Dismantling the online sphere – Information-ﬁ lters as organizers of public vision

This paper shows how searches for “synthetic biology” on Google.co.uk and Wikipedia.com lead to different demarcations of this controversy in the form of the hyperlink networks that become browsable for the user. When talking about the mediation of biotechnological controversies in the public sphere the interesting divide may accordingly not be between the ofﬂ ine and the online, but rather between logics of ﬁ ltering on the web and the networks of actors that are playing a role in assembling the “web-visions” they give rise to.


1: Introduction
Debates about whether the Internet provides a valuable public sphere have been reoccurring since the mid-90s and they have often centered around diff erences and similarities between the so-called "online public sphere" and its offl ine counterpart. Th e methodological trend in such studies has been to carry out representative samples of the discussion about a given topic in each of the two spheres and use these samples as a basis for discussing whether the online sphere makes room for a more or less direct (Coleman 2005), plural (Gerhards, Schaefer 2010), deliberative (Marlin-Bennett 2011), de-centralized (Benkler 2006, Surowiecki 2004, epistemic correct (Goldman 2008), participatory (O'Reilly 2006) or representative (Papacharissi 2002, Lazer et al. 2009) sphere than the one designated as offl ine. Scholars behind these studies have disagreed on the democratic consequences of the Internet but they have nonetheless agreed on approaching it as an online sphere that can be more or less representatively uncov-ered by the researcher (Lazer et al. 2009, Schneider, Foot 2005, Th elwall 2009. Section 3 of this paper outlines the contours of this debate and indicates that the online/offl ine divide may not be the most fruitful framework for understanding the web as a public arena. It will be argued that conceptualizations that posit the online sphere as a unifi ed phenomenon capable of being compared to other spheres are increasingly challenged by the role which informationfi lters and their surrounding networks play in organizing the visibility of voices around specifi c issues (Halavais 2009). More specifi cally the section uses the concept of "web-visions" (Koed Madsen, 2012) as a way to denote the slice of a controversy that is made visible by an information-fi lter to a temporally and spatially situated web-user. Th e concept is operationalized by following hyperlinks with a departure in the concrete scope of information returned by specifi c fi lters of interest. A "web-vision" is therefore not representative of anything outside the fi lter but it can provide an understanding of the way computers, web-masters, algorithms and web-users play a role in demarcating controversies and how this demarcation diff ers in relation to the used fi lter that serves as the entry-point to the web as well as the time and place of the entry.
Th e empirical part of the paper is an analysis of the way the controversy of synthetic biology has been organized and demarcated by two of the most heavily used information-fi lters in the UK, wikipedia.com and google.co.uk, from January -June 2011. Synthetic biology is the most recent attempt at applying planned engineering on living organisms and section 2 gives an introduction to the choice of this case and the choice Dismantling the online sphere -Information-fi lters as organizers of public vision Anders Koed Madsen PhD Fellow, Department of Organization, Copenhagen Business School of looking at the UK and it also outlines some general thoughts about the democratic importance of sociotechnical controversies. Th e data presented in the paper represents the beginning of a longer longitudinal design and it should therefore be read as an explorative attempt at using the dynamics of these situated "web-visions" to conceptualize the relation between the web and sociotechnical controversies in a way that is diff erent than working from the basis of samples of web-sites as more or less accurate representations of an online sphere.
Section 4 contains the results of the initial quantitative analysis of the web-visions and it shows how google. co.uk gives rise to more fl uid demarcations of the issues of synthetic biology than wikipedia.com and how e.g. artists are visible as voicing social and ethical questions about the technology in the Google-vision. Th is stands in contrast to the more stable and institutionalized demarcation emerging from wikipedia.com but the longitudinal study also shows how the web-visions get more and more similar over time. Section 5 provides a discussion of the fi ndings of the analysis and presents three arguments for the usefulness of working with web-visions instead of starting from pre-defi ned spheres. Section 6 concludes the paper and points towards possibilities for further research.

2: Synthetic biology as a socio-technical controversy
Scholars within the tradition of Science and Technology Studies (STS) have since the 1970s carried out a range of studies showing diff erent ways in which the development of science is always intermeshed with politics in a way that poses hard questions about its role in a democratic society. A few highlights of this literature have been studies illustrating how the development of nuclear radioactivity carried with them discussions about the divide between experts and laymen in decision-making (Wynne, 1992), how the meaning of technology assessment in relation to GMO have to a large extent been infl uenced by local cultures and traditions of law-making (Jasanoff , 2007) and how knowledge-claims in the realm of science are often settled in coalitions of a diverse set of human actors in combination with non-human actors such as laws and scientifi c equipment (Latour, 1987). Irwin and Michael have coined the concept of ethno-epistemic assemblages in order to more precisely express the idea that such "hybrid coalitions" put forward knowledgeclaims in relation to specifi c socio-scientifi c controversies (epistemic) in a way that is situated in local contexts where it is possible to trace back knowledge-claims to local practices (ethno). By calling these ethno-epistemic coalitions assemblages, Irwin and Michael emphasize that they connote a territory of heterogeneous actors that form a whole through their relations to each other and the signs that enable communication between them (Irwin & Michael, 2003). Th e web, which we will return to in the next section, is without doubt an important actor in establishing such signs and relations today.
With the insights from 30 years of STS research in mind it is not surprising that synthetic biology has enrolled a range of diff erent actors in discussions about its development. As the most recent example of genetic engineering of biological organisms follows a list of scientifi c projects that have previously created controversies by framing themselves as providers of answers to large societal questions such as food hunger, climate change and disease prevention. Since the fi rst organisms were genetically engineered in the 1970s there have been immense political and ethical debates between proponents of these biotechnological solutions to societal problems and critics who have deemed such solutions to be unsafe, unethical or just overly optimistic. Synthetic biology has also been launched with great promises that have ignited debates between actors with quite diff erent normative foundations. Proponents of the technology have proposed to discuss its development on the basis of strict scientifi c risk-evaluations and this type of cost-benefi t analysis has to a large extent been taken up by e.g. Th e Presidential Commission for the Study of Bioethical Issues who was recently asked to deliver a report on synthetic biology by Barack Obama (Th e Presidential Commission for the Study of Bioethical Issues 2010). On the critical side we have seen environmental NGOs, such as the ETC group, questioning the possibility of making such risk-evaluations in the face of radical uncertainty and highlighting the potential injustice in the access to potential benefi ts of the technology (Th e ETC Group 2007).
Besides the existence of these varied foundations for the debate there are also considerable diff erences in the ways of defi ning synthetic biology within the community itself and this makes it hard to draw demarcations around the practice. Th e diff erent approaches, however, seem to share the ambition to utilize the technologies of human engineering to optimize evolutionary processes in biological organisms in order to make them perform specifi c desired functions. One of the ways this is done is by decoding the genome of an organism, translating it into digital codes on a computer, recoding it digitally and use that line of code to make a synthetic DNA-structure that can be inserted in an otherwise empty cell. Th is cell will then produce specifi c desired proteins and potentially be part of an organism that can be used to create products such as plants with effi cient photosynthesis, algae-based biofuels and malaria-medicine (Th e Presidential Commission for the Study of Bioethical Issues 2010). Th e J. Craig Venter Institute made headlines in May 2010 by creating the world's fi rst synthetic and self-replicatory bacterial cell and even though its genome is minimal it was by many commentators seen as a landmark of this synthetic biology. Th e fact that a minimal cell can create such a fuzz serves to show that the technology is still in its infancy and that synthetic biology is still a fl exible technology regarding its technical development as well as the way it is interpreted by society.

3: Th e web and synthetic biology
Th e issue of synthetic biology revitalizes democratic questions about the process though which scientifi c technologies are discussed and stabilized. Th e web is one of the central arenas in which this process takes place and the fl exibility in the defi nition and use of synthetic biology makes it an interesting case for analyzing the democratic and epistemic characteristics of the web as a space where the public can encounter diff erent views about science (Weingart, 1998) and where diverse social institutions interact in shaping the social situation around this new technology (Hjarvard, 2008).
Th e importance of the web as such an arena has not escaped scholarly attention and even though the discussion has not been able to solve the disagreement between "optimists" and "pessimists" it has shown some similarities in the way the web has been approached as an object of study by both sides. A tendency has namely been to conceptualize discussions on the web as something that takes place in an online sphere (Gerhards, Schaefer 2010, Al-Saggaf 2006, van Os, Jankowski & Vergeer 2007, in a virtual sphere (Papacharissi 2002) or in cyber-space (Marlin-Bennett 2011). It has furthermore been a common move to compare this sphere with a traditional, pre-digital sphere or, in the most explicit cases, with the "[…] real world in which people can walk around and encounter each other in the fl esh" (Marlin-Bennett 2011, 9).
Th is theoretical move has made it necessary to decide on strategies of sampling that allows for clearly demarcated representations of the spheres in question but it has caused methodological headaches with regard to the online sphere (Lazer et al. 2009) in which it has proved to be far from obvious how to conduct a representative sample of e.g. the discussion about synthetic biology. Th e most well-articulated response to the challenge has been to draw the boundaries of a web-sphere on the basis of a set of web-sites that the researcher knows to be central to the topic of interest and then use them to initiate a snowball-sample through hyperlinks. A "web-sphere" is, on this account, conceptualized as "a set of dynamically defi ned digital resources spanning multiple web-sites deemed relevant or related to a central event, concept or theme, and often connected by hyperlinks" and the web-sphere is supposed to be representative of the discussion online (Schneider, Foot 2005). 1 Th e merit of this way of approaching the Internet as a public sphere is that it allows for structured comparisons between spheres that are clearly demarcated, but the drawback is that it requires the researcher to collapse previously distinct entities into unifi ed wholes. Th is is e.g. the case when political web-sites in EU-countries are collapsed into a representation of the "online European public sphere" (van Os, Jankowski & Vergeer 2007) or when the search result pages of Google, Yahoo and Fireball are collapsed into a representation of the "internet-based public sphere" (Gerhards, Schaefer 2010). Th e price for being able to do broad comparisons between the offl ine and the online is the reduction of complexity and distinction within these spheres.

4: From spheres to web-visions
An alternative to the focus on the representation of spheres is to take departure in the idea that every image of a controversy is a unique "socio-technical setup" produced by a specifi c network of actors through specifi c media with specifi c logics of fi ltering. Th is is as true for the image that ends up as the outcome of a traditional offl ine media of representation, such as the city-hall meeting, as it is for the image that ends up as the result of a search on Google. (Girard, Stark 2007). Th e way public offi cials facilitate a citizen hearing with the use of technologies such as a strictly planned physical space, microphones and plans for speaking and voting makes controversies visible and managable in a way that is neither more or less performative and mediated than the way Google´s algorithm makes a controversy visible. Both images are produced in a situation with specifi c constraints on the social interaction (Hjarvard, 2008) Th e interesting insights, therefore, do not lie in comparisons of how well they map pre-defi ned spheres but in looking at them as two distinct socio-technical set-ups that assemble and organize public experience in diff erent ways. Th is situated approach to depicting controversies also seems more in line with the argument behind the concept of "ethno-epistemic assemblages", which prompts us to take our analytical point of departure in a situated setting. If knowledge-claims about synthetic biology are situated in local contexts it may be a fruitful move to start from the way the web looks through a specifi c fi lter, at a specifi c place at a specifi c point in time.
Th e move of beginning a media-analysis on the basis of the logic of the fi lter instead of departing from an a priori divide between diff erent spheres also resonates with work done under the heading of mediatization-theory.
With departure in the concept of "media-logics" scholars within this tradition have emphasized how the form of a given media infl uences the distribution of material and symbolic ressouces in a way that control the categorization, selection, circulaton and presentation of knowledge (Hjarvard, 2008). Th is focus on how formal and informal rules shape representations of knowledge is carried on in the study below but it is important to emphasize that these representations are not just a result of the logic of fi ltering and the form of the medium. Because both Google and Wikipedia rely on user-generated content their representations of synthetic biology is just as much an outcome of the behavior of all the actors that produce and rate this content. It is an assemblage of the logics of fi ltering, the behavior of web-users, the choices made by web-designers and the words used by the involved organizations to describe synthetic biology. Th e interaction between these human and non-human actors is what eventually leads to the situation where a specifi c scope of the dicsussion about synthetic biology becomes visible.
Th is scope is situated in time and space and the paper will conceptualize it as the "web-vision" that the fi lter provides the user with. It is defi ned as follows: A "web-vision" is the specifi c actors, themes and documents that become visible to the user when querying the web through a specifi c information-fi lter at a specifi c time in a specifi c place.
Th is focus on spatial and temporal groundedness goes against popular conceptions of the web as a place where time and space is annihilated and this seems as a natural consequence of breaking with conceptions such as the virtual or cyberspace as something diff erent from the offl ine. If these are not interesting divides it follows that we cannot think of the web as being exempt from time and space that are some of the most central conditions for any other part of our cognition.

Operationalizing the "web-visions" of a British user
In order to situate the empirical research of the paper in a spatial and temporal reality it was decided to follow the "web-visions" of entry-points to the web that are heavily used in the UK. Th e geographical situation in the UK was chosen for two reasons. Th e fi rst is that the UK has a quite unique tradition in relation to biotechnological risk-assessment due to the fact that the BSE scandal in the 1990s made many of the abovementioned questions about science and democracy surface in the region. Th is led to a focus on the relevance of broad debates on scientifi c developments and analyzing web-visions with departure in the UK makes it possible to determine the extent to which this focus is transferred to the web as it is seen by the UK public. Th e second reason is that the UK shares semantics with the United States in relation to the word used to denote the scientifi c project we are interested in. A search for "synthetic biology", therefore, makes it possible to determine the extent to which the British public is infl uenced by American framings. 2 Th e fi lters to follow were chosen on the basis of a search on Alexa.com, which revealed wikipedia.com and google.co.uk to be among the most heavily used entry-points to the web by British users. While inquiring these fi lters about the issue of synthetic biology it was ensured that all the searches were conducted from the same computer with a constant IP-address based in London and that the Google-searches were de-personalized by adding "&pws=0" after the search URL. In that way the search-results represents the way Google makes the controversy visible before the (minor) 3 infl uence of browser history kicks in.
Besides being heavily used the two fi lters were chosen because they rely on diff erent forms of crowd-sourcing and therefore serve the potential of producing diff erent web-visions around synthetic biology. Google originally created its position on the market of search by harnessing the words and hyperlinks that people constantly leave on the web and it used these traces to statistically calculate a rank of relevance for any web-site accessible through their interface (Brin, Page 1998). Th is approach to information-fi ltering and relevance was launched in opposition to e.g. Yahoo that relied on human-based classifi cation of web-sites in the tradition of library indexes. Google can in that sense be said to utilize the "crowd-wisdom" (Sunstein 2006) of its users by letting e.g. hyperlinks count as votes for web-sites. It is this crowd-logic, rather than the fact that it plays out in an online realm, that makes it an interesting case to follow. Moving to Wikipedia it has, in the same sense, taken advantage of the fact that users of the web are becoming producers of web-content by enabling a transparent system of collaborative fi ltering of its articles. Everybody can edit an article on the encyclopedia that works on the basis of a post-fi ltering philosophy. Th is means that a contribution to an article immediately appears on the site without any form of editorial oversight. Instead of pre-defi ned fi lters, Wikipedia has an internal hierarchy of moderators and users that constantly overlooks articles and removes knowledgevandalism (Bruns 2008) in a way that seem to be eff ective in correcting errors (Fallis 2008). Th e two fi lters are, accordingly, harnessing the intelligence of web-users on the basis of diff erent philosophies of fi ltering and what makes them possible to compare is that they both function as hubs for sending the user further into the web. It is true that the page of search-results of Google and the article on Wikipedia are quite diff erent but a central part of both of them is to decide on the relevance of external links to guide their users further into the web.
It is these links that form the basis of the operationalization of the "web-vision" that the fi lter gives rise to. In the case of Google the web-visions are simply made by following the links of the top 20 URLs returned in the search-result and in Wikipedia they were made by following the external links in the bottom of the article. Th e links were followed with the help of the Issue Crawler 4 that set to follow two layers of links as well as to perform a "co-link analysis", which in the language of the Issue Crawler means that only pages with two in-links are kept in the visualization. Th is was done in order to reduce the "web-visions" by restricting them to include only sites that are deemed relevant by at least two other web-sites and thereby reduce the risk that the visuali-zations would drift away from the issue of synthetic biology. Th e raw data that the Issue-Crawler returns is a matrix illustrating which web-sites that are linking to each other and this data was directly exported into UCI-net which allowed for subsequent manipulation of the networks returned from the crawler. Th is manipulation included the deletion of all web-sites that did not have at least one mention of the word "synthetic biology" throughout its pages as well as "irrelevant" links such as the ones that almost all web-sites forge to e.g. the licenses of Creative Common and Flash-players. 5 Finally the visions were made interpretable and useful for the study by organizing the networks on the basis of statistical calculations of the distances between nodes and by coloring, shaping and sizing the nodes in the network on the basis of the parameters in the table below.
Th e shaping and sizing of the nodes in the "webvisions" is a deductive element in the sense that the parameters are based on already established theoretical expectations of what a controversy is. Th e six web-visions below are accordingly a construct on which the researcher has a huge infl uence. No web-users will en-  counter these visualizations when browsing the web for information about synthetic biology and they are only to be seen as heuristic images of the scope of the controversy that the chosen fi lters make visible to a generalized UK web-user. Th ey are prompts for discussing what lies "behind" the URL-lists that the fi lters return.
Th e next section will outline some of the most interesting fi ndings in this initial stage of the study and thereby provide an insight into the dynamics with which the two fi lters demarcate the controversy of synthetic biology.

5: Similarities and diff erences in the visions
Th e temporal make-up of the public space is important and we will therefore start by looking at the extent to which the visions are stable or fl uid by comparing the way they make actors visible over time. Th e sizing of the nodes in the web-visions is supplemented with a calculation of a "fl uidity-rate" that indicates changes in the visions of a fi lter from month to month and a "longitudinal stabilityrate" indicating the level of stability in the visions across the whole period. Th ey are operationalized according to the formulas below and the boxes compare the visions on these concepts.

Fluidity-rate = (number of new web-sites in the web-vision prior to previous month/total amount of websites in the given month) x 100
Logitudinal stability-factor = ((websites common to all visions of a single devise across y months/avarage amount of web-sites in the visions across y months) x 100). From the two boxes it is evident that the Google-visions are way more fl uid and changing than the Wikipediavisions and this serves as an indication that the collaborative and consensus-oriented logic of fi ltering behind Wikipedia and the type of actors that makes up this as-    semblage results in a slower updating-pattern than the crowd-based algorithm of Google and the actors that are active here. Besides that it indicates that the Wikipediavision is cumulative in the sense that the number of links are increasing over time. Few web-sites drop out and this is also indicated by the fact that the longitudinal stabilityrate is high. In order to better understand the characteristics of these diff erences in temporality it is, however, necessary to look at the distribution of web-sites in the diff erent visions over time with reference to the spatial parameters listed above.
Looking at the graphs above we can see that the high fl uidity factor in the Google-visions is generally translated into less straight lines compared to the Wikipedia-visions where the lines are quite stable. Th is is especially evident in relation to the geographical distribution of web-sites, which is changing dramatically in the Google-visions in the time-span covered. Th e US actors are on the rise and a comparison of January and June reveals that a tightly connected UK-cluster consisting of e.g. BBSRC, Th e Natural Environmental Research Council, Th e Welcome Trust and Th e Royal Society have disappeared from the vision in during this time-span (the cluster is marked with a red circle in the January vision above). Th e fact that there seems to be a distinction between this UK cluster and the US cluster to the right of it makes it even more interesting that the fl uidity in the Google-vision to a large extent covers an Americanization. A subsequent qualitative analysis must inquire into whether or not there are radically different narratives about synthetic biology in the disappearing cluster and the US cluster. Another fi nding is that the actors that are positioning themselves as being global voices are to a large extent connected to the US cluster. When you search for an organization that presents itself as having a global perspective on synthetic biology you will, accordingly, quickly fi nd yourself browsing the websites of American organizations. Along with this Americanization, the fl uidity in the Google-vision also covers a shift in the types of websites that are made visible and it is e specially news actors are on the rise whereas natural science actors that are in decline. Th e emerging news-sites are to a large extent US-based and even UK-based sites, such as Th e Guardian, are running stories on American project on synthetic biology.
Th e characteristics of the fl uidity in the Google-vision is, however, not fully understandable by looking at the graphs above. Figure 9 e.g. indicates that not much have happened regarding the red nodes in the Google-visions because they make up the same share of the network in April and June. But looking closer at the visions we can see that there is a change in this category as well. Even through the percentage of this type of web-sites is similar in April and June we can see that the web-sites that represent this group are becoming more institutionalized as well as less UK-oriented over time. UK-based designers such as Daisy Ginsberg and James King have e.g. dropped out of the vision. Th eir project is to illustrate social issues around synthetic biology through art and we see that their visibility is substituted by the visibility of Th e Hastings Center, Th e ETC Group and Biopolitical Times that are all more institutionalized American NGOs (all of these are marked with black circles on the visions above).
When looking at the fl uidity of Google from the perspective of the Wikipedia-visions we can see that the June-vision of the fi lter is similar to the January version of the Wikipedia-vision which haven't changed much from January to June. Th is is true when we look at the graphs in Figures 5-8 as well as when we look at the more detailed network-visualizations in Figures 1-4. Th e Google-visions had a unique scope of visibility in the fi rst months with more UK web-sites and a less institutionalized set of social actors being visible, but the diff erent logics of fi ltering seem to be generating increasingly similar visions. Th is may be due to the fact that Th e Presidential Commission for the Study of Bioethical Issues released a report on synthetic biology in late December and this report is becoming visible in the Google.co.uk-visions in April where it is linked to by a group of centrally positioned actors. Th ese links are not visible in the June-visions of Google.co.uk, but a range of American institutions have become visible instead. Looking at the Wikipedia-visions we see that these institutions have been visible from the outset and this is one of the signifi cant examples of the increasing similarity across the two fi lters (these institutions are marked with black squares in all visions above). If the publishing of the American report is the explanation for the increasing similarity it serves to show that the Google-vision is easily infl uenced by such big events and that they have the power to trump the visibility of the actors that made these visions stand out from the Wikipedia-visions in January. Th e Wikipedia-visions, on the other hand, has a high "logitudinal stability-rate" and the visions are not altered dramatically by the links forged to the Obama-report. Possibly because they fi t the already institutionalized slant of the vision.

Making things public?
Th ese quantitative analyses represent the beginning of longitudinal study but they already serve to indicate answers to the important question of what visualizations of web-visions could potentially be useful for. One possible answer is that they make the structure of controversies public in a way that allows people to better understand and navigate these controversies. But we need to remem-ber that the web is not the world and accept that if visualizations are to serve the function of a good public map they need to be constructed in a way such that many more layers of information is added to the digital traces of one fi lter (Venturini, 2010). It would therefore be too ambitious to present the "web-visions" above as tools for navigating the controversy of synthetic biology. Quite to the contrary they are constructed by following specifi c fi lters in a specifi c country at a specifi c time and they are even removed from our everyday experience of the web through the use of the issue-crawler. Th ey are, in other words, a construct make by several actors -including the researcher.
But the lack of representation is at the same time the roots of their usefulness in at least three ways. From the perspective of this paper they are, fi rst of all, useful because they highlight the fact that it is fi ltering logics and the assemblages they are part of, rather than ontological distinctions between the offl ine and the online, that is important when we try to understand the dynamics of controversies and their mediation. And even though they lack the quality of representation they may still provide insights into the dynamics of controversies. Working on the basis of case-study logic one could e.g. argue that the Wikipedia-visions are "least likely" to change radically and that rapid fl uctuations in this vision would indicate a very dynamic controversy whereas changes in the Google UK visions are less remarkable. Th is just serves to indicate that representativity and objectivity is not the only benchmark of good visualizations and deeper discussions of the methodological underpinnings of controversy-visualizations can be found in another paper (Koed Madsen, 2012).
Secondly, the fact that the visions are situated in a UK-context makes them interesting cases for studying the development of biotechnological discussions in the specifi c fi lters that are used by the public in a country that has a unique history regarding such discussions. Th e fact that the vision of these fi lters seem to be Americanized and institutionalized after the release of the Obamareport raises interesting questions about the role digital medias play in the UK. Is it e.g. a problem for the UK public that they share language with Th e United States when it comes to searches? Is it a problem that the UK does not have a national Wikipedia like other countries? In order to probe further into such questions it could e.g. be interesting to compare the UK fi ndings with webvisions that are situated in other countries.
Th e third answer to the question of relevance is broader and it points to the fact that these visions are assemblages that are infl uenced by a range of diff erent phenomena such as algorithms and information-fi lters, cultures of relevance, cultures of media use and -most importantly -digital traces left by the organizations involved in the controversy of synthetic biology. For a researcher they represent a way to visualize and simplify controversies in a way that takes departure in the current cultures of communication and the vocabulary of the actors engaged in the controversies (Latour 2007). Th is makes them quite diff erent from other simplifi cations of controversies such as surveys and city hall meetings. Th e dynamics behind their creation makes potential diff erences interesting as refl ections on e.g. media-logics and organizational networks. Compared to an analysis of e.g. the search-page that meets the everyday web-user on Google, these visions make explicit the networked nature of infl uential web-based fi lters.

6: Conclusion and further research
Th e empirical analysis above have presented a longitudinal comparison of web-visions created by two infl uential web-fi lters that sort information according to quite different logics of relevance. Th e analysis of these visions serves as an empirical example of the way the web is better understood as a conglomerate of fi lters that make visions available to its users rather than as a unifi ed sphere that can be representatively uncovered. Th e comparison between the visions produced by google.co.uk and wikipedia.com revealed diff erences in the temporality of the visions. Th e Google-visions proved to be altered signifi cantly over time whereas the Wikipedia-visions remained stable. Th e analysis showed that the fl uidity in the Google-visions was correlated with rising discussions about a report released by Th e Presidential Commission for the Study of Bioethical Issues in the USA and that e.g. the visibility of artists posing social questions about the technology were declining. Th e discussion following the analysis provided three arguments for the usefulness of "web-vision" analysis that hinted at the potential of using them to say something more general about the world by constructing the visions on the basis of a case-study logic rather than on the logic of sampling spheres, their potential to help analysis of geographically and temporally situated developments of the web as a public arena and their potential to provide insights in to the assemblages and dynamics behind the simplifi ed outcomes of the fi lters.
Th e results represent the beginning of a longitudinal study and further research in the area needs to see the quantitative studies in this paper being supplemented with qualitative studies of the themes and connections between the actors in order to generate a deeper understanding of the diff erences between the demarcations that diff erent fi lters give rise to. Th ey also need to be supplemented with searches for other issues in order to see Noter 1 A way of softening the divide between the offl ine and the online as well as putting less emphasis on a priori selected actors has been to distinguish between spheres within the confi nes of the Internet, such as the blogosphere and the news-sphere and let "ordering devises" such as BlogPulse and Google News draw the boundaries around these spheres (Rogers 2009). Th is is a way of letting the "logics of the web" draw the boundaries instead of snow-balling from a priori known actors. It grounds the representation in online dynamics, but it retains the ambition of uncovering a sphere. 2 Th e keyword searched for ended up being "synthetic biology" because searches on Google Trends, Blog-Pulse and the history of Wikipedia proved that there is much more web-activity around this key-word than around similar words such as "synthetic life", "constructive biology" or the abbreviation "synbio".
if some of the fi ndings above have a more general nature and with words that are related to synthetic biology in order to test the power of the search-word used. Would searches for synthetic life e.g. provide radically diff erent visions? Finally it would strengthen our understanding of the web as a public arena if we were to conduct similar studies on search engines specifi cally oriented towards blogs. Th is would also require analyses of why people are leaving digital traces in diff erent media-spaces and thereby give the user a larger role in the analysis. Besides these shortcomings the paper has hopefully given a sense of the kind of insights that a web-vision analysis can provide as to how the web is active in creating "visions" of controversies.