JCMC 8 (4) July 2003
Message Board
Collab-U  CMC Play  E-Commerce Symposium Net Law  InfoSpaces Usenet
  NetStudy VEs VOs O-Journ HigherEd Conversation Cyberspace Web Commerce
Vol. 6 No. 1  Vol. 6 No. 2  Vol. 6 No. 3  Vol. 6 No. 4
 Vol. 7 No. 1  Vol. 7 No. 2  Vol. 7 No. 3  Vol. 7 No. 4  Vol. 8 No. 1  Vol. 8 No. 2  Vol. 8 No. 3  Vol. 8 No. 4

Interactive Internet?
Studying Mediated Interaction With Publicly Available Search Engines

Paul Wouters and Diana Gerbec
Networked Research and Digital Information (Nerdi)
Royal Netherlands Academy of Arts and Sciences


Abstract

This study explores the use of freely available search engines to locate mediated interaction on the World Wide Web. We use the concepts of mediated interaction and quasi-interaction as developed by Thompson (Thompson, 1995) and Slevin (Slevin, 2000). We conclude that the publicly available search engines lack stability of results, that their behavior is not transparent, and that they do not present the results in a way that is suitable for the creation of data sets. The study confirms that the Internet and the World Wide Web can mainly be characterized as instantiations of mediated quasi-interaction rather than of mediated interaction. Internet researchers might consider not only to try to develop better search software, but also tools that can archive publicly available mediated interaction in real time at a large scale. Even with improved tools, however, we should not expect that the Internet will give us anything like "total data" about social life. The use of information and discussion on the Internet in off-line contexts cannot be deduced from the Internet itself. To answer these questions, Internet research must be combined with off-line interviews, observation and surveys.


Introduction

In common parlance, the Internet is closely related to the concept of interaction (McMillan, 2002). Using the Internet, one can quickly locate people with compatible interests, find partners for collaboration, and exchange goods at a high speed (Lievrouw & Livingstone, 2002). The concept of the virtual community (Rheingold, 1994) was initially based on the idea of fast, seamless interaction that would be impossible in the old-fashioned off-line world. But how interactive is the Internet really? A recent estimate by the information economists Lyman and Varian shows that the largest amount of information on the Internet (measured in bits) is available in the form of videos and films (Lyman & Varian, 2000), unidirectional forms of information. Although e-mail is the most popular activity on the Internet, it accounts for only a small part of the total of information on the Internet (Lyman & Varian, 2000). In itself, this does not have to be very interesting. After all, it may simply be the consequence of the technical properties of the small e-mail message compared with the bulky nature of a video file. To estimate the extent to which the Internet promotes interaction, a measure of social rather than technical interaction is needed.

In this paper, we explore the extent to which it is possible to estimate the amount of interaction between humans on the Internet with existing, publicly available, information retrieval tools. We are therefore not after a measurement of relative amounts of information bits (Shannon & Weaver, 1949), but after an assessment of the relative amount of instances of interaction on the Internet. In other words, we are more interested in the concept of information as defined by Bateson (Bateson, 1972) than in the one proposed by Shannon (Shannon, 1948). Our aim is moreover not quantitative in itself. As noted in the introduction to this special issue, the Internet seems to make traceable, because of its mediating function, interactions that were previously ephemeral (like chatting and gossiping). It may also provide new means of expression and interaction. We aim to recover traces of interaction on the Internet in order to analyze the content of these interactions in a comparative mode. We are therefore not primarily interested in ephemeral instances, but in traces that are left on the Internet and can be analyzed with hindsight.

Technically, this means that our goal is to locate information on the Internet that has been created by interaction between at least two different people. For this, we decided to use the publicly available search engines (see http://www.searchenginewatch.com). Search engines search the Web, which is clearly not identical to the Internet. It is only one of the many Internet protocols. The Web has, however, become the dominant protocol and virtually all other protocols (e-mail, file transfer protocol, chat, virtual spaces such as MUDs and MOOs, and dedicated databases) are now accessible via Web interfaces. Search engines are moreover the tool that most people have to use if they wish to locate information. This is also true for social scientists and Internet researchers. We therefore chose to limit ourselves to the publicly available search engines. In the last section of this article, we will discuss to what extent this limits our method, and whether new specific software tools may be needed to analyze Internet based human interaction.


Characterizing Interaction on the Internet and the Web

The Internet can be conceptualized as a container of different communication media. Using this perspective means that Thompson’s concept of cultural transmission is central to our approach (Thompson, 1995). Slevin has used this theory of media of mass communication to analyze the Internet and the World Wide Web (Slevin, 2000). In Thompson’s theory, three forms of interaction among human beings1 are important: face-to-face interaction, mediated interaction, and mediated quasi-interaction (Thompson, 1995). In face-to-face interaction, people share time and space, since they are co-present. In mediated interaction the sending of the message and its reception are separated in time and space. The media moreover co-shape the message. Mediated quasi-interaction has the same properties. It differs from mediated interaction in that it is not a priori given who the audience of the message will be.2 A telephone conversation or an e-mail exchange between two or more people is a form of mediated interaction. Mass media such as television, radio and the newspapers are instances of mediated quasi-interaction. An e-mail message to a mailing list is also a form of mediated quasi-interaction.

Slevin has used this typology to discuss the properties of the Internet in this respect. The Internet enables new forms of mediated interaction in both synchronous (chat channels) and asynchronous form (e-mail). Face-to-face interaction does by definition not exist on the Internet. All interaction is mediated. Static Web pages3 are good examples of mediated quasi-interaction. They are comparable to newspapers or advertisements. Dynamically generated Web pages, for example the results of searching a database, are also forms of quasi-interaction. The Internet enables new forms of combinations of these different types of interaction. A Web page may for example contain a static page with information, a button by which one can comment on this information, a link to a chat room about the topics of the Web site, and a button to subscribe to some e-mail service. Clearly, the Web protocol itself can host many different forms of communication and cultural transmission, even at the level of the individual Web site. This property can therefore also be used to analyze these different modes of communication.

According to Slevin, the Internet actually shares most of its properties with the old mass media: I shall argue that the mechanisms by which the Internet and intranets facilitate public spheres continue to have a great deal in common with the mechanisms deployed in non-dialogical mediated publicness as set out by Thompson (Thompson, 1995). Those who think that Internet technology will create a dialogical mediated publicness will be inclined to be disappointed by any practical implementation in this direction. I shall argue, therefore, that following through Thompson’s ideas (Thompson, 1995), the Internet is facilitating a space of the visible that can best be described as a deliberative mediated publicness rather than a dialogical one. (Slevin, 2000; italics in the original)
Slevin bases his argument mainly on theoretical grounds (Slevin, 2000). Given the nature of Web pages, his argument seems plausible. Yet, the empirical basis of his claim is non-existent. Using Thompson’s characterization of interaction (Thompson, 1995), we can rank forms of Internet communication (see Table 1).

Chat channels Mediated interaction Ephemeral
E-conferences Mediated interaction Ephemeral
E-mail Mediated interaction Ephemeral
E-mail forums and news groups Mediated interaction Traceable
Virtual communities Mediated interaction Ephemeral
Message boards Mediated interaction and quasi-interaction Traceable
Web sites with archive of dialogues Mediated quasi-interaction4 Traceable
Web sites with information Mediated quasi-interaction Traceable

Tabel 1. Interaction and Quasi-interaction on the Internet.


This typology is useful, we think, to characterize the types of interaction on the Internet and the Web. Given our research question, we have restricted ourselves to the traceable forms of mediated interaction.


Mapping Controversy on the Web

The way people interact using the Internet has been studied by communication scientists, library and information scientists, and Internet researchers (Borgman & Furner, 2002; Dutton, 1996; Gauntlett, 2000; Herring, 2002; Jones, 1999; Mann & Stewart, 2000). The topics have varied from mediated interaction in communities and collaboratories (Davenport & Hall, 2002; Finholt, 2002; Rheingold, 1994; Van den Besselaar, Tanabe, & Ishida, 2002) to forms of quasi-interaction such as scholarly publishing (Kling & McKim, 2000) and Web site production (Park, Barnett, & Nam, 2002; Pew Internet & American Life Project, 2002; Tang & Thelwall, 2002).

While these studies have provided valuable insight about Internet-based interaction and communication, they provide no basis for estimating the extent of interaction on the Internet relative to non-interactive modes of communication. The case studies, by definition, do not allow comparative analysis. Another body of work has analyzed the role of the Internet on the basis of user surveys (Voorbij, 1999). This type of work usually does not deal with the content of the interaction but focuses on frequency of use, browsing behavior, and the social networks that are maintained with online and offline communication (Garton, Haythornthwaite, & Wellman, 1997; Haythornthwaite, 2000). Another body of work focuses on non-use of the Internet (Wyatt, Thomas, & Terranova, 2002).

Of more direct relevance to our research question is the work by Marres and Rogers (Marres & Rogers, 2000). They aim to map the controversy of the debate on genetically modified food (GM food) on the Internet. They do this by creating a map of all the hyperlinks between Web sites dealing with GM food, with the help of a software tool they developed specifically for this purpose. From our perspective, they do not in the first place provide an analysis of interaction about genetically modified food. Rather, they give a characterization of hyperlink strategies of the actors in the dispute about GM food. Although they claim to analyze the dynamics of the GM food controversy, the map drawn is the result of the summation of links over a short period of time. The resulting analysis is therefore not a dynamic but a static one. Their overview of hyperlinking strategies is highly relevant to the sociological analysis of the debate on GM food. Their study is not, however, about mediated interaction but about the link structure of mediated quasi-interaction on the Internet. At the end of the paper, we will come back to the issue of how this type of work relates to a more encompassing analysis of social controversy.

Our study of interaction on the Internet is complementary to the study by Marres and Rogers (Marres & Rogers, 2000). We also focus on the debate about GM food (see also the paper by Hellsten, this issue). We took this debate as our case, because we know that GM food is a hot topic, which has generated a lot of controversy at the international as well as national level (Frewer, Howard, & Shepherd, 1997). The debate remains the topic of numerous newspaper articles and TV reports. This makes the GM food controversy a suitable case study to assess the extent to which interaction on the Internet is a significant mode of communication in large-scale social debates.


Research Question and Method

We can now formulate our research question on the basis of the argument developed so far. As stated above, we wish to explore to what extent it is possible to estimate the amount of interaction on the Internet with existing, publicly available, information retrieval tools. We have taken the debate about GM food as the case study for this methodological exploration. More specifically, our question is: We have operationalized our questions as follows. By interaction, we mean instances of interaction between actors about GM food. An interaction is present if and only if more than one actor is directly involved in the interaction. The interaction should moreover be publicly available and accessible. Of course, the boundary between interaction and quasi-interaction is not always clear. We have therefore added the criterion that the interaction is also institutionalized as an interaction. This can often be taken from subject lines, the name of dedicated groups or e-mail lists, etc. As stated above, we have excluded all ephemeral instances of interaction from our analysis, since we are interested in large-scale analysis of the dynamics and content of Internet-based interaction.

We have performed our analysis in three different stages. In all stages we submitted the following logical string to these search engines:

discussion AND food AND (genetic* OR modif* OR engineered)

Before these three stages we had tried out several other forms of the search string. The word "discussion" turned out to be necessary. Without it, we obtained only Web sites sending information about GM food. The typographic format of this string varied according to the grammar of the specific search engine. The use of the wildcard * is quite different in different search engines, as are the possibilities of directly applying Boolean search techniques. The order also had to be changed from time to time to improve the total number of hits. Since the search and presentation algorithms of the search engines are not publicly available, it was not always clear to us why some ordering seemed to improve the results. In the first stage in December 2001, we used all search engines and metacrawlers that were at that time publicly available via the Website http://www.searchenginewatch.com (Sherman & Price, 2001). This led to a total of 107 different search engines. In the second stage, performed in February 2002, we repeated the search again using the search engines as listed at http://www.searchenginewatch.com, restricting ourselves, however, to the category "major search engines." In the third stage, in September 2002, we restricted the number of search engines to a core set of the six best search engines in the previous rounds. This was based on the finding from the first two rounds that the other search engines did not give any results that were not already covered by one of the core set. We also took note of the fact that the number of search engines had changed in the course of the 8 months covering the three stages of our research. In this stage we varied the search string by replacing “discussion” with “conference” and “forum” respectively.

In all cases, we worked as follows. Usually, we obtained hits ranging from zero to several hundreds of thousands of hyperlinks. We checked the returned hits as they appeared on the second, third and fourth page of results. The relevance of these hits was virtually non-existent. We therefore restricted our analysis to the first page of results. We moreover visually inspected the second page if this page could be expected to have additional results (this varies by search engine). In those cases in which the number of relevant hits seemed to decrease more slowly than usual, we even inspected the third and fourth pages. This was done only a few times, however. In this way, we obtained a list of search engine results (SERs). Each SER was subsequently inspected to see whether or not the hyperlink seemed to link to an instance of mediated interaction. If this was the case, the link was added to a list of primary inquiry results (PIRs). After this, the list of PIRs was visited to check whether the dialogue was indeed about GM food. If this was the case, the content of the dialogue was downloaded for content analysis.5


Results

In the first stage of the 107 search engines used, no more than 23 search engines gave some dialogical results. In this way, we obtained 120 instances of a dialogue about GM food. Of these instances, 40 were unique (the others were copies). The Usenet archive deserves separate mention. This archive is accessible through Google Groups. Using this feature of the Google search engine, we were able to identify 596 different news groups in which dialogue about GM food had taken place. It should be noted that these instances of dialogue were not recorded "in real time". Because of the use of search engines, we found traces of dialogues that had happened in the past. The search engines that produced the highest hits are listed in Table 2.

Google Groups 59+
Yahoo 20
Altavista 9
Google 8
Web Crawler 7
One Seek 6
Dogpile 5
MSN 4
qbSearch 4

Table 2. Search engines with highest scores on GM food interaction.


In the second stage, February 2002, we repeated the search for all search engines in the category “major search engines” as listed on the http://www.searchenginewatch.com Web site. We did not repeat the use of the other search engines since their use had not proved quite so useful in the first stage. We did not, however, search the Usenet archive through Google Groups. Eleven search engines gave hits that seemed to indicate interaction about GM food. The number of hits was significantly lower than our search in December 2001. Yahoo produced 5 hits, the other ten search engines varied between 3 and 1 hits. This gave a total of 23 instances of dialogue.

In September 2002, we varied the search string as noted above and applied it to the 6 search engines that had given the best results in the first two stages. As noted, we dropped the use of meta-crawlers since they systematically came up with the same results as the major search engines in the first two stages. Using the same search string (with the word “discussion”) yielded the results listed in Table 3.

Google Groups 28
Altavista 8
Web Crawler 3
Google 1
Yahoo 1
All the Web 0

Table 3. Results of the 6 search engines in the third stage.


Replacing the word "discussion" with "conference" radically diminished the number of hits. Only Yahoo gave 8 discussion groups. All other search engines returned 0 hits. Using the word “forum” in the search string led to 5 discussion groups by Google Groups and 3 by Yahoo. All other search engines returned no discussion groups in their results.

Although analyzing the discussion groups goes beyond the scope of this article, we did inspect the results to decide whether or not a link actually pointed to mediated interaction about GM food, instead of to quasi-interaction. The vast majority of results pointed to Web sites containing information, newspaper clippings, and scientific articles about GM food. Even links that seemed to point to a discussion group more often than not pointed to an archive of articles about GM food in the traditional mass media. The Web is clearly a great way to distribute newspaper articles and other printed information about GM food. Of the number of links that actually pointed to an instance of mediated interaction, most were not dedicated to GM food as the main topic. That is, most hits we got were side remarks about GM food in the framework of discussion groups about more general topics or simply quite different domains. There are only a few discussion groups that devote themselves to GM food. Of course, we counted all of these latter groups as interactions about GM food. If the discussion group was actually focusing on another topic, we counted those discussions as interaction if a dialogue was created about GM food. A lone message about GM food in a discussion group about cooking was therefore not counted as interaction about GM food. We also found that the length of the threads shows a skewed distribution: most are very short; only a few threads are long.

This finding may be related to our finding that the number of instances of mediated interaction has decreased from our first to our third search stage. We would like to repeat that this decrease does not have to mean that there is less discussion going on in chat group, e-conferences and the like. It does mean, however, that the attention to GM food in those discussion lists and news groups that are archived has decreased. A possible explanation might be that most of these discussions are by lay people for lay people. Although the general interest in GM food does not seem to have diminished, given the regularly recurring waves of news stories about biotechnology, the first wave of excitement and news has clearly passed. Discussions among lay people may therefore not bring up many new issues and are prone to die out. Professional discussions on the topic seem to be longer lasting.7


Conclusions: How Interactive is the Web?

We wish to raise two points in the discussion of our results. First, we discuss which conclusions we can draw about the use of the publicly available search engines to probe the Web and the Internet. This relates to the question about the relation between interactivity and the Internet as a whole, in contrast to its Web components. We also bring up the issue of the need to develop tailor-made search engines and robots for social scientific research. Second, we address the question of what our findings say about the interactive nature of the World Wide Web and of the Internet.

The problematic nature of search engines has been discussed in the framework of social science research by a number of information scientists (Bar-Ilan, 1999, 2001; Bar-Ilan & Peritz, 2002; Introna & Nissenbaum, 2000; Thelwall, 2001, 2003). Our findings provide, we think, a strong confirmation of the unsuitability of search engines for creating data sets that can be used in social science. First of all, stability of results is not guaranteed. If we repeated the search using the same search engine and the same search string, the results were often quite different within a time span as short as five minutes (Feldman, 1997). Also, the presentation algoritms can produce quite unexpected results. For example, we asked to reorder the 30,000 hits we obtained via Altavista and as a consequence the total number of results decreased to 10,066! We also found that the Google searches were not replicable. One would expect that this might be less problematic in the case of searching through an established archive. Our results indicate otherwise. Our searches through the archive of Usenet Newsgroups through Google Groups show that even these searches do not produce stable results. We do not know whether the actual search algoritm is to blame for this. It might be that the search engine does return the same hits if queried by the same search string. The way it is presented makes it virtually impossible to check out how many unique hits are returned, because they are drowned in a sea of copies.

Thus, the publicly available search engines lack stability of results, their behavior is not transparent, and they do not present the results in a way that is suitable for the creation of data sets. As has already been concluded by Bar-Ilan (Bar-Ilan, 2001) in the context of searching for information (mediated quasi-interaction), Internet researchers clearly need to give priority to developing their own software tools for data collection, if they wish to make robust statements about social interaction on the Internet and the Web.

Nevertheless, the fact that we uncovered relatively little interaction is not in the first place an artifact of the search engines, but should also be attributed to the nature of the Internet and the Web. Our findings confirm Slevin’s expectation with regard to the Internet. It is very difficult to find instances of mediated interaction on the Web. The overwhelming majority of results we obtained were postings of information about GM food. Most of these were clippings and articles from traditional media. Even within the domain of discussion and news groups, most hits were less dialogical than they seemed. It is striking how often discussion groups and e-mail forums are used for the distribution of printed articles about GM food. The mountain of information about GM food that we uncovered gave birth to a very small mound of mediated interaction.

This conclusion is not as trivial as it may seem to those who think that the Web is mainly about finding information and not about interaction. Initially, the Web was inspired by the idea of promoting collaboration between authors; hence it did turn around the notion of mediated interaction (Berners-Lee, 1999). We did find some instances of interaction. We expected this, since virtually all interaction oriented Internet protocols are now accessible via ordinary Web browsers. The fact that we could track down some interaction confirms this. The fact that we found so little, however, may be an indication of the relative scarcity of mediated interaction that is archived on the Internet and accessible through the Web. We know that there must be interaction going on the Internet in the form of chat channels, news groups, email traffic between individuals, and email lists. We do not know however how much this interaction contributes to the social traffic on the Internet. As in offline life, most of the interaction going on will not be archived. This means that even with improved search engines, we might be able to capture only a small fraction of dialogue on the Internet. For this reason, Internet researchers might consider not only trying to develop better search software, but also tools that can archive publicly available mediated interaction in real time at a large scale. The use of this software and the subsequent results should of course be in accordance with proper ethical guidelines. Even with improved tools, however, we should not expect that the Internet will give us anything like “total data” about social life.

To sum up, the vast majority of communication on the Internet is mediated quasi-interaction. Only a relatively modest part seems to form instances of mediated interaction. A precise estimation of these quantities is presently impossible. More interesting than knowing the exact quantitative relationships is the observation that the Internet adds two features to more traditional mass media. First, it is an accessible, though unorganized, archive of information gleaned from the traditional media. Second, it creates additional possibilities for additional interaction with posted messages. In other words, although the Internet is, like traditional mass media, mainly a medium for quasi-interaction, it creates more possibilities for triggering interaction. The influence this has on social interaction and communication may be invisible on the Internet itself or be ephemeral and non-traceable. The use of information and discussion on the Internet in off-line contexts cannot be deduced from the Internet itself. To answer these questions, Internet research must be combined with off-line interviews, observation and surveys.


Acknowledgments

We would like to thank our colleagues at Nerdi, the participants to the First Virtual Methods seminar (December 2001, CRICT, Brunel University, UK), the participants in the AoIR 3.0 conference (October 2002, Infonomics, Maastricht, Netherlands), and the participants in the ICT colloquium of ASCoR (March 2002, University of Amsterdam, Netherlands) for their comments on earlier versions of this article. We also thank the anonymous referees and the editors of this issue for their valuable review and commentary.


Footnotes

1. In the literature about interactivity and new media, three different types of interactivity have been defined (McMillan, 2002): user-to-user, user-to-documents and user-to-system interaction. Thompson (Thompson, 1995) focuses on the first two forms of interaction, albeit in different theoretical terms.
2. Literally speaking, this may also be uncertain in mediated interaction, of course. Even in face-to-face interaction someone might unexpectedly listen in. The distinction is nevertheless useful since the dynamics of interaction and mediated interaction are different.
3. A Web page is the minimum amount of encoded text and images necessary for a Web browser to display a complete screen (Foot & Schneider, 2002). With a static Web page we mean a piece of html-encoded text which resides in a stable form on a server site. A dynamically generated Web page is created at the moment of the request for the page, usually by software interacting with a database. A Web site is a collection of Web pages that share a base url address.
4. This is an instance of quasi-interaction as far as deposition in the archive means that the mediated interaction is closed. However, the archive can of course be analyzed as the representation of mediated interaction in the past.
5.This analysis will not be presented here, but is the topic of a follow-up paper.
6. Searching through the Usenet archive produces the same type of lists as produced by normal Google searches. This means that the total number of unique hits is not easy to determine. One gets weighted lists of hits of which many are copies. The number of new unique instances of dialogue about GM food decreases with the page number of the result lists. We were able to identify 59 different groups. The total number must therefore be higher. We did not attempt to estimate this number, which could be done by curve fitting of the function that describes the decrease of unique hits by page number, since our interest in this paper is not quantitative in itself. We indicate the number as 59+ in Table 2.
7. We will come back to this issue in our content analysis of the GM food debates we downloaded.


References

Bar-Ilan, J. (1999). Search engine results over time - A case study on search engine stability. Cybermetrics, 2/3. Retrieved December 15, 2001 from http://www.cindoc.csic.es/cybermetrics/articles/v2i1p1.html.

Bar-Ilan, J. (2001). Data collection methods on the Web for informetric purposes - A review and analysis. Scientometrics, 50(1), 7-32.

Bar-Ilan, J., & Peritz, B. C. (2002). Informetric theories and methods for exploring the Internet: An analytical survey of recent research literature. Library Trends, 50(3), 371-392.

Bateson, G. (1972). Steps to an ecology of mind. New York: Ballantine .

Berners-Lee, T. (1999). Weaving the web: The original design and ultimate destiny of the World Wide Web by its inventor. New York: HarperCollins Publishers.

Borgman, C., & Furner, J. (2002). Scholarly communication and bibliometrics. In B. Cronin & D. Shaw (Eds.), Annual review of information science and technology 36 (pp. 3-72). Medford, NJ: Information Today Inc.

Davenport, E., & Hall, H. (2002). Organizational knowledge and communities of Ppractice. In B. Cronin & D. Shaw (Eds.), Annual review of information science and technology 36 (pp. 171-227). Medford, New Jersey: Information Today, Inc.

Dutton, W. H. (1996). Information and communication technologies: Visions and realities. Oxford: Oxford University Press.

Feldman, S. (1997). It was here a minute ago! Searcher, 5(9), 52.

Finholt, T. A. (2002). Collaboratories. In B. Cronin & D. Shaw (Eds.), Annual review of information science and technology 36. Medford, NJ: Information Today.

Foot, K. A., & Schneider, S. M. (2002). Online action in Campaign 2000: An exploratory analysis of the U.S. political Web sphere. Journal of Broadcasting and Electronic Media, 46(2), 222-244.

Frewer, L., Howard, C., & Shepherd, R. (1997). Public concerns in the United Kingdom about general and specific applications of genetic engineering: Risk, benefit, and ethics. Science, Technology & Human Values, 22(1), 98-124.

Garton, L., Haythornthwaite, C., & Wellman, B. (1997). Studying online social networks. Journal of Computer-Mediated Communication, 3(1). Retrieved December 15, 2001 from http://www.ascusc.org/jcmc/vol3/issue1/garton.htm.

Gauntlett, D. (2000). Web.studies: Rewring media studies for the digital age . London: Arnold Publishers.

Haythornthwaite, C. (2000). Online personal networks: Size, composition and media use among distance learners. New Media & Society, 2(2), 195-226.

Herring, S. C. (2002). Computer-mediated communication on the Internet. In B. Cronin & D. Shaw (Eds.), Annual review of information science and technology 36 (pp. 109-168). Medford, New Jersey: Information Today, Inc.

Introna, L. D., & Nissenbaum, H. (2000). Shaping the Web: Why the politics of search engines matters. The Information Society, 16, 169-185.

Jones, S. (1999). Doing Internet research: Critical issues and methods for examining the net. Thousand Oaks, CA: Sage.

Kling, R., & McKim, G. (2000). Not just a matter of time: Field differences and the shaping of electronic media in supporting scientific communication. Journal of the American Society for Information Science, 51(14), 1306-1320.

Lievrouw, L., & Livingstone, S. (2002). The handbook of new media: Social shaping and consequences of ICTs. London: Sage.

Lyman, P., & Varian, H. (2000) How much information? Retrieved February 20 from http://www.sims.berkeley.edu/research/projects/how-much-info/.

Mann, C., & Stewart, F. (2000). Internet communication and qualitative research: A handbook for researching online. Thousand Oaks, CA: Sage.

Marres, N., & Rogers, R. (2000). Depluralising the Web, repluralising public debate: The case of the GM food debate on the Web. In R. Rogers (Ed.), Preferred placement: Knowledge politics on the Web (pp. 113-136). Maastricht: Jan van Eyck Editions.

McMillan, S. J. (2002). Exploring models of interactivity from multiple research traditions: Users, documents, systems. In L. Lievrouw & S. Livingstone (Eds.), The handbook of new media: Social shaping and consequences of ICTs (pp. 163-182). London: Sage.

Park, H. W., Barnett, G. A., & Nam, I. Y. (2002). Hyperlink-affiliation network structure of top websites: Examining affiliates with hyperlink in Korea. Journal of the American Society for Information Science and Technology, 53(7), 592-601.

Pew Internet & American Life Project, W. D. C. (2002) The internet goes to college: How students are living in the future with today's technology. Retrieved September 20, 2002 from http://www.pewinternet.org/reports/pdfs/PIP_College_Report.pdf.

Rheingold, H. (1994). The virtual community: Homesteading on the electronic frontier. Harper Perennial.

Shannon, C. E. (1948). A mathematical theory of communication. Bell System Technical Journal, 27, 379-423 and 623-356.

Shannon, C. E., & Weaver, W. (1949). The mathematical theory of communication. Urbana: Univ. of Illinois Press.

Sherman, C., & Price, G. (2001). The invisible Web: Uncovering information sources search engines can't see. Medford, New Jersey: Information Today Inc.

Slevin, J. (2000). The Internet and society. Cambridge: Polity Press.

Tang, R., & Thelwall, M. (2002). Disciplinary differences in US academic departmental Web site interlinking. Albany: SUNY .

Thelwall, M. (2001). A Web crawler design for data mining. Journal of Information Science, 27(5), 319-325.

Thelwall, M. (2003). Can Google's PageRank be used to find the most important academic Web pages? Journal of Documentation, 59(2).

Thompson, J. B. (1995). The media and modernity. Cambridge: Polity Press.

Van den Besselaar, P., Tanabe, M., & Ishida, T. (2002). Introduction: Digital cities research and open issues. Digital cities II: Computational and sociological approaches. Lecture notes in computer science 2362. Berlin: Springer.

Voorbij, H. (1999). Searching scientific information on the Internet: A Dutch academic user survey. Journal of the American Society for Information Science, 50(7), 598-615.

Wyatt, S., Thomas, G., & Terranova, T. (2002). They came, they surfed, they went back to the beach: Conceptualizing use and non-use of the Internet. In S. Woolgar (Ed.), Virtual society? Technology, cyberbole, reality (pp. 23-40). Oxford: Oxford University Press.


About the Authors

Paul Wouters is Programme Leader of Nerdi (Networked Research and Digital Information) at the Royal Netherlands Academy of Arts and Sciences. He is a biochemist by training and has a PhD in the sociology of science. Diana Gerbec works in the information industry in Sydney, Australia. She has a Masters Degree in Communication Science, University of Amsterdam. In writing this article, the authors have relied mainly on mediated interaction.
Address: Networked Research and Digital Information (Nerdi), NIWI-KNAW, Joan Muyskenweg 25, Postbus 95110, 1090 HC Amsterdam


©Copyright 2003 Journal of Computer-Mediated Communication