that only other very smart people could do them. Still, if you're not already reading about something for other reasons, you should probably let someone else with a comparative advantage write about the topic, unless you wouldn't be doing much else productive with your time.

In any event, whether it's worth contributing depends highly on what material we're talking about. Filling in a series of past episodes of I Love Lucy is a different matter from expanding articles on suffering in nature or the history of international cooperation. I think the latter topics could be important enough to deserve being written even if you weren't already planning to study them in depth.

Encouraging others to contribute?

It's plausible that if you wanted to increase social-science and philosophy contributions to Wikipedia, you could do this better by indirect means (i.e., encouraging others to do so), although this could be tricky if you don't want to also push on the tech articles. That said, tech articles may not be a big concern anyway on balance as discussed in the previous section.

I'm not suggesting that an organization promoting social sciences on Wikipedia would be an optimal charity to support. It's plausible there are better ways to promote societal wisdom faster, and I don't even know if that's the most efficient thing to push on—perhaps not relative to more direct efforts to strengthen compromise?

The big picture

It's important not to be insular in thinking about the value of Wikipedia: Even though most of the people it enlightens aren't part of your ideological circle and will never pay you back for the favor, if we take a distanced perspective of seeing ourselves, with our own ethical convictions, as one small piece of a larger project of humanity jointly working to figure out tough issues, we realize that our pet projects are not so momentously important compared with everything else as they might seem. Helping to edify those strangers may have its own ripple effects. I know there have been a few Wikipedia articles that have significantly changed my world view.

Selectively adding altruistically relevant material to Wikipedia is like treating the whole world as an arm of your altruism-research organization. Sometimes it's hard for us to see when we have our particular prejudices, but there's immense value in more people learning about these important issues—such as cooperation, global politics, cognitive biases, animal suffering, humane slaughter, and so on. Adopting a more Copernican view of our place in the altruism universe can cut against pride, but when we do so, we see that the extensive audience of Wikipedia is a more powerful place to make an impact than we might have thought.

Poll: Why altruists don't contribute to Wikipedia

I was curious to learn more about why more altruists don't contribute to Wikipedia, so I composed a poll for the "Effective Altruists" group on Facebook. Here are the vote tallies:

My replies to possible concerns

I personally find Wikipedia contributing extremely rewarding. When there are important topics that I want to become an expert on, the Wikipedia article is the first thing I should be reading anyway. Then once I'm familiar with that material, I can read additional papers in the field and summarize them on that article's page. When I have to summarize a paper, I read it more actively and more fully internalize its points, and I remember it much better. Indeed, my Wikipedia contribution gives me a sort of extended memory of what I learned. I've found that there's no shortage of altruistically relevant pages where lots of relevant content is missing.

When I read about a topic, some altruistic value comes from my own improved understanding, but if I contribute to Wikipedia, an additional source of value comes from helping many other people understand the topic better. Most of the time cost of contributing to Wikipedia comes from reading the source material, so if I'm going to read it anyway, I find little additional burden in adding a sentence or two summarizing it on a relevant Wikipedia article—assuming what I'm reading is worth sharing. I try to check that the citation I plan to add doesn't already exist by doing a search by keyword or author names. For instance, if the paper is by Ken Olum about the doomsday argument, I would search {olum doomsday site:wikipedia.org} and see if the paper is already mentioned.

It can be daunting to create new articles, so you might focus on adding one citation at a time to an existing article. If you're worried about needing to learn a lot of Wikipedia style to get started, I'd recommend reading just one or two introductory articles, starting to edit, and then reading more background if you discover that Wikipedia contributing will be a bigger part of your life.

If you're concerned about not getting recognition, one way to help with this is to list your contributions on your "User page." You may also list limited biographical information, including a link to your personal website. It's true that if you contribute to Wikipedia, your personal or organizational website won't get traffic from the content, but if you take a more impartial view about the value of different activities, this may be less bad than it seems. And in any case, the contributions you do make on Wikipedia will probably get more traffic than they would have on your site. The articles that I've created tend to get at least 10 views per day on average, compared against less than one view per day for most of the pieces on my own websites.^a You can also share your Wikipedia contributions on Facebook just as if they were a blog post or paper.

It's true that deletionist editors make the world a sadder place, though I suppose they do help keep Wikipedia's quality reputation high. In practice, though, I haven't found deletionism to be a big problem. Of all my contributions between Nov. 2013 and Jun. 2014, maybe ~4% were removed as of Jun. 2014. What's more, if your contributions are removed, you can just find the text and citations from the page's edit history and copy your additions onto your own website instead. I did this for an article section that I wrote on satisfaction in arranged marriages that was removed. (I wrote it out of personal interest rather than because of its high altruistic import.) Because contributions to Wikipedia are irrevocably put under a Creative Commons license, I listed the text that I moved to my site under the same.

In addition, Vipul Naik suggests:

I generally try to concentrate my attention on contributions where the probability of my edits getting overridden are minimal. These usually tend to be pages about organizations, books, people, etc. rather than general subject overviews. The criteria for the latter are much more moot and as a result there can be lots of edit wars etc. in those. I try to steer clear of general subject overviews for that reason (though I might make minor edits to already existing pages).

Wikipedia translations

In addition to adding content to the English Wikipedia, another possibility is to help translate existing articles to other languages. It's possible for even those who only speak English to help with this, although of course if you know the target language, that's even better. Here are some pros and cons for translation compared against contributing new content to the English article:

• Pros
  • Translation is faster than writing from scratch, potentially much faster. This means you can make available many more words of text per unit time than if you made novel contributions.
  • There may be fewer existing foreign-language articles on the web about a given topic, so the marginal value of an additional article may be higher in a foreign language.
  • Depending on your moods and inclinations, you might enjoy translating more (or less) than contributing to the English article.
• Cons
  • One of the most important downsides is that when translating, you learn far less personally than when adding new content. 90% of the effort of adding to Wikipedia for me lies in reading the source articles that I'm writing about or the Wikipedia page where I'm adding them, and those are things I'd want to be doing anyway, which is why I find Wikipedia contributing to have low opportunity cost. In contrast, translating would involve more menial work. Of course, translating would let you read the contents of the article you're translating, but at a really slow pace.
  • Foreign pages have fewer readers. To get some sense of proportions, here's a quick-and-dirty comparison of English vs. German pages using the stats.grok.se tool:
    • The English "Artificial intelligence" page gets 3500-4000 views per day. The German page "Künstliche Intelligenz" gets about 300, less than 1/10 as many.
    • The English "Roboethics" page gets about 60 views per day, compared against 8-10 for the German "Roboterethik" page.

    You could do a similar comparison for articles that resemble the one you'd like to translate.

  • I would conjecture that foreign-language readers who can't read English are somewhat poorer or less educated on average, which means they may also have less of an impact even if they read the information. That said, even very intelligent people may prefer reading their native languages, so this is only a weak point.
  • The English Wikipedia is likely to be the "state of the art" of the field (i.e., better than versions in other languages on average). It seems arguably most important to advance the "state of the art."
  • The English Wikipedia can often serve as a base from which content can be copied to other languages, so adding to the English version deserves some credit for those translations that will derive from it.

Another possibility is to work on Wikipedia infrastructure or tools rather than articles. This could have a big impact through automation. On the downside, (1) it benefits all articles, including those that may increase risks, and (2) it doesn't teach you about the contents of articles in the way adding information to articles does, though if you want to learn programming-related skills, developing Wikipedia tools could be a valuable experience in its own right.

Acknowledgments

I was inspired to expand this piece from an initial, shorter blog post in part because of a suggestion on the Effective Altruist Project Board about contributing to Wikipedia. José Oliveira Stor Zé and Peter Hurford inspired parts of the section about translating articles.

Footnotes

1. As Vipul Naik pointed out, the statistics reported on stats.grok.se include "much impure data, for instance bots loading a page continuously for whatever reason and any stupid crawler not using the API, etc." according to the FAQ. "Wikimedia Traffic Analysis Report - Crawler requests" reports that 19.8% of page requests come from bots. As another page notes, "On less popular wikis the share of bot requests will be higher."

   That said, there are other reasons to think even low view counts per day may not be overestimated on balance:

   • Some of my Wikipedia pages have traffic counts as low as 3, 2, or even 0 on some days, so this noise is probably not too high. Or else it's erratic, but then that doesn't explain the relative consistency from day to day.
   • The page views could also be underestimates because they don't "include requests to the mobile site, which is expected to serve about half of the pageviews at some point in 2015" according to the FAQ. In Jun. 2014, mobile traffic was 26.8% of the total.
   • Sometimes the counts logged are too low for various reasons.

       (back)