October 23, 2017

Wikipedia as a Public Good

My post about Wikipedia and public goods prompted an interesting response from Judd Antin at Berkeley’s School of Information. He makes a number of sharp points, but let me focus on this response to the idea that free-riders don’t hurt Wikipedia:

This completely depends on what your goal is. On the one hand, sure, once you reach critical mass, the marginal cost of providing the good is zero (or near-zero), so who cares how many free-riders there are. On the other hand, there are lots of benefits to adding to the group of contributors. Wikipedia isn’t perfect – not even close. It’s wrong on a lot of topics. It’s poorly written in many places. It’s skewed heavily towards CS and popular culture, and away from things like history and literature. There’s a lot to be gained for Wikipedia by converting free-riders to contributors. And let’s not foget about the many, many systems that never get to critical mass.

Notice the background assumption in this passage that the numbers of free-riders and contributors are inversely correlated. That is, it assumes that each free-rider is in some sense detracting from the overall effort by failing to contribute. And it suggests that a better-designed institution might be able to convert more of those free-riders into contributors and thereby increase the total value of the public good that’s being provided.

The problem with this, I think, is that it gets the sign of the correlation between free-riders and contributors backwards. That is, on the Internet, the potential audience is so enormous that the most effective way to get more contributors is to increase the total number of users. For Wikipedia, there isn’t so much a “free-riding problem” as there is a “free-riding opportunity”: the more free-riders there are, the easier it will be to recruit new contributors down the road.

Antin also seems to be under the impression that I’m suggesting that we need to “rethink 100 years of theory” about public goods, which is certainly not the case. Standard arguments about public good problems work perfectly well for in cases like national defense and clean air, and I’m certainly not suggesting that we need to re-think those cases. My claim is simply that the standard model (and especially the focus on free riding) just isn’t a helpful way to think about online content creation, a public goods problem with dramatically different characteristics from clean air or national defense.

Comments

  1. That’s a great point. After all, contributing to Wikipedia requires some minimum amount of knowledge about how the whole thing works. And the main way anyone gains that knowledge is by free-riding for a while.

    But we can have both at the same time, can’t we? On the one hand, more free-riders means more potential contributors. So the free-riders don’t hurt per se. On the other hand, if Wikipedia could convert more free-riders to contributors, that would help them too.

    And, I admit to some hyperbole of my own. I don’t really think you were upending 100 years of theory. I’m just sensitive to assumptions that online collective action is somehow this completely uncharted territory in which none of our pre-existing ideas apply. You weren’t saying that, but lots of people do when they write about this space. Really, I think public goods / collective action / social dilemmas / social exchange theory continue to provide a great framework for understanding Wikipedia, Flickr, blogs, Facebook, etc. We just need some updated ways to understand the context.

  2. It still bugs me that people still say that wikipedia isn’t perfect because it isn’t a perfect replacement of a perfect repository of all human knowledge – or that it isn’t perfect because more Buffy fans contribute than Dostoevsky fans.

    Wikipedia IS perfect – it’s just that imperfectness and amateurness are built into the fundamental architecture of the thing. How many times do we have to tell you – wikipedia is NOT a source!

    —-

    Anyway – wikipedia is nowhere near a new idea. The only new thing is that it condenses the normal information gathering cycle. Wikipedia is just like the collective engineering knowledge developed over the decades in a workshop or factory – everyone adds a little bit of information, and if it turns out to need correcting it just gets corrected.

  3. Wikipedia isn’t a repository for knowledge, it’s a collection of opinion. So it’s a great database for studying the social dynamics of superstition, but not for much else.

  4. I don’t think this is a popular sentiment, but I disagree with Antin’s statement that “[t]here’s a lot to be gained for Wikipedia by converting free-riders to contributors.”

    The passage implies that a large proportion of the people who read Wikipedia but don’t contribute would make useful contributions, and in particular it would reduce the “wrong[ness] on a lot of topics” and “poor” quality of writing “in many places.” I believe that there is a lot of particular knowledge held by individuals that would be informative to the public if it could be meaningfully collected in one place, but it is another leap to assume that these people could meaningfully contribute through the predetermined and frankly quite restrictive medium of Wikipedia.

    Additionally, on certain topics, people with superior firsthand knowledge might be better off not having free-riders become interfering contributors. Scientists who are experts in an esoteric field have an interest in seeing their research not be misrepresented in popular media (think of all the frustration scientists have with journalists reporting on their field), and would probably prefer to create these articles before laypeople do.

    Finally I think if Wikipedia content is “skewed” toward CS and popular culture, that might be a reflection of a skewed contributor group but more likely it’s just a reflection of the state of general knowledge in this country. Why is it not Britannica — with a small board that dictates its selection of articles — that is skewed toward history and literature?

    p.s. since when did “inverse correlation” become synonymous with “negative correlation”? I mean it’s used correctly here but I had always thought that “inverse correlation” meant “directly correlated with the reciprocal” and I’m not sure if I was taught wrong in high school or if it’s one of those language evolution things.

  5. I find Wikipedia excellent for history, at least much better than anything else I have easy access to. I do read the occasional book on history but then you have to follow the narrative of that particular book, with Wikipedia you wander around where you want to go because everything is hyper linked.

    There are lots of arguable points in history, and that’s never going to change. In most cases, the “many authors” approach to Wikipedia will at least ensure that all the major arguments are mentioned, while the author of a book will push their understanding and conveniently omit any counter arguments.

    As for popular culture, why shouldn’t it be there? If you don’t like it, don’t read those articles. Wikipedia has quite a decent index, I cannot say I’ve ever been forced to thumb through pages of irrelevant rubbish trying to find the article that I wanted (unlike pretty much every paper medium and many poorly indexed academic journals).

  6. CattyNebulart says:

    Wikipedia fulfills its goals quite admirably, it’s error rate is about on par with Encyclopedia Britannica and it is far more up to date. Granted there are only 4 things I tend to look up on Wikipedia, CS, Animals, History and popular culture, on each of which Wikipedia is quite strong. If anything it lacks information in the Popular Culture category when I do research, but then Wikipedia doesn’t try to be a complete source.

  7. I am sorry but I can not properly understand the meaning of your phrase puiblic good. Could you write about it little more? Thank you Jackie from home electric flame fireplace inserts