Jonathan Rascher’s NCC homepage

There are many like it, but this one is mine.

A brief ethnography of Stack Overflow

Last updated: 10 Feb. 2011

Introduction & history

Stack Overflow is a question and answer (QA) Web site dedicated to the issues faced by programmers, software engineers, and computer scientists. The Stack Overflow service launched in 2008 as a joint effort of Jeff Atwood and Joel Spolsky, two highly prolific bloggers and software developers. As the Stack Overflow FAQ explains, the site gives “professional and enthusiast programmers” a place where they can ask questions about “matters that are unique to the programming profession.” Other Stack Overflow users may then provide their own answers to these questions. Additionally, users may leave comments on both questions and answers, and questions and answers may be voted on by the site’s population.

In its relatively short lifetime, Stack Overflow has been a resounding success. Spolsky recently announced that site traffic grew 131% in 2010, with Stack Overflow receiving over 16 million unique visitors in the month of December 2010. Attempting to capitalize on the success of the Stack Overflow concept, they’ve assembled a horde of new QA sites, the Stack Exchange network, on topics ranging from IT administration to mathematics to photography to cooking. The largest of these sister sites are Super User, for amateur computer enthusiasts, and Server Fault, for system administrators. Although the combined question totals of these two sites presently make up only 10% of Stack Overflow’s question count, the user bases of these and many smaller Stack Exchange sites are active and growing.

In this brief ethnography, we will explore the community and culture of Stack Overflow, placing particular emphasis on the development of personal identity among users, the ability of the community to police itself rather than rely on a large staff of interventionist administrators, and the sense of trustworthiness and reliability induced by Stack Overflow’s “reputation” system.

Experiences & impressions

The author has been a Stack Overflow user since August 2009. Operating under the display name “bcat,” he has asked two questions, left 82 answers, and cast 4097 votes on other users’ content. At first, the dissonance between the his question, answer, and vote counts might be somewhat jarring—the ratio of the author’s votes to answers is roughly 50:1, and the ratio of his answers to his questions is similar. In particular, the lack of questions might be taken to indicate a lack of participation during the 18 months of the author’s membership; however, the number of questions a user asks is a poor measure of his or her activity on the site. Among Stack Overflow’s 35 most respected users (as ranked by user reputation), only one user has asked more than 100 questions, and three of the top 35 users have asked no questions at all.

While the author cannot claim to rank among the top Stack Overflow users, he has been sufficiently active on the site to amass 3244 of the reputation points alluded to previously, granting him access to limited moderation functionality, including the ability to edit questions and answers, to categorize questions, and to nominate poor or misplaced questions for closure or relocation to another Stack Exchange site. Additionally, as a computer science student and a “computer geek,” he has found Stack Overflow to be an invaluable aid in his academic and recreational pursuits. Increasingly frequently, he has noticed that Web searches pertaining to programming problems return Stack Overflow questions which have received the very answer he is seeking, or which contain sufficient additional information to enable him to solve the problem and perhaps even post an answer of his own. Moreover, in the author’s experience, the Stack Overflow community is remarkably civil, with the vast majority of Stack Overflow users ready, willing, and able to help fellow community members with challenging technological problems. The remainder of this report shall discuss three defining characteristics of the Stack Overflow social structure which, in the author’s opinion, have contributed greatly to Stack Overflow’s success. We shall begin by analyzing the nature of personal and collective identity in the community.

Identity & social norms

When a new user signs up for Stack Overflow or another Stack Exchange Web site, it quickly becomes apparent that they have not joined a social network site. In a time where many Web service providers are racing to add “social” feature to their sites, Stack Overflow differentiates itself by not being explicitly “social.” In particular, the Stack Overflow community lacks any means for its users to “friend” each other, or to otherwise define connections between themselves. Any sort of private messaging facility is similarly absent. The Stack Overflow community is highly focused on its goal—providing a useful QA service—and off-topic uses of Stack Overflow are frowned upon. Even popular questions risk being closed if they are deemed not to serve the site’s mission. This phenomenon is exemplified by a popular question soliciting funny or insightful quotations about programming, which was deemed to be “not a real question” and unceremoniously closed, preventing users from posting further answers.

Having seen this, we might wonder whether Stack Overflow should be considered a community at all, or we might suspect that, like David Bell’s hypothetical “community of car drivers,” it fails to provide the sense of identity and closeness associated with a genuine community. However, such concerns betray an overly narrow notion of community which fails to recognize the ability of even focused and businesslike Web sites such as Stack Overflow to help patch the hole left in the human psyche by the oft-maligned disintegration of traditional, real-life communities.

We have already alluded to the almost overwhelming civility and hospitality of Stack Overflow users, who take time out from their busy lives to help others with programming problems harboring no expectation of rewards other than those which intrinsically accompany helping one’s fellow man. Furthermore, though it is not a social site, Stack Overflow is far from a colorless place where questions are asked and answered without passion or personality. Instead of defining their identities through social connections, members of the Stack Overflow community define themselves by the content they produce, i.e., their questions, answers, and comments. Also, despite the community’s disdain for off-topic posts, Stack Overflow users are quite accepting of humorous questions and answers, particularly those expressed with a sort of ironic jocularity—something demonstrated by the popularity of a recent question on Server Fault that inexplicably asks the name of a common device (an extension cord) using an illustration of intentionally poor quality (see Fig. 1). In short, membership in the Stack Overflow community can be a rewarding and intellectually-stimulating experience, albeit one which favors depth of knowledge and a sense of humor over more mundane social interactions.

[A cropped sketch of an extension cord, power strip, and AC adapter with intentionally poor, overly minimalist image quality.]
Figure 1: A user’s artistic rendering of an extension cord.

Moderation & participation

We have stated multiple times that the bulk of discourse on Stack Overflow is polite and civil; however, this report would be both dishonest and unbelievable if we did not acknowledge that Stack Overflow attracts a certain number of users with undesirable, antisocial behavior. Like any online community, Stack Overflow must deal with the inevitable influx of “trolls”—those who seek to harass and annoy legitimate users—as well as “flamers”—those who are legitimate users themselves but do not choose to behave civilly in their interactions with other users. Many young communities die a premature social death because their administrators are unprepared or unwilling to devote the time and effort needed to properly police community content. On Stack Overflow, however, moderation abilities lie almost entirely into the hands of ordinary users. As a user’s reputation on the site increases, he or she is granted increased powers, including the ability to edit and request closure or deletion of the posts of others. Therefore, while certain trusted community members are given the specific title of “moderator,” site tending is not restricted to this group; it is distributed among the entire user base.

Reputation & reliability

We have referenced Stack Overflow’s reputation system frequently in the previous four sections, and now we shall discuss it in more detail. Intimately entwined with participation in the Stack Overflow community, reputation is, according to the Stack Overflow FAQ, “a rough measurement of how much the community trusts you.” A user gains reputation points when his or her posts are voted up by one of the user’s peers. Conversely, a user loses reputation points when a peer votes one of his or her posts down. In this way, voting serves as more than a moderation tool—it also gives users an easy way to reward good contributions and penalize contributions which are found to be lacking in some way. As mentioned in the previous section, this reward includes increased ability to police the content of others, culminating in the receipt of access to a helpful moderation dashboard when a user amasses 10,000 reputation points.

Were this the only function reputation serves on Stack Overflow, it would hardly merit its own section in this report; however, reputation plays a much more critical and direct role in the community than simply restricting access to moderation facilities. Reputation and its primary accoutrement, vote totals, are displayed publicly and prominently throughout the entire Stack Exchange network. Most crucially, reputation is displayed on every user’s personal profile page, and it accompanies all the user’s questions and answers. Though Stack Overflow does allow participation by anonymous users, participants are strongly encouraged to choose user names and create accounts for themselves. It has been repeatedly proclaimed and demonstrated that tying together user reputation and user identity can strongly discourage antisocial behavior while simultaneously encouraging new, low-reputation users to become active, productive community members.

The centrality of reputation to the Stack Overflow ethos also practically forces users to refactor their posts mercilessly, often leading to competition among users as they race to provide a complete answer to a question as quickly as possible. Users are implicitly encouraged to draw on the answers of others, and not infrequently the highest-voted answer to a question will be little more than a compilation of several older answers. Some have argued that the site’s heavy emphasis on reputation discourages new participants and is thereby undesirable, but the benefits to the programming community as a whole seem clear. Stack Overflow’s reputation system has shaped the site into a meritocracy where there is generally a direct correlation between reputation and trustworthiness. This enables Stack Overflow users to determine with a good degree of accuracy how likely it is that an answer they read on the site is correct.

Conclusion & future research

In this report, we’ve endeavored to provide a succinct yet accurate analysis of Stack Overflow. We first explained the general social atmosphere of the site, explaining the nature of this rather specialized, nontraditional community. Second, we summarized the moderation philosophy of Stack Overflow, hypothesizing that the service’s community-centric, distributed approach to moderation has allowed Stack Overflow and its sister sites to avoid many of the pitfalls facing young online communities. Third, we discussed the reputation system implemented on Stack Overflow, considering the development of the community as a meritocracy and the sense of dependability this provides to the site.

This brief ethnography has touched on three of the primary characteristics which have shaped the unique QA community of Stack Overflow and other sites in the Stack Exchange network. The scope of this report has been limited by time constraints, and it seems that further ethnographic research into Stack Overflow might prove enlightening. The author is particularly curious about the effects of the recently-launched Stack Overflow Chat feature, which promises to provide a forum for informal and immediate conversations which may not meet the standards of a more formal QA site.

Works cited

Credits

The Stack Overflow logo and site design are © 2011 Stack Overflow Internet Services. The author believes that his inclusion of the logo in this report qualifies as fair use due to the nonprofit, educational purpose of the work. Additionally, user-generated content from Stack Exchange Web sites is used under the terms of the Creative Commons Attribution-ShareAlike license. The views and opinions expressed in this document are the author’s own, and they should not be taken to represent the views and opinions of any other Stack Overflow users, or of Stack Overflow’s administrators.