It has frequently been observed that when an online forum is set up, such as for sharing knowledge or mutual technical support, a large proportion of the contributions come from only a small number of the signed-up participants, and a large number don’t contribute much, if at all.
In his 2008 book ‘Here Comes Everybody’ Clay Shirky gives examples from Wikipedia and Flickr. He displays a graph of how 118 photographers contributed over 3,000 photos of a Coney Island parade to a Flickr page, of whom 12 of them contributed over half; one photographer contributed 238 photos all by herself. Shirky refers to this as a ‘power law distribution’ curve and considers this to be quite typical of how populations engage in online communities. (To dig deeper into the maths of power law distributions, see the Wikipedia article at http://en.wikipedia.org/wiki/Power_law.)
If anything, the skew in participation in online discussion communities is more extreme, because many of these contain a large proportion of ‘lurkers’ – people who join a message board, email list, newsgroup or what have you, and read all the messages, but never contribute at all. In ‘Lurker demographics: Counting the silent’, a research paper for the Computer-Human Interface conference of the ACM in 2000, Blair Nonnecke and Jenny Preece suggested [1] that ‘lurkers’ may make up as much as 90% of most online groups. Jakob Nielsen promotes the idea of a ‘90–9–1 rule’ [2] signifying that lurkers make up 90% of such groups and 9% contribute intermittently, while 1% are the heavy contributors from who more than half the postings come.
For people who sponsor or manage online knowledge-sharing systems, this differential in contribution may be something to worry about. After all, it seems reasonable to suspect that many ‘lurkers’ actually have valuable ideas and experience to contribute, but fail to do so for lack of confidence or for some other reason, such as the attitudes displayed by other users. What can be done to encourage the quieter folk to share what they know?
From stagefright to encouraged participation
In any community of practice, it is natural that some people will have more experience and more words of wisdom to contribute than others; we expect (nay, we hope) that these people will contribute disproportionately more. Others will join to listen and to learn. A third form of participation is by asking questions, and if a well-formed question is responded to with one or more considered answers, the whole community benefits from the knowledge shared.
Independently of this, there is the personality factor. It is clear that some people are simply more confident than others in launching a message into public arena, for a variety of reasons to do with their family background, life experience, command of language or whatever. And in any new setting, even a confident person may take time to speak up, though it be only to ask a question.
When people are averse to speaking out in public, we call it ‘stagefright’. In a face-to-face setting, such as a workshop, we can diminish this barrier by splitting a large group into smaller groups. For most people, speaking up in a small group feels less risky, and a good workshop leader will ensure some icebreaking activity precedes the discussion to increase trust. (Face-to-face in a small group, there is actually also some pressure in favour of contributing: if two out of seven around a table remain silent, it will feel very odd.)
‘Stagefright’ operates in online settings too. Nonnecke and Preece found that the proportion of lurkers rises in larger, more anonymous online communities. The difference here is the cloak of invisibility. If you are totally invisible to others unless you speak, which is usual online, lurking is a comfortable, risk-free strategy — and it takes more energy to move people on from there.
Don’t bite the newbies!
We probably do not put enough effort into devising ways of inducting people into online communities so that they feel comfortable and confident in adding what they know and asking about what they don’t. It would be useful to know whether online systems where people establish a profile and add a photo are less scary than more anonymous forums. It certainly sounds reasonable that having a known identity and a reputation to care for should at least dissuade the kind of verbal ‘terrorism’ that some online denizens inflict on their fellows.
Sadly, in many online knowledge-sharing communities, especially those with a technical support or knowledge curation function, there has been a long history of ‘newbies’ being savaged verbally, or subjected to dismissive sarcasm. Sometimes this is understandable: long-term community members can get sick and tired of seeing the same basic ‘dumb’ questions raised again and again by newcomers. This of course is why ‘Frequently Asked Questions’ lists were first devised, and it is not unreasonable in such circumstances to direct new members to the FAQ as a first resort.
Foreground the knowledge, acknowledge the contribution
Where a community of practice, or a knowledge community, functions primarily through an email list or online forum such as those operated by LinkedIn, there is always a risk of previous expressions of knowledge getting lost, of repetition of topics, and of such discouraging attacks on newbies. But if the online system puts the knowledge in the foreground, and keeps it permanently accessible, so that discussion happen around the knowledgebase, we should expect those problems at least to go away. However, this still leave the matter of how to encourage contributions of knowledge.
There is a well-known set of stories in knowledge management circles about how Xerox harnessed the knowledge of its service engineers, the guys who make visits to sick photocopiers and laser printers on customer sites. Julian Orr, an anthropologist at Xerox’s Palo Alto Research Center, undertook a study of these service technicians, their work, their values and their discourse. His findings are recorded in a 1996 book, ‘Talking About Machines: an Ethnography of a Modern Job.’
Orr found that Xerox service technicians didn’t think like company men. They exhibited a ‘gunslinger mystique of self-reliance: the lone technician walks into the customer site to cope with whatever troubles lie therein… but with the community available as a resource.’ He also noted that they couldn’t be motivated by prospects of promotion because ‘the only real career option, promotion to management, means leaving the community [of competent technicians], and most technicians would rather remain a technician-hero than become an organization manager.’ (See excerpts.) [3]
However, their community of local fellow technicians was important to them; they got together frequently for breakfast, lunch and after work. In amongst the socialising and the card games they talked constantly about photocopiers and printers, why the darned things didn’t behave like they were described in the repair manuals, and what to do about them. They acted as a compact knowledge community — albeit a frail one, with nothing written down, and knowledge lost as people moved out of town.
To Xerox’s credit, the company acted on Orr’s observations, initially by handing out walkie-talkies so the techs could stay in touch through the day, and later by setting up a Web-based knowledge database called Eureka, with Oracle at the back end and Xerox’s own DocuShare Web technology as the front end. Engineers could now write up their newly found solutions into Eureka templates, using their laptop computers; and they could access the database and explore it while in the field, to search for solutions to problems they were encountering.
Eureka wasn’t an overnight success. Xerox engineers initially lacked an incentive to contribute online: they saw adding their knowledge to the database as an extra burden on top of their existing work. With a bit more research, Xerox found that the key was to allow engineers to own the authorship of their solutions, to gain recognition for their knowledge, as they had in the face-to-face local communities down at the diner. Furthermore, they added a peer review process, so a solution would only be added officially to the database after approval by other skilled Xerox technicians: more kudos.
Xerox’s director of corporate strategies, Dan Holtshouse, reported in 2001 that within a few years, the Eureka database had accumulated some 50,000 solutions, and that between 70% and 80% of technicians were contributing to it, on average once a week. [4] Part of the secret, of course, was the almost perfect fit between that technician cult of personal competence, and the online peer review and acknowledgement process that rewarded their contributions (backed up, it should be said, with rewards ceremonies and other accolades for top tipsters).
Stacking up knowledge
Do knowledge communities function best when someone acts as the facilitator and moderator? It is an idea I have often heard expressed, and I myself have often seen discussion lists function better because someone has stepped informally into this ‘pastoral’ role.
However, I was intrigued earlier this year when a young programmer described to me a knowledge community that he belongs to, called Stack Overflow, which uses an ingenious set of underlying rules and a game-like points-scoring systems to ‘crowdsource’ its management and, with a light touch, nudge people towards contributing well.
Stack Overflow is essentially a ‘question and answer’ site for computer programmers. A user formulates a question — the most popular ones are on such subjects as C#, Java, JavaScript, jQuery and PHP — and other users attempt to give helpful answers. The site started in 2008 and currently has over three quarters of a million members, with nearly two million registered questions. Unlike an email discussion list, the Stack Overflow Web site retains the questions and answers, and so it has the nature of a huge FAQ, with wiki-like features.
In a ‘tech talk’ to a Google audience in April 2009, for which there is a YouTube video [5], Stack Overflow co-founder Joel Spolsky explained that the site was designed with the understanding that people’s behaviour is guided by the nature of the environment they find themselves in. Among the mechanisms that the Stack Overlow developers implemented were voting, tags, and post-editing of both questions and answers; plus the accumulation of reputation through ‘points’ and ‘badges’ as a reward for helpful behaviour.
30,000 tags organise 2 million Q&As
They have found that their user community is generally quite good at assigning relevant descriptive tags to the problem they are describing. Only high-privilege users can create new tags, but over time a library of over 30,000 has built up, and you shoul apply as many as are relevant. Each tag has an ‘info’ page, which combines the attributes of a concise introductory description, and what in a thesaurus would be called a ‘scope note’, an explanation of when it is appropriate to use that tag. (All these tag ‘info’ pages combined are known as the system’s ‘tag wiki’.)
The site makes it very easy to browse Q&A sets by tag, also to view which ones are actively looking for answers, which ones are most popular, etc. If you register as a Stack Overflow member, you get to customise your personal interface to the site, selecting the tags in which you are most interested. Regular visitors will thus be presented on arrival with stuff that interests them. The site also usies the tags to track what kind of questions you have answered in the past and will alert you to new questions in that field which you might want to answer.
Wiki-like editing improves the quality
Interestingly, unlike the Xerox Eureka approach which gave ownership of knowledge to the contributor, Stack Overflow permits senior members who have earned enough privileges (see below) to edit both the questions and the answers, and to keep editing them. Spolsky remarks that this was directly inspired by Wikipedia, with the aim that the quality of question-and-answer sets can improve over time.
Voting sifts content, assigns reputation
A problem with many Q&A-based knowledge sites is that they do not sift good answers from bad ones. In Stack Overflow, members can vote answers up or (more rarely) down, so the best answers rapidly rise to the top. Questions can also be voted up and down, so that the most valued ones have highest visibility. Spolsky admits that this was an idea that they borrowed from the social bookmarking services Reddit and Digg.
Motivation in Stack Overflow is based on the awarding of points, which are an index of your reputation. You accumulate points when fellow users ‘vote up’ your questions and your answers. The site FAQ explains that reputation acquired this way is a measure of how much the community trusts you, admires the relevance of your questions and answers, and approves your ability to communicate.
The acquisition of reputation points progressively grants you more and more privileges on Stack Overflow. Once you have acquired ten points, you are free from the basic precautionary ‘new user’ restrictions. With 15, you can starting voting posts up; interestingly, you must have 125 before you can start voting posts down. The coining of new tags is a considerable privilege and requires 1,500 points, while 2,000 are required to be allowed to edit existing questions and answers — the wiki-like behaviour described above.
Stack Overflow is thus a meritocracy. Those at the top of the stack are in practical terms indistinguishable from moderators, able to protect Q&As from further editing, delete the lowest-scoring posts and edit the ‘tag wiki’ documentation. But note that these privileges are not handed out by the site’s founders: they are earned in the coin of gratitude from the other thousands of users who find the contributions valuable.
In addition to points, Stack Overflow members can achieve ‘badges’, often with humorous names, which signify particular kinds of achievement such as ‘Enthusiast’ for visiting the site on 30 consecurive days or ‘Pundit’ for leaving 10 comments each awarded 5 pointss or more. Your badges will be visible everywhere you post on the site!
Stack Exchange Network
The success of the Stack Overflow model encouraged the company behind it, Fog Creek Software, to make the same engine available to drive other similar knowledge communities. The Stack Exchange Network currently hosts 57 such communities fully operational or in beta, covering topics such as WordPress, Game Development, Database Admin and also less geeky subjects such as Cooking, Writing, Home Brewing and Bicycles. The current list of sites is at http://stackexchange.com/sites. The most popular of these, with 20,000 users, is AskUbuntu, a user and developer support site for the Ubuntu operating system.
(BTW another Q&A site which makes use of tags, voting on answers and wiki-like editing — but with a lesser degree of sophistication and aimed towards a more general-interest audience — is Quora.com.)
Moral: model the user, find the motivation
Why have Xerox’s Eureka database and the various knowledge communities in the Stack Exchange Network succeeded in encouraging knowledge contribution? Because the forms of user participation they offer, and the rewards for doing so, have a good fit with the nature of their users.
As Spolsky remarked in his introduction to the Google Tech Talk, while the User Interface problem in computing has many satisfactory solutions, we are still struggling to engineer effective Human to Human Interfaces that are mediated through virtual space. Solving the problem of improved participation and contribution has to start with understanding the nature of the people you want to recruit to your knowledge community, and ‘what makes them tick’.
References
[1] Blair Nonnecke, Jenny Preece. ‘Lurker demographics: Counting the silent’. Proceedings of CHI 2000. The Hague: ACM. http://citeseer.ist.psu.edu/viewdoc/download;jsessionid=7C67E954D971D9F108D86A35047AC9E7?doi=10.1.1.83.9346&rep=rep1&type=pdf
[2] Jakob Nielsen’s Alertbox, 9 Oct 2006: ‘Participation Inequality: Encouraging More Users to Contribute.’ http://www.useit.com/alertbox/participation_inequality.html
[3] Julian Orr, ‘Talking About Machines: An Ethnography of a Modern Job’; published by Cornell University Press, 1996. Excerpts here: http://roi-learning.com/dvm/cop/orr.htm
[4] ‘How Xerox got its engineers to use a knowledge management system’ — Judith N. Mottl. 2001. In TechRepublic online, http://www.techrepublic.com/article/how-xerox-got-its-engineers-to-use-a-knowledge-management-system/1048267
[5] Joel Spolsky. Google Tech talk: Learning from StackOverflow.com. YouTube video at http://www.youtube.com/watch?v=NWHfY_lvKIQ
Be First to Comment