Online communities survive on newcomer pipelines and a low cost of contribution

Year 4 / Module 30 Social computing and computer-supported cooperative work / Unit 5 / Online communities survive on newcomer pipelines and a low cost of contribution

Online communities survive on newcomer pipelines and a low cost of contribution

HCI4106 · Unit 5 of 10

A community's health is set less by its founding design than by whether it absorbs newcomers faster than members leave and keeps the marginal cost of a contribution low, which is why popularity itself can destroy a peer-production system.

Online communities survive on newcomer pipelines and a low cost of contribution

Whether a community still exists next year depends less on what it was founded to do than on whether it is replacing the people who leave. The fifth of ten units argues that two flows set a community's health. One is the absorption of newcomers against the departure of members, and the other is the marginal cost of one more contribution. Popularity is not health. Growth raises coordination cost and the chance that a newcomer's first attempt is rejected, so a system can be undermined by its own success (Butler, 2001; Halfaker et al., 2013). The next unit turns from the community to the self each member presents inside it.

Founding design is not what decides a community's fate

Kraut and Resnick (2012) organised online community design around five challenges, from starting a community to regulating behaviour, and stated the evidence behind each design claim. It takes the public-goods reading from Kollock and Smith (1996), where free riding is the default prediction, and treats commitment as a matter of interpersonal bonds. The format is also the limit: a claim that a design raises participation is only as strong as the studies behind it. The challenge that carries the others is the newcomer one, because contribution and commitment are exercised by people who remain.

Faraj, Jarvenpaa and Majchrzak (2011) give the reason a founding design cannot bind a community's future. What affords collaboration in an online community, they argue, is its fluidity: boundaries, norms, participants and artefacts keep changing, and the resources people bring, passion, time and identity among them, arrive and leave, each carrying a cost as well as a benefit. Their running case is Wikipedia, where the roles that resolve a tension are enacted by whoever is present rather than written in. It is a theory with examples rather than a test.

A community is a flow, not a stock

Active registered editors, counted as five or more edits a month, peaked in March 2007 at 56,400 and declined for years. The share of desirable newcomers whose first-session edits were reverted rose from 6.1 per cent in the first semester of 2006 to 18.2 per cent a year later. Quality control was working, and that was the problem. Two-month survival of desirable newcomers stood at 25.6 per cent in early 2006 and fell to 11.7 per cent. By late 2007 more than half of new users received their first message from an algorithmic tool, and by mid-2008 the figure reached 75 per cent.

The leak is at the entrance, and the receiving practices set the slope

Panciera, Halfaker and Terveen (2009) compared editors who became heavy contributors with those who did not, and found the difference visible from the first day. Eventual Wikipedians' work lasts longer, they invoke norms more often, and their activity starts intensely, tails off and stays high. TeBlunthuis, Shaw and Hill (2018) retested the findings on hundreds of active peer production wikis from Wikia and reproduced the subset of models they could test. They reported support for the external validity of two claims: quality control can limit growth by deterring contributors, and norms entrench over time. The result is bounded: two claims, other wikis, observational models rather than an intervention.

The largest randomised test of a receiving measure is the Newcomer Homepage. Warncke-Wang and colleagues (2023) assigned its invitation across 27 Wikipedia language communities and 244,060 accounts. It raised the odds of a first edit within a day and changed nothing in retention, productivity or the reversion of contributions, helping newcomers who had not intended to contribute and distracting those already contributing. A newcomer working in a second language is the ordinary case rather than the exception.

Choi, Alexander, Kraut and Levine (2010) identified the seven socialisation tactics used most often in Wikipedia WikiProjects and measured them against newcomers' later commitment. Most newcomers edited less over time; declines slowed or reversed for those socialised with welcome messages, assistance and constructive criticism, while invitations predicted steeper declines. Lampe and Johnston (2005) reached a compatible result from Slashdot, where new users' behaviour was affected by viewing behaviour, by the moderation feedback their comments received and by replies.

Contribution turns on cost and motive, and growth can destroy the community

Cosley, Frankowski, Terveen and Riedl (2007) deployed SuggestBot in Wikipedia, routing editing tasks to contributors, and raised edits by roughly four times compared with suggesting random articles. What changed was not the desire to contribute but the difficulty of finding work, which is why a community can be healthy in spirit and short of labour.

Nov (2007) surveyed Wikipedia contributors and found no single motive: fun, at 6.10 of 7, and ideology, at 5.59, ranked highest of eight, while social, career and protective motives were weak. Ideology was highly rated but not significantly correlated with how much a contributor contributed, so stated motive poorly predicts effort. Crowston and Fagnot (2018) turned that into a stage theory, distinguishing initial, sustained and meta motivations and testing them on a Wikimedia editor survey: distinct stages carry distinct motives, so one lever does different work at different points in a contributor's life. Ren, Kraut and Kiesler (2007) add the group-level parallel, since common identity theory predicts attachment to the group and common bond theory attachment to individual members.

Butler (2001) built a resource-based model of online social structures and tested it on longitudinal data from a random sample of email-based listservs. Size and communication activity each have both positive and negative effects on survival. The benefits a community returns fall as size outruns the communication that supplies them, so growth can undermine the sustainability it appears to demonstrate.

Contested question: can the pipeline be redesigned, or is the newcomer's trajectory set on arrival?

The newcomer research divides into three positions. The first, held by Panciera and colleagues (2009), is that the trajectory is set almost immediately, since eventual heavy contributors differ from other newcomers from their first session. The second, held by Choi and colleagues (2010) and Lampe and Johnston (2005), is that the receiving community writes the slope, because welcome messages, assistance and constructive criticism slow an otherwise universal decline. Antin and Cheshire (2010) hold the third: the pipeline runs further upstream, since reading is not simple free riding and readers supply value, and on Lave and Wenger's (1991) account of legitimate peripheral participation reading is the gateway through which newcomers learn the practice before editing.

Kraut and Resnick (2012) sit on the second position while conceding the first. What would settle the question is a design that does both at once. Recruit comparable cohorts into several communities, and assign the receiving tactics at random. Follow them with the cohort studies' own trace measures, so that selection and treatment are separated and the outcome is behaviour, not intention. TeBlunthuis and colleagues (2018) supply the population such a trial would need; they do not supply the randomisation.

Compare the three positions by method rather than by conclusion. Panciera and colleagues (2009) compare cohorts after the fact; Choi and colleagues (2010) and Lampe and Johnston (2005) observe tactics the receiving community chose rather than assigning them; Antin and Cheshire (2010) rest on self-report. None of the three separates who arrives from what the community does with them. Warncke-Wang and colleagues (2023) assigned a receiving measure across wikis and found it moved activation rather than survival. While the dispute is open, a designer should measure a newcomer's first session and the community's response as two separate variables, as the homepage trial does.

Where the argument is heading

Two movements are visible. One is from single-project history to populations: TeBlunthuis and colleagues (2018) show the questions asked of Wikipedia can be asked of hundreds of wikis. The other is a move from a single motive to a sequence. Crowston and Fagnot (2018) argue that initial, sustained and meta motives differ, so a newcomer intervention should be judged at the stage it targets rather than on aggregate edit counts. The instruments exist: Cosley and colleagues (2007) showed routing work to contributors can be measured in a live community, and Antin and Cheshire (2010) point the measurement upstream, at readers.

The sources this unit rests on

Each note states what the document did and found. Identifiers were resolved on 2 November 2026.

Antin, J. and Cheshire, C. (2010). Readers are not free-riders. doi:10.1145/1718918.1718942 Surveyed Wikipedia readers and contributors and found that reading is not simple free riding and, on the legitimate peripheral participation account, that reading is a gateway into the community. It moves the newcomer pipeline upstream of the first edit.
Butler, B. S. (2001). Membership size, communication activity, and sustainability: a resource-based model of online social structures. doi:10.1287/isre.12.4.346.9703 Built a resource-based model of online social structures and tested it on longitudinal data from a random sample of email-based listservs, finding that size and communication activity have both positive and negative effects on sustainability. It supplies the mechanism by which growth itself can destroy a community.
Choi, B., Alexander, K., Kraut, R. E. and Levine, J. M. (2010). Socialization tactics in Wikipedia and their effects. doi:10.1145/1718918.1718940 Identified the seven socialisation tactics used most often in Wikipedia WikiProjects and measured their effect on newcomers' commitment: welcome messages, assistance and constructive criticism slowed or reversed the general decline in edits, while invitations led to steeper declines. Receiving practices, not recruitment appeals, move the slope.
Cosley, D., Frankowski, D., Terveen, L. and Riedl, J. (2007). SuggestBot: using intelligent task routing to help people find work in Wikipedia. doi:10.1145/1216295.1216309 Deployed a recommender in Wikipedia that routed editing tasks to contributors and found that intelligent routing raised edits by roughly four times compared with suggesting random articles. The binding constraint is the cost of finding work, not the wish to contribute.
Crowston, K. and Fagnot, I. (2018). Stages of motivation for contributing user-generated content: a theory and empirical test. doi:10.1016/j.ijhcs.2017.08.005 Distinguished initial, sustained and meta motivations for contributing as three interrelated phenomena and tested the stage theory on a Wikimedia editor survey, finding support for the hypothesis that stages have distinct motives. It explains why one lever affects newcomers and veterans differently.
Halfaker, A., Geiger, R. S., Morgan, J. T. and Riedl, J. (2013). The rise and decline of an open collaboration system. doi:10.1177/0002764212469365 Analysed Wikipedia's editing history and showed that the community's quality control response to growth cut newcomer retention: reversion of desirable newcomers' first-session edits rose from 6.1 to 18.2 per cent between 2006 and 2007, two-month survival stood at 25.6 per cent in early 2006, and the survival rate fell to 11.7 per cent within a year. It is the paper the pipeline thesis rests on.
Kollock, P. and Smith, M. (1996). Managing the virtual commons: cooperation and conflict in computer communities. doi:10.1075/pbns.39.10kol Argues that computer communities face a public-goods dilemma whose default is free riding. It is the primary for the free-rider prediction the unit's synthesis takes up.
Kraut, R. E. and Resnick, P. (2012). Building successful online communities: evidence-based social design. https://mitpress.mit.edu/9780262528917/building-successful-online-communities/ Organised online community design around five challenges, stating the evidence behind each design claim, and framed contribution as a public goods problem and commitment through interpersonal bonds. It is the synthesis the unit argues with.
Lampe, C. A. and Johnston, E. (2005). Follow the (slash) dot: effects of feedback on new members in an online community. doi:10.1145/1099203.1099206 Compared three explanations for how new users learn to participate in Slashdot and found that behaviour was affected by viewing behaviour, by the moderation feedback their comments received and by replies to them. Feedback from members, not instruction, is what moves a newcomer.
Lave, J. and Wenger, E. (1991). Situated learning: legitimate peripheral participation. https://doi.org/10.1017/CBO9780511815355 Argued that learning is legitimate peripheral participation in a community of practice, in which newcomers move towards full participation through real work, making the community rather than the individual the unit of learning. It frames the unit's reading of lurking as part of the pipeline.
Nov, O. (2007). What motivates Wikipedians? doi:10.1145/1297797.1297798 Surveyed Wikipedia contributors, drawing 151 valid responses from a sample of 370, and found fun (6.10 of 7) and ideology (5.59) the strongest of eight motivations; ideology was not significantly correlated with contribution level. Stated motive, it shows, poorly predicts effort.
Panciera, K. A., Halfaker, A. and Terveen, L. (2009). Wikipedians are born, not made: a study of power editors on Wikipedia. doi:10.1145/1531674.1531682 Compared editors who became heavy contributors with those who did not and found the difference visible from the first day: eventual Wikipedians' work lasts longer, they invoke norms more, and their activity starts intensely and stays high. It is the first position in the unit's contested question.
Ren, Y., Kraut, R. E. and Kiesler, S. B. (2007). Applying common identity and bond theory to design of online communities. doi:10.1177/0170840607076007 Linked common identity theory, attachment to the group, and common bond theory, attachment to individual members, to design decisions such as recruiting newcomers against retaining members, showing the choices lead to different forms of attachment and participation. It separates two commitment mechanisms that design advice conflates.
TeBlunthuis, N., Shaw, A. and Hill, B. M. (2018). Revisiting "The rise and decline" in a population of peer production projects. doi:10.1145/3173574.3173929 Replicated Halfaker and colleagues' tests on hundreds of active peer production wikis from Wikia, reproduced the subset of findings they could test, and reported support for the external validity of the claims that quality control can limit growth by deterring contributors and that norms entrench over time. It lets the central claim be tested beyond one project's history.
Faraj, S. A., Jarvenpaa, S. L. and Majchrzak, A. (2011). Knowledge collaboration in online communities. doi:10.1287/orsc.1100.0614 Argues that what affords collaboration in an online community is its fluidity, since boundaries, norms, participants and artefacts keep changing and each resource a member brings carries both a benefit and a cost. It explains why a founding design does not decide a community's fate.
Warncke-Wang, M., Ho, R., Miller, M. and Johnson, I. (2023). Increasing participation in peer production communities with the newcomer homepage. doi:10.1145/3610071 Ran a controlled experiment on a Newcomer Homepage across 27 Wikipedia wikis and 244,060 accounts, finding a small activation gain with no change in retention or productivity.

Check your understanding & apply it · 4 questions, 1 essay

1A community's coordinators report that monthly active contributors have flattened while the backlog of unwritten pages has grown, and propose to protect quality by reverting weak new contributions faster. Which diagnosis does this unit's evidence support?

2A volunteer mapping project plans to invite thousands of newcomers at once, to protect quality by reverting weak contributions faster, and to send every newcomer the same automated welcome. Which prediction does this unit's evidence support?

3Two claims are made about a community that is growing fast. The first says growth shows the community is healthy. The second says growth can raise the cost of coordinating and undermine the community it appears to strengthen. Which claim does the unit's evidence support?

4A researcher proposes to settle whether a newcomer's trajectory is set on arrival or written by the receiving community by comparing, after the fact, the first sessions of editors who became heavy contributors with those who did not. Judged against this unit, how strong is that design?

5Essay An open-source documentation project has grown quickly. The number of monthly active contributors has flattened while the backlog of unwritten pages has grown, and the senior maintainers have responded by reverting weak new contributions faster and more strictly in order to protect quality. Advise the project using this unit's argument. Say what is happening to the project's pipeline and why the flattening is the symptom the unit's sources would predict, what the evidence says about the maintainers' response, which receiving practices the evidence supports instead, what the marginal cost of a contribution implies for the project's tooling and guidance, how the project should treat motivation, and how it should test whether the changes work. Name your sources at each step.

The project is treating a pipeline problem as a quality problem. Halfaker and colleagues (2013) describe this configuration exactly: as Wikipedia grew, its community used its primary quality control mechanism more restrictively, the share of desirable newcomers whose first-session edit was reverted rose from 6.1 to 18.2 per cent between 2006 and 2007, two-month survival of those newcomers had stood at 25.6 per cent in early 2006, the survival rate fell to 11.7 per cent and did not recover, and active editors declined from a peak of 56,400 in March 2007. Stricter first-edit rejection is therefore likely to deepen the flattening the maintainers are trying to reverse. Butler (2001) adds that growth raises cost when the resources for communication do not scale, so more members can make the project less sustainable rather than more. Choi and colleagues (2010) give the alternative: welcome messages, offers of assistance and constructive criticism slowed or reversed the general decline in newcomers' edits, while invitations led to steeper declines, so the maintainers should replace the automated welcome with personal assistance and useful criticism. Lampe and Johnston (2005) show that feedback on a new member's first contributions, not instruction, changes what they do next. On cost, Cosley and colleagues (2007) raised edits by roughly four times by routing tasks to contributors instead of suggesting random pages, so the project should route the backlog to people who can do it and cut the time spent finding work worth doing. Motivation should not be treated as one lever: Nov (2007) found fun and ideology the strongest of eight motivations but no significant link between stated ideology and contribution, and Crowston and Fagnot (2018) found initial, sustained and meta motives differ, so onboarding and retention need different appeals. The contested question is whether the newcomer's trajectory is set on arrival (Panciera et al., 2009) or written by the receiving community (Choi et al., 2010; Lampe and Johnston, 2005), with Antin and Cheshire (2010) placing the pipeline upstream of the first edit. The project cannot settle that with its own logs alone, but it can test its changes: define newcomer cohorts, assign the receiving practices at random where the workflow allows, and follow each cohort with trace measures, time to first accepted edit, reversion of first-session edits, two-month survival and return after criticism, so that selection and treatment are separated.