Discursive Equality and Everyday Talk Online: The Impact of “Superparticipants”


  • Accepted by previous editor Maria Bakardjieva


Empirical studies of online debate almost universally observe a “dominant” minority of posters. Informed by theories of deliberative democracy, these are typically framed negatively—yet research into their impact on debate is scant. To address this, a typology of what we call super-participation (super-posters, agenda-setters and facilitators) is developed and applied to the http://www.moneysavingexpert.com/ forum. Focusing on the first of these, we found 2,052 superposters (0.4%) contributing 47% of 25m+ posts. While superposters were quantitatively dominant, qualitative content analysis of the discursive practices of 25 superposters (n=40,044) found that most did not attempt to stop other users from posting (curbing) or attack them (flaming). In fact, in contradiction to the received wisdom, super-posters discursively performed a range of positive roles.


Debate about the equality (or not) of online talk persists. Empirical studies of, for example, online political talk typically observe a highly active, “dominant” minority of posters. Davis (2005, pp. 58–59) found dominant minorities in Usenet forums, though he did not analyse what impact they had on the nature of the debate (see also Kuperman, 2006). Focusing on the Jihadist forum, Awan (2007) found that 87% of registered members had never posted; 13% had posted at least once; 5% had made more than 50 posts and 1% had made more than 500 posts. Awan (2007, p. 394) states that: “A small but vocal, active core posts new content, initiates debates and responds to questions posed by newcomers.” However, it is not clear on what empirical basis these observations were made. Wright (2006) identified an active minority in the UK government's Downing Street online discussion forum but also did not explore their impact (see also Jankowski & Van Os, 2004). Moving away from discussion forums, Anstead and O'Loughlin (2011) identified a small minority of active participants on Twitter debates surrounding a political discussion show on television, but they also did not explore this finding further. Shifting to the blogosphere, studies have identified significant disparities in the link structure and volume of postings (Hindman, 2009), but analysis of the discursive practices of this blogging elite remains scant.

When studies have looked in more depth at the behaviour of highly active participants, a positive function has been suggested. Albrecht's (2006, p. 70) analysis of the Demos debate in the city of Hamburg found that 20% of participants created 75% of the posts. However, they acted as “old hands,” largely replying to others. Similarly, Kies' (2010) study of the large (over 500,000 message) Radicali political party discussion forum in Italy also suggests a more positive role. The 10 most frequent posters created over 25% of all the posts (averaging over 5,000 messages each). However, he concludes that the forum was an “exemplary” case and that although “the debates are dominated by a minority of users […] this apparent domination should not overshadow the participative and deliberative importance of the forum…” (2010, pp. 141–142). The existence of a highly active minority would appear, thus, to be the norm and is an extension of the 1/9/90 rule, which predicts that only 1% of users of a forum actually post, with 90% lurking and 9% editing.1 However, the empirical research lacks depth, and there are differences of opinion with regard to their actual impact on debate.

Much of the literature that has analysed such communication has been framed by theories of deliberative democracy and the public sphere. The work of Habermas, and particularly his ideal speech situation and discourse ethics, are especially influential. Habermas effectively argues that anyone who has the competence to speak should be allowed to take part in the discourse if they wish to do so (Chambers 1996). Put simply, communication must be egalitarian in nature. Building on this, Dahlberg (2001a,b) has emphasised the importance of discursive equality and inclusion to the online public sphere. There are, however, differences of opinion (or different readings of Habermas's thinking) about the extent to which communication in the public sphere must be based on equality and egalitarianism—and what these concepts mean in practice.2

The more widely held view is that there must be equality of access to (i.e. the chance to participate in) the public sphere and that each individual's view must be treated with respect. Following this logic, the volumes of participation of specific individuals are not problematic so long as they do not attempt to stop others from participating. In fact, the regularly ephemeral and disjointed nature of online debate is often said to make the presence of an active minority a distinct positive (Dahlberg 2001a). Similarly, Oldenburg (1999) argues that “the regulars” are crucial to the construction of what he calls third places: communal venues beyond home or work (such as pubs and cafes) where informal political talk emerges. However, it must be questioned: is there a point at which an active minority effectively puts off potential, lurking or infrequent participants from engaging in debate—even though they are not discursively attempting to block them by what they say (see also Herring, 1993)? If this were the case, quantitative analysis of participation rates would appear to be important. Indeed, it seems counterintuitive to ignore the volume of an individual's postings if we want to assess their impact on discursive equality because this is likely linked to the construction of authority (Huffaker, 2010). This leads us to the second position.

The more literal approach argues that there must be discursive equality—equality of participation. For example, Cohen (1997, p. 69—see also Knight & Johnson, 1997) argues that there must be “some form of manifest equality among citizen.” Dahlberg (2001a) focuses more specifically on the everyday public sphere: “More difficult to deal with, and thus more problematic for achieving egalitarian and inclusive discourse, is the qualitative domination of conversations by particular individuals and groups. Here, dominant participants may not necessarily be directly abusive or say more; rather, they assert their influence and sideline other participant's views by dictating the agenda and style of dialogue.” Recourse is again made to the perceived nature of online communication, but this time to justify the use of moderators and posting limits to facilitate equality. For Coleman and Gotze (2001, pp. 17–18): “mechanisms of moderation and mediation are crucial to the success of […] asynchronous dialogue”. To this end, the famous Minnesota E-democracy forum limited users to two posts per day in an attempt to stop quantitative domination (Dahlberg, 2001b).

There are clearly differences of opinion and uncertainties within the literature—and practice—compounded by a distinct lack of empirical research. This is unsurprising because there is a lack of agreement—and often a lack of specificity to theoretical models—as to what constitutes an active (or dominant) minority, and this makes it difficult to observe and interpret such behaviour in practice. To help move these debates forward and guide future research, we develop a three-part typology of what we call superparticipation: superposters, agenda-setters and facilitators.3 Our empirical analysis focuses on superparticipants. We argue that to fully understand the relative equality of Internet debate, and how the active minority impacts this, researchers must analyse posting patterns and discursive practices. To this end, we combine an overarching quantitative content analysis with a detailed qualitative content analysis of superposters.

The analysis focuses on the www.moneysavingexpert.com (MSE) discussion forum. The selection was informed by a desire to respond to a new agenda for online deliberation research, namely the study of informal political talk in nonpolitical online third spaces (Wright, 2012a,b). The concept of third space is heavily influenced by, but ultimately grounded in a critique of, Oldenburg's (1999) notion of the third place: physical environments beyond home and work, such as pubs and cafes, where communities gather and political talk emerges. Rather than thinking about what virtual equivalents of a third place might look like, third space reconsiders the concept in the context of the Internet and thus differs in several of its conclusions. In particular, third space does not privilege place over issue-based communities, and suggests that online forums facilitate rather than debilitate informal political talk. While testing whether the MSE forum was a third space is not the main focus of this research, the research design is informed by our desire to respond to the new agenda and we will, thus, also make tentative suggestions about the MSE forum as a third space.

A Typology of Superparticipation

In the literature on user behaviour within online forums, there are myriad attempts to develop typologies (Golder, 2003; Turner & Fischer, 2006) and study particular categories of behaviour/participation. These include trolls, experts (Welser et al., 2007), information seekers, opinion leaders, and repliers (Viegas & Smith, 2004). It is striking that even the most detailed typologies (Golder, 2003) give little (if any) weight to the volume of participation. Kim (2000, p. 118) and Sonnenbichler's (2010) related Membership Life Cycle models do begin to move us in this direction. As the name suggests, these models are based on the length of membership and how comfortable people feel when participating. Kim's more linear approach covers visitors, novices, regulars, leaders and elders, while Sonnenbichler's more fluid model has categories for visitors, novices, actives, passives, leaders and trolls. While these models are helpful, they do not provide posting volumes for each category and the regular/active pool is too broad. More nuance is needed when considering active participants, and to this end, we put forward the concept of “superparticipants”.

At the most basic level, to describe something as “super” implies going above and beyond the norm. In terms of participation in a discussion forum, there are two principal forms of posting activity: users can start new threads or post within existing ones. Bringing this together, we identify two types of superparticipants: Superposters and Agenda-setters. The third category, facilitators, refers to the people who manage the day-to-day activities on the forum. The three categories are not mutually exclusive.

SP1: Superposters

Superposters (SP1s) are those participants who post very frequently in a discussion forum. There are a number of different ways to approach and define what counts as a superposter. The issue is made more complicated because we can assume that the volume of individual user postings is linked to the age and size of a forum: the bigger and older the forum is, the greater the number of SP1s. If this assumption is correct, it suggests that we should calculate SP1s by determining either a small percentage, or a set number, of the most frequent posters. There are, however, some concerns with this approach. First, the level of individual user postings can vary significantly (and change over time), and this would be reflected in any sample. There is a danger that the most frequent participants on some forums would actually have made only a limited number of posts. To consider a participant as super, we argue that they must have created a minimum number of threads or made more than a certain number of posts. Thus, to resolve this issue, a minimum number of posts must be adopted—but this leads to our second concern—which is with larger forums.

Discussion forums vary in scale, from a few hundred posts to nearly two billion: There are hundreds of forums with over 10,000,000 posts.4 On large discussion forums with millions of posts, it is highly unlikely that one person will have created even 1% of all posts, and adapting this approach, thus, would be complicated. It would also mean that users who may have made tens of thousands of posts would be excluded—and this gets to the heart of our concern, which is not (necessarily) to analyse the dominant minority, but to analyse all users who participate regularly.

To account for variations in forum size, we argue that a super-poster should be defined as any user that has created more than 2% of all messages on a forum with between 20,000 and 99,9999 posts and any user who has made more than 2,000 posts on a forum with over 100,000 thousand messages. This gives a minimum post count of 400 on a forum with 20,000 posts. We set the level at 2,000 posts because any user making this many posts must have spent a significant amount of time and effort writing posts. This approach is informed by our inclusive rather than exclusive approach; we want to find all participants that qualify as super-participants rather than an elite minority: In theory, every user on a forum could qualify as a superparticipant. But how does this marry with the notion that “super” implies going beyond the norm? First, it is highly unlikely to actually happen. Second, existing analyses of online discussion forums suggest that in most cases, participation at this level is atypical, and this reflects our broader approach to consider the bigger picture rather than a specific forum. The second type of Superparticipant (SP2) attempts to set the agenda of forums.

SP2: Agenda-setters

The presence of people who attempt to set the agenda of online forums has, again, been widely noted, but with relatively limited attempt to focus on their impact in depth. According to Bua (2012), the importance of agenda setting is often not fully appreciated in the literature on political communication and participation—with limited theoretical and empirical work to date. In the context of Usenet discussion forums, Himelboim et al. (2009) have undertaken a detailed study of new thread creation. They found that many Usenet forums featured a small group of ‘discussion catalysts’ whose seed posts inspired a lot of debate. Interestingly, these users largely posted stories from elsewhere on the web with little or no comment—a relatively conservative approach to setting the agenda. Himelboim et al. do not make clear whether the ‘catalysts’ focused on specific issues. Moreover, they did not analyse how, if at all, these users engaged in the subsequent debate.

We have categorised agenda-setters as superparticipants because they can have a disproportionate influence on the nature of debate that occurs. The inclusion of agenda-setting reflects our view that influence is not limited to the volume of posts alone. Agenda-setters may post on similar topics, or within specific sections of forums. Of course, creating new seed posts does not guarantee that they will receive responses, or that people will interpret—uptake—the message as intended. We might expect regular participants to be more familiar with the (discursive and other) norms that shape interactions on the forum (Huffaker, 2010). However, the extent to which SP2s successfully set the agenda and secure uptake (Freadman 1994) remains an important point for analysis. Based on our previous experience of analyzing thread creation, and findings in broader studies, we have set the bar for SP2s at 200 threads.

SP3: Facilitators

Facilitators perform a specific, formal function within discussion forums because they help to set the tone and can normally moderate, manage or otherwise advise broader participants. The precise role of facilitators is determined by the forum—as is the selection process for who becomes a facilitator. Wright (2006) has previously outlined a series of potential roles, distinguishing between moderators and facilitators. This can involve the management of SP1s and SP2s.5 The importance of facilitators to political (Edwards, 2002) and other talk (Herring et al., 2002) is widely recognised. Having outlined the concept of superparticipation, we now present the methodological approach and case study adopted.

Research Design and Methodology

Due to space limitations the analysis of superparticipation focuses on the first of these—SP1s—using both quantitative and qualitative measures. Our typology makes no judgment on whether the impact of SPs on the nature of debate is positive or negative. This must be determined through empirical analysis. Thus, in order to analyse how SP1s communicate, we have undertaken both a broad quantitative content analysis and a detailed qualitative content analysis. As discussed in the introduction, our selection of the case and coding frame is informed by our earlier work that has argued there is a need for more analysis of informal political talk in non-political online forums (third spaces) using inclusive definitions of politics and deliberation (Graham, 2008, 2010, 2012; Graham & Harju, 2011; Wright, 2012a,b). Thus, we propose to analyse the MSE discussion forum.

The MSE forum is part of a larger website and e-mail list operated by the finance guru, Martin Lewis.6 The forum has 54 different topic areas, most devoted to different areas of a person's financial life (e.g. credit cards, mortgages) but with broader chat areas (e.g. Money Saver Arms). The forum has 25,195,926 million posts from 942,588 registered users in 1,788,165 threads within 54 topic areas.7 It is one of the largest forums in the UK and was of interest to us because anecdotal evidence suggests that users make major financial decisions on the basis of advice from strangers and that political talk often emerges.8 An underlying research question was, thus, to determine whether the MSE forum constituted a third space.

The qualitative content analysis focused on 25 randomly selected Superposters. To identify the sample, a list of all participants by frequency of postings was hand-scraped from the MSE forum alongside their basic user data. We then identified all SP1s (n=2,052) before identifying a random sample of 25 from the list in order to investigate their posting behaviour in more detail. Beginning with the most frequent poster, every 80th SP (this was selected based on the total number of SP1s) active during 2008–2012 was selected for analysis. Each profile included apparent links to all of the posts and threads created (and participated in) by the user and the number of times they had thanked another user, or one of their posts had been thanked. On inspection, we found that these were capped at the most recent 300 messages (or the most recent 300 threads participated in). Consequently, we had to work within this limitation; it was not possible to analyse all of the posts made by each of the 25 SP1s. The thread and post creation were scraped off, including what forums they were posting in.

Once the quantitative analysis of SP1s was complete, a qualitative content analysis (Mayring, 2000) of the 25 SP1s' posting behaviour was conducted. The unit of analysis was individual forum posts. The context unit of analysis was the discussion thread. Thus, in order to maintain the social integrity of the discussions, the unit of selection was the discussion thread, as opposed to the individual posting. In other words, one needs to read posts in the context of the whole thread in order to say something about the nature of the debate.

The selection of threads was based on three criteria. First, only threads that were active during 2008–2012 were selected. Second, we decided to take different sized threads as activity might vary between small and large threads. Three sizes were distinguished: small threads (<100 postings), medium threads (≥100 and <300 postings) and large threads (≥300 and ≤ 600 postings). The cap was set at 600 as our initial investigation of the forum revealed that most threads did not exceed 600 posts. Those threads that did were typically large consisting of thousands of posts set over several years. Finally, since we were analysing posting behaviour, at least 50 postings per SP1 were gathered.

Given that the unit of selection was the thread and not the posting, several rounds of sampling were carried out in order to gather 50 postings for each of the 25 SP1s. This was necessary as some SP1s posted only a few messages over numerous threads. In the first round, two small, two medium and two large threads for each SP1 were randomly selected. For those SP1s where 50 postings were not collected, a second round of sampling was conducted by randomly selecting another set of small, medium, and large threads. If 50 or more postings were still not obtained, we proceed to randomly select small, medium, and large threads until this was achieved in a final round of sampling (see Table 1 for posting totals).

Table 1. Number of Postings for each of the 25 SP1s
25 SP1sNo. of postsPercentage of postings

All threads were archived during August, December, and February 2011–2012 and transferred to MAXQDA (a qualitative content analysis software programme) for hand coding. Mayring's procedures for carrying out inductive coding via the use of feedback loops were used for the analysis. Graham's (2008) coding scheme, which was developed to analyse political talk in nonpolitical forums, was initially adopted for the analysis. During several rounds of coding and recoding, categories were modified, merged and deleted, while new ones were created until a final set of coding categories was deduced.

A posting can potentially serve multiple functions; a participant can address multiple participants and/or issues. For example, in a single post, a participant can give advice on applying for disability benefits to one person while, later in the post, debate the recent proposed changes to the NHS. Thus, the ten categories introduced below are not mutually exclusive. In order to increase confidence in the findings and in the reliability of the final coding scheme, an intercoder reliability test was conducted. Once the analysis was complete and the final coding scheme was deduced, two additional coders were trained. A random sample of threads, amounting to 10% of posts, was countercoded. The final coding scheme was relatively reliable, with 8 of the 10 categories scoring greater than 94% for the average pairwise percent agreement, while scoring .68 or higher for Krippendorff's Alpha.9

Superparticipation on the MSE Forum: Quantitative Analysis

The quantitative analysis was designed to give us a broad overview of the behaviour of Superparticipants. Before beginning, it should be noted that the list of users on which the data is based featured just over 500,000 users, significantly less than the near one million users stated on the board statistics page—we use only the visible (list) data here. In total, we identified 2,052 SP1s, who had made 11.8m posts averaging 5,745 posts each. However, using averages is somewhat distorting because this covers users that have been members for 106 days to 3,882 days; made between 2001 posts and 116,074 posts; and averaged 1 post per day to over 100 posts per day. Tables 2-4 below present the core data for all SP1s.

Table 2. Average Posts per Day
Av. Posts Per DayNo. of SP1s
Table 3. Length of Membership (Days)
Length of MembershipNo. of SP1s
0-2 Years115
3-4 Years648
5-6 Years828
7+ Years461
Table 4. How Number of Posts Affects Behaviour
No. of PostsNo. of SP1sAv. Posts per dayAv. Membership length (days)

The first striking finding is the number of users that have created many thousands of posts. Put simply: There are many SP1s on the MSE forum. While they account for only 0.4% of the listed users, they made 47% of all posts. It can be said, thus, that the SP1s on this forum are quantitatively dominant and that the MSE forum broadly follows the pattern identified in previous websites. However, it must be noted that nearly 200,000 users only posted once. This is, perhaps, unsurprising as MSE is used by many as an advice forum—and we might expect such users to ask questions with a smaller number of “experts” that reply (Welser et al., 2007). Furthermore, SP1s were much more likely to have participated in the preceding months than the whole (around 28% had posted since the turn of 2011 and 75% of our sample—see below—had participated in the previous 24 hours of the analysis date). Put simply, the data suggests that many SP1s were heavily committed to the forum; they spend a significant portion of time each week participating; they continue to do so for a significant length of time; and they accumulate vast numbers of posts. Having analysed the overarching data for all SP1s, we now turn our attention to the 25 SP1s for which we conducted a detailed analysis.

The number of thanks received by users is a strong indicator of how the style and content of their posts are perceived by others. In our sample, 5 users were thanked at least once in more than 90% of all their posts, but 6 users were thanked in less than half of their posts (see Tables 5 and 6). There were similar variations in the average number of thanks per post: 3 users received less than 1 thanks for each post made, but 7 users received more than 5 thanks per post. There were, thus, significant variations in how SP1s' posts were perceived by other users; some rarely received feedback but others were thanked repeatedly.

Table 5. Percentage of Posts Thanked
No. of SP1s% of Posts Thanked
Table 6. Average Number of Thanks per Post
Av. No. of thanks per postNo. of SPs
  1. Note: Data was collected 20 June 2012 due to a problem with the data originally collected.

Thanking other users for their posts is also an important communal act. It was not possible to conduct a complete analysis of the extent to which the SP1s thanked other users; this was capped by the website at the most recent 300 thanked messages. While 79% of the SP1s had posted more than 300 thanks, 5 had not with one user only thanking 7 other users. SP1s also gave some personal information including date of birth or age (9), location (14), interests (5), a biography (5) and occupation (6).10

While it is not possible to conduct a detailed analysis of SP2 behaviour in this article, we can give some initial findings from our sample, and more generally outline their thread creation behaviour. First, it should be noted that our SP1s created new threads (seeds) far less frequently: only two SP1s had created more than 200 seed messages and thus also qualified as SP2s, while only one other user had created more than 100 seeds. There were also significant differences in the number of replies that SP1s received. While we must exclude one user who made at least 300 new threads because we do not know the total number, we found 10 users averaged less than 10 replies, while a further 6 averaged more than 100 replies for each seed message (up to 1,124 messages). Users who made less seed posts tended to receive more replies when they did make one. There were similar disparities with thread views, with the lowest averaging 324 views but the highest averaging 39,496 views. The data provide prima facie evidence that there were significant differences and that some users were more successful at steering debates than others were. However, further qualitative content analysis is necessary to explore this issue.

Super-Participation on the MSE Forum: Qualitative Content Analysis

There were 239 threads from 41 different subforums, which consisted of 40,044 postings, included in the analysis. The first striking finding is that SP1s (all SP1s) were responsible for 49% of the postings; quantitatively speaking, they clearly dominated these discussions and the sample appears to follow a similar pattern to the whole forum (where SP1s created 47% of posts). However, what kind of role did SP1s play in these threads? Did they dominate the discussions in a negative way (e.g. putting off infrequent posters), or did they behave more like Oldenburg's regulars and facilitate and enhance conversation, political talk and community building in general? We now turn our attention to their discursive behaviour.

The 25 SP1s under investigation contributed 1,699 postings to the 239 threads, accounting for 4% of the total postings. When taking a closer look at their postings, several clear patterns emerged. First, a common criticism of online discussion forums has been that they tend to foster shouting matches as opposed reciprocal exchange among participants. However, this was not the case for our sample. In particular, 81% of SP1s postings were coded as replies. In other words, they were frequently reading and responding to fellow participants' posts. Second, as Figure 1 shows, 4 clusters consisting of 10 posting behaviours were identified by the analysis. These were a) storytelling/providing personal information, banter/humour; b) advice giving/helping, requesting advice/help, acknowledgements; c) arguing/debating, interpreting/clarifying, providing factual information; d) degrading and curbing.

Figure 1.

Overall SP1 posting activity in percentages

The first cluster of behaviours identified by the analysis was closely linked to social bonding. The most common behaviour was storytelling/providing personal information, which accounted for 36% of the total messages posted by the 25 SP1s. For 16 of these participants, it represented a quarter or more of their postings. SP1s frequently shared their stories (e.g. personal experiences), account of events (e.g. status updates on what they did that day) or other personal information (e.g. likes/dislikes, interests, information about their family/friends/relationships), as the two examples below exemplify:

Example 1: Hi all, what have I missed?? Hope all are ok,Mallorca was good-weather a bit rubbish for the first 10 days,but last 8 days were glorious so not all bad..had forgotten how expensive it has got! Ate onboard the boat one night, then just went ashore to a little beach bar for a nightcap-2 rounds of drinks for adults,and 3 kids rounds.....how much ????? 102 euros!!!!! WTF?????

Anyways, new boat is awesome, glad to be back -even if DH is freaking out over the stock market crash-I was woken with a cup of tea this morning-and the instruction to get out of bed and look for a job as we are penniless!! Ho hum....

Example 2: I looooove the white company; everything is just so clean, fresh and elegant looking [Elizabeth], I also really dislike Liz Earle. It even feels greasy when you put it on your face. I do, however, ADORE the Eve Lom cleanser - pricey, but definitely worth it for my skin! May I give stardrops a special mention… Oh, I also like hello kitty

Storytelling and the providing personal information seemed to serve two purposes. First, such behaviour tended to foster everyday conversation among forum participants. SP1s were not only sharing their (sometimes intimate) experiences and information, but also were often eliciting and encouraging others to share their stories. As a result, participants got to know each other; they bonded and developed friendly relationships. Outside everyday conversation, SP1s also used storytelling and personal information as a means of fuelling and informing (and sometimes sparking) debate and discussion on particular issues (Dahlberg, 2001a).

The use of humour/banter was another behaviour that seemed to foster a friendly communicative environment. It accounted for 11% of the postings. On occasions, SP1s engaged in good-natured teasing and the exchange of witty remarks with fellow participants, as the exchange of postings between 3 SP1s below illustrates:

SP1a: I don't care if it is your birthday. You're cruising for a bruising fella.image

SP1b: Awwww, look at you go image

SP1c: It's so touching, loves young dream.image

[SP1a] you may be younger and fitter than me, but I've got thirty years experience of duffing up the public!image

SP1a: Yes sir.image

SP1c: feeling a little sick but I'll bash on.

SP2b: that's dedication for ya.

SP1c: Revenge sweet revenge. imageBravo [SP1b].image

Such behaviour seemed to act as a form of social bonding, strengthening the relationships between participants, which is similar to Graham's (2010) findings. On further inspection, we discovered that banter was mostly done ‘in house’: SP1s engaged in banter with each other as opposed to with non-SP1s. However, this is not surprising as SP1s are highly active and develop relationships with each other, allowing playful and flirtatious communicative practices to emerge thereby strengthening those relationships further. That said, the use of banter/humour, as Basu (1999: 390–394) has argued, seemed to act as ‘social glue’: it fostered a warm atmosphere, allowing a more productive communicative environment to emerge.

A third cluster of behaviours identified by the analysis focused on consultative and supportive discourse. Advice giving/helping was the second most common behaviour, representing 19% of the postings. During the course of many of the discussions that took place, SP1s took on the role of a consultant by providing ‘expert’ advice to fellow forum participants (typically to non-SPs) on everything from consumer rights to loan lending procedures, which is in line with Albrecht's (2006) findings. Many SP1s claimed to be ‘experts’ such as accountants, financial advisors and lawyers. For example, the posting below was from a participant claiming to be a lawyer:

As I said, your only legal option if a customer fails to return a item is civil action, as the contract is treated as it were never entered into. Some get around them by doing as eBay but that is contrary to the consumers statutory rights and if in the event that OFT get sufficient complaints and prosecute eBay the odds are that eBay would be found to be in breach of both the DSRs and indeed Section 6(4)(g) of the Consumer Protection From Unfair Trading Regulations which makes it a criminal offence if they were to insist that such a right does not exist.

However, it was not always about providing ‘expert’ advice. They also drew from their life lessons and experiences as a means of offering ‘friendly’ advice and recommendations, as the example below illustrates:

Hi, I have been thinking about the question you ask for sometime now and this what I have come up with:

  1. When life becomes a bit too much I watch The Life of Brian; sounds flippant but it isn't and it really helps (or any programme you find amusing and makes fun/shows defiance to misfortune).
  2. Analyse - make sure that I know why I feel as it is all becoming a bit too much.
  3. Write down the actions that will make me a) feel better; b) change a situation.
  4. If there is nothing that can be done immediately either a) plan when to do; or b) forget about it (by putting it in an imaginary F*ck It Bucket'

When it is about debt there is also another strategy: 1) Calculate your net worth - this changes perspective on debt because you realise that by paying it off you are increasing your net worth just as you do if you are saving. As to the starting point - well not much can be doen about this one except make sure that you improve on it. 2) Look at your budget again. If you notice that you spend very little proportion (or nothing) on having some sort of enjoyment in life re-do the budget. You need to feel good to be able to keep paying off with persistence, determination and patience. 3) Are there any ways to earn more? Some of the additional earnings should be kept for 'life enjoyment' and for 'security savings' but the lion share should go to debt repayment. Well, this is what I do anyway. And what I have noticed is that my 'wants' are becoming fewer, my expenditure has been much reduced and my net worth is increasing at a steady pace.

SP1s were also often helpful. For example, when forum participants asked for directions, recipes or contact information, SP1s were typically the ones there to help. Requesting advice/help among SP1s, on the other hand, was not as common, representing 5% of postings: SP1s largely gave rather than sought advice.

The use of acknowledgements was another behaviour identified by the analysis, which was supportive in nature. It accounted for 14% of the postings. This included postings where a participant acknowledged the presence, departure or action/situation of another forum participant. The four most common acknowledgments used by SP1s were words of encouragement, statements of sympathy and understanding, complimenting and thanking. In many cases, SP1s provided emotional support to fellow forum participants who, for example, were having difficulties with claiming benefits, going through a divorce or debt crisis, as the posting below illustrates:

Good to see you posting again this morning hon - and I am glad you and you OH are now singing from the same hymn sheet. Do not despair - you are already on your way out the debt situation because you have now taken steps to arrest the debt spiral. Rest assured it can be done, with grace and elegance. One month ago I became debt free - paid of 10k of debt in two years and one of those years was spent on benefits and the other as a single mum on minimum income, so take heart.

In many of these cases, SP1s engaged in empathetic behaviour while providing emotional support, which is in line with Preece's (1999) findings. Overall, such behaviour seemed to have created an environment that was felt by participants to be supportive and welcoming thereby strengthening the bonds between them.

Another cluster of behaviours identified by the analysis was closely linked to political talk. As discussed above, one of the aims of this article was to investigate the extent to which the MSE Forum functioned as a third space—an environment that fosters informal political talk. Thus, in addition to the analysis discussed above, all 239 threads were coded for political discussions. Graham's (2008) criteria for identifying (and assessing) political talk in non-political online forums, which captures both conventional and lifestyle-based political issues that arise during the course of everyday conversation, were adopted. All threads that contained a posting in which (a) a participant made a connection from a particular experience, interest or issue to society and which (b) stimulated reflection and a response by at least one other participant were coded as political threads.

After applying the criteria, 77 threads, which consisted of 39% of the postings, were coded as political threads. Political talk here dealt with everything from health care reform to the underlying causes behind the 2011 London riots. The topics also ranged from more conventional political issues such as government policies on social housing to more lifestyle-based political issues such as being a vegetarian and what it means to be a ‘good’ parent. Political talk was not confined to any particular subforum. Indeed, it emerged in more than half of the forums under investigation from the Food Shopping & Groceries, Discount Codes 'n Vouchers and Pets & Pet Care forums to Benefits & Tax Credits, Money Saving in Marriages, Relationships & Families and Money Saver Arms.

Arguing and debating, which represented 15% of the postings, was closely linked to political talk. The analysis revealed that when SP1s argued/debated, it was almost exclusively done during the course of a political discussion, representing 92% of these postings. But several questions remain: How deliberative were SP1s in these debates? Were they reciprocal, rational and critical? Did they use evidence to support their claims?

Overall, SP1s were broadly deliberative when engaging in political talk. As mentioned above, the level of reciprocity as a whole was high. When taking a closer look at the 77 political threads, we found that 87% of SP1s postings were coded as replies. These exchanges tended to be rational and critical in nature. In particular, the level of rationality was high; 86% of the claims made were reasoned as opposed to assertions (nonreasoned claims), indicating that being rational was the norm. There also was a high level of critical reflection; i.e. SP1s tended to be critical of other participant's claims, which represented 72% of their arguments; these threads hosted a diversity of opinions. Moreover, when SP1s posted arguments, nearly half provided evidence (facts/sources, examples, comparisons or personal experiences) in support of their claims. One common practice was the use of personal experiences, as the example below illustrates:

I'm brilliant at managing my money - but I'll be honest. My child Tax Credits are spent on luxuries for my children - LIKE FOOD!!!! Like essential petrol to get him school -too far to walk both kids along roads with no pavement but car is ONLY used for school runs. Like “essential” clothing to keep them wind and water tight. There is no LUXURIES contingency in the tax Credit system just for basics. We have no spare income for activities at the moment - and if either of them need new shoes before the end of term I am scuppered because there is no excess in the budget.

In this debate, the participant was defending the child tax credit benefit from those who thought it was being abused. As a means of supporting her argument, she uses her personal experience of being on such benefits. In some of these cases, SP1s used their personal experiences as an authoritative position to speak from, which is in line with similar studies on third spaces (Graham, 2012; Graham & Harju, 2011; Van Zoonen, 2007).

The other two behaviours linked to political talk were interpreting/clarifying and providing factual information, both of which were pedagogical in nature. Interpreting/clarifying accounted for 7% of the postings. In these cases, SP1s would help fellow participants by explaining, interpreting or clarifying the issues and topics under discussion. The most common practice was clarifying (mis)information:

Poster: They may still be entitled to something if they needed to help their OH, e.g. get in/out of the bath, dress themselves, with eating, etc. You don't have to be there 24/7.

SP1: You are mistaken. To claim carers allowance (CA)the person you care for needs to get DLA middle rate care which means more than hours worth of care a day. The OP's husband does not even claim DLA at the moment.

As this example shows, SP1s acted as a form of quality control, correcting and clarifying information posted by forum participants. They also helped others by explaining the arguments and positions being put forth. As the posting below illustrates, this sometimes included providing a summary of what the debate was about for newcomers or those who were having difficulties following the discussion:

I've been reading through most of this thread, and I get the impression that it's moving a bit too fast for you and sometimes you can't keep up with the advice people are trying to give you, so I'll try and break it down. […]

They acted too as ‘experts’ by interpreting laws, rules and procedures on things such as consumer rights/laws, and traffic and parking violations, as the example below shows:

Poster: That's what this says - but what does “The road forks around the farm for engineering reasons owing to the surrounding area's geology,” actually mean?

SP1: It means what it says - the geology of the area meant building around the farm was a cost effective solution. The Wiki entry photo doesn't show it very well, but the westbound carriageway is higher than the eastbound. The M6 at Shap also splits to accommodate the contours.

Finally, SP1s even provided interpretations of situations in other forum participant's lives:

As a glasses wearer for nearly 30 years i think i have worked out what has happened regarding the OP situation […]

On many of these occasions, SP1s would use their own life experiences as a mirror for explaining and interpreting events in other participant's lives. Overall, these findings support those of Rauch (1983) and Nonnecke and Preece (2003—cited in Albrecht, 2006).

Not as common, but related to interpreting/clarifying, was the behaviour of providing (unsolicited) factual information. On occasions, SP1s dropped links to or articles/reports/information from news media, government or other institutions and organizations as a means of fuelling and enhancing the quality of debate.

The final cluster of behaviours identified by the analysis consisted of degrading and curbing, two negative behaviours commonly associated with online forums. Degrading (or flaming)—to lower in character, quality, esteem, or rank another participant's post or person—is behaviour typically associated with online political talk (see e.g. Jankowski & Van Os, 2004). Indeed, 86% of the postings coded as degrading occurred during the course of a political discussion. This is not surprising given the argumentative and aggressive nature of political talk. That said, such behavior impacts political talk in a negative way. For example, Jankowski and Van Os's interviews found that this type of communicative practice turned people away from debates, negatively affecting access to the discussions. Curbing—an attempt to suppress, restrict or prevent another participant from raising an issue or voicing an opinion—was another negative behaviour identified by the analysis. Curbing here was typically directed at suppressing a particular position, topic and, on a few occasions, a type of behaviour (i.e. arguing/debating), as the examples below show:

Example 1: “So far, you're the only one preaching on this thread. Why would you even want to come on to a thread about vegetarian food if you feel so persecuted?”

Example 2: “I wonder when this thread is going to be allowed to die or is the playground rhetoric going to continue?”

Like degrading, this behaviour seemed to be limited to political talk with 82% of curbing taking place within political threads. Overall, however, such practices—degrading and curbing—were infrequent, accounting for only 3% of the postings.


This article has attempted to illuminate an underresearched area of online forum communication: the nature and impact of superparticipants. We have developed a theoretically informed typology that we hope will help researchers to identify the presence of superparticipants so that their activities can be subjected to detailed analysis. Our quantitative analysis identified 2,052 SP1s. Super-posting was not, thus, the preserve of the few—though this was only 0.4% of all users—suggesting a long tail. Moreover, they made nearly 50% of all posts. Thus, they did quantitatively dominate debates. However, the qualitative content analysis has found that SP1s did not normally attempt to stop other users from posting (curbing) or attempt to attack them (flaming). They undertake a range of largely positive functions and roles within the forum, including helping other users; replying to debates and summarising longer threads for new users; being empathetic towards others' problems; and engaging in (largely) rational critical debate. The often humorous and familiar nature of debate suggests that this was a virtual community, and SP1s played a significant role in facilitating this. It seems as though, at least in this case, SP1s helped to create continuity and facilitated an inclusive environment. We argue, thus, in contradiction to most of the received wisdom, their role was largely positive; they were performing a similar role to Oldenburg's regulars. More generally, the analysis suggests that the MSE forum was a third space—with significant amounts and deliberatively constructed political talk; strong evidence of community; and reciprocal and altruistic activities. This is interesting because Habermas' ideal speech situation requires that differences of status, authority and ethos be lifted. It could be argued that SP1s, through the volume of their participation and (often) presentation of themselves as experts contradict the ideal speech situation. However, our analysis suggests that the more general fair play criteria of the ideal speech situation (in the sense that SP1s largely did not curb or attack other users) were met and thus the role of SPs needs further theoretical problematization. Oldenburg's positive account of the regulars has proved illustrative. It suggests that further research is necessary into the nature and function of political talk in third spaces, and the role of SPs in particular. It would also be worthwhile considering the extent and role of super-participation in civic life more broadly: Do SPs facilitate inclusive civic discourse and promote community formation in other environments?


The second author was supported by a grant from the British Academy (MC110323). Versions of this paper were presented at the AOIR Conference (2012) and ECREA (Istanbul).


  1. 1

    Attempts to predict posting patterns, such as Lotka's Law, are not specifically designed for the nature of online publishing and have had limited success at predicting post distribution. For example, Kuperman (1996) found dominant minorities in e-mail lists, but the lower publication barriers made Lotka's Law a poor fit, but the Generalised Inverse Gaussian-Poisson and Poisson-lognormal distributions were stronger. Other widely cited models, such as the Pareto Principle, under-predict the impact of dominant minorities.

  2. 2

    Habermas' notions of the public sphere and ideal speech situation have received much criticism. For example, Fraser (1992) and Benhabib (1992) criticize his account for excluding women from public life and the distinction made between public and private issues. See Calhoun (1992) for a comprehensive overview.

  3. 3

    We have chosen not to use the title of regulars because we believe this lacks specificity (see typology) and does not enable us to capture the different types of SP we identify.

  4. 4

    See: www.big-boards.com - though this list is far complete, it is a useful indicator.

  5. 5

    In the context of a third place, facilitators are similar to the pub landlord.

  6. 6

    After the analysis was conducted, the forum was sold for £87m.

  7. 7

    The analysis was conducted on 21 August 2011.

  8. 8

    This has arguably become all the more important as the government has cut back its own consumer advice bodies and asked that the “Big Society” step in to fill the gap.

  9. 9

    The categories curbing and providing factual information received .60 and .33 for Krippendorff's Alpha. The low scores here can be contributed to the infrequency of the two codes in the sample. For both, the average pairwise percentage of agree was 99%.

  10. 10

    In some cases, personal information appeared false or was intended to be a joke. No user gave each piece of personal information while six gave no personal information at all.


  • Todd Graham (e-mail: t.s.graham@rug.nl) is an Assistant Professor of Political Communication at the Groningen Centre for Media and Journalism Studies. His main research interests include (new) media and democracy, popular culture, online journalism, online deliberation and public sphere theory.

    Address: University of Groningen, Department of Journalism, Oude Kijk in 't Jatstraat 26, 9712 EK Groningen.

  • Scott Wright (e-mail: sw335@le.ac.uk) is Senior Lecturer in Political Communication and Mid-Career Fellow of the British Academy. His current work focuses on empirically fleshing out the ‘new’ agenda for online deliberation: informal political talk in third spaces.

    Address: Bankfield House, New Walk, University of Leicester, Leicester, LE1 7JA.