2.1.1 A Wide Range of Contacting Tools

Nature Network presented several communication instruments at different levels. At the group level, the most relevant were Forums, Groups and Questions & Answers because they were meeting points where users could interact with each other and share information. Other services such as Blogs and the Workbench were addressed to the creativity of the users and their ability to spread information and contents. Otherwise, the Hub was the main instrument for connecting the network to real life though events and meetings.

2.1.1.1 Forums

Nature Network provided tools for establishing specialized forums where users could discuss issues and controversies on scientific advances. It contained 1,822 forums where 16,678 discussion topics were laid out. It is interesting to observe that three of the five most populated forums were promoted directly by the creators of the site themselves (i.e. Nature India, Ask the Nature Editor and Careers Advice by NatureJobs), which provides evidence of little incentive among this community to create its own active and dynamic forums. In general, the activity in these forums was quite low, with less than one (0.3) message per person and a participation rate of only 2.5 per cent of the Nature Network community (see Table 2.1).

Table 2.1

The five most active forums in Nature Network

Name	Members	Topics	Replies	Messages	Activity
Brain Physiology, Cognition and Consciousness	606	229	3,542	3,771	6.2
Nature India	1,386	1,336	2,296	3,632	2.6
Ask the Nature Editor	797	991	501	1,492	1.9
PhD Students	1,125	848	457	1,305	1.2
Careers Advice by NatureJobs	771	930	302	1,232	1.6
Total	27,012	16,678	16,909	24,621	0.9
Average	15.6	4.5	9.9	14.4	0.3

Figure 2.1 shows the number of forums that posted their last message by year. In general, the distribution describes low activity, with a large proportion (66 per cent) of forums without messages two years before the closure. The linear trend of the cumulative distribution and the constant number of forums by year confirm that this application remained inactive.

2.1.2 The Natural Community

This last section is possibly the most important because it concerns the creation of a personal sketch in order to be able to participate in forums and groups. The main page talks of 25,000 colleagues but in December 2014 a crawling of the site estimated more than 1.3 million members. A sample of more than 800,000 profiles was obtained from that crawling, acquiring data on disciplines, organizations, countries, etc. This sample helps us to describe what kind of users shaped the Nature Network community. The first thing that attracts the attention was that most of the profiles do not include any information and only 32.1 per cent of the users filled the profiles with some data. For example, discipline is the section most filled out with 29.7 per cent, while group information is only presented in 1.6 per cent of profiles, affiliation in just 0.92 per cent, and sex and age in barely 0.5 per cent. This scant attention to filling out a personal profile is an example of the poor commitment of the users the site and it could be considered a qualitative measure of the success of these services.

In any case, although this information could be not representative of the 1.3 million Nature Network users, we could start from the hypothesis that users that include some information in their profiles would be motivated by a closer involvement with the site and a more active participation. In this way, data on disciplines, sex or age could describe the profiles of the most active users in Nature Network.

Table 2.3 shows the most frequent research disciplines in the Nature Network’s profiles. Nature Publishing Group uses its own subject matter classification with which it arranges the content of its products. It is surprising that Business/Investment is the most frequent class with more than the 93 per cent of profiles. This could be due to some failure of assignation of that discipline and it is possible that it was added automatically by error. If that discipline is put to one side, Other (2.56 per cent) and Biology (1.65 per cent) are the disciplines that have the largest proportion of users in the system. Apart from Business/Investment, the distribution does not show any thematic bias and fits the common subject-matter distribution of large bibliographic databases (e.g. Web of Sciences, Scopus), although the elevated presence of Other could be a symptom of mis-classification or that the research activity of the users did not fit with that classification scheme.

By gender, Nature Network’s population presents a high percentage of men (67.9 per cent) in contrast to the 32 per cent of women. This percentage is similar (30 per cent) to the percentage of women in science (UNESCO Institute for Statistics, 2015) which demonstrates that the participation in this network is the same for men and women. With regard to age, the average of age is 32.8 years old, which suggests that most of the users are young scholars at the start of their academic career.

Table 2.4 includes the five organizations with more profiles on the site. Universities prevail in the ranking, with Harvard University (2.64 per cent), Imperial College London (1.39 per cent) and Columbia University (1.38 per cent) outstanding. It should be pointed out that two of the universities with the most profiles come from United Kingdom, which evidences the strong presence of British users in the network. It is also surprising that the fifth organization by number of profiles is the Nature Publishing Group (1.14 per cent). This could be evidence of the strong support of the owner company for this platform, although it would also report an excessive intervention of this publishing group in energising the network.

Table 2.5 ranks the ten countries with the most users signed into Nature Network. The United States (32.2 per cent) and the United Kingdom (15.2 per cent) are the countries with the largest presence, followed by India (11.4 per cent) and Germany (3.7 per cent). On examination of the penetration index, the United Kingdom (4.1 per cent) and India (4 per cent) are the countries where this platform yielded the most success. This penetration is not surprising in the United Kingdom because the head office of the Nature Publishing Group is located in London (NPG, 2015). Perhaps more remarkable is the penetration in India, which could de due to influence of United Kingdom and the Anglo world as other commonwealth countries such as Australia (0.15 per cent) and Canada (0.13 per cent) show high penetrations as well. In any case, these results are not entirely representative of the total population in Nature Network and so may only be considered from an illustrative point of view. According to its distribution across countries, Country Spreading (CS) index shows that 74 per cent of users belong to the first ten countries, which suggests that the platform did not go beyond the British environment.

2.1.3 The Chatting Room

Nature Network was one of the first online social networks for scientists that permitted the building of personal and complete profiles as well as incorporating two fundamental communication tools (Forums and Groups) that made possible the interaction between users. These communication instruments can be considered group tools, that is applications that promote the interaction among users at the same time and in the same place. However, although these instruments were created with the purpose of building a large and cohesive community, the reality was that most of the users were not so much interested in those functions. Results from the crawler verified this fact. The participation degree in these services was quite low, with only 2.5 per cent of users involved in Forums and 0.2 per cent in Groups. If the activity inside these services is observed, data show poorer values with an average activity of 0.3 posted messages in Forums and 1.5 in Groups. The other instrument, the Blog service, only produced 25 logbooks. These figures are clear symptoms that the network did not come up to the expectations of the users and could provide the most decisive proof of the failure of this site. It is possible to reinforce this claim because, as a result of the low amount of information used to describe the profiles, only 32.1 per cent of them included any data.

One other cause that would explain the closing of the site was that the presence of the Nature Publishing Group in the network was evident at all times. Most of the active forums and groups were created and propelled by Nature staff which suggests that the company did indeed exercise a strong intervention in the network. Since most of the blogs were also published by members of Nature, a monitoring policy of the site was instituted aimed at fostering a community of loyal ‘customers’ interested in Nature products. However, it is also possible that this excessive intervention was caused by the low participation of its members, and in consequence the company was looking to somehow keep the site active. Perhaps the last version introduced in 2010 incorporating the Workbench and Q&As could be understood as one final effort to stimulate the service. But the inactivity of the Forums and Groups sections and the scant information in the profiles decided in advance the eventual fate of the site.

In general terms, the Nature Network was the first initiative to pave the way to academic social sites with the sharing of information tools and personal profiles. However, it failed to create a comfortable environment that was attractive to the academic community. The model was closer to a chat room, in which direct communication between users in a public environment took precedence. Nature thus understood that a social network is just a public forum where research topics are discussed, recent publications are commented on and scientific news is spread. Nevertheless, any consistent virtual community should be based on the ability to produce and generate content which could then be shared, favouring networking and collaboration between its members.

2.2.1 Scientist’s Directory

Unlike other social services, BiomedExperts opted for automatically creating profiles as the starting point, thus counting on an initial population of profiles that gave consistency to the project. A key point in the success of an online social network is to have an important critical mass which allows its users to interact. As a specialized network in biomedicine, it started by creating 1.8 million ‘knowledge profiles’ from authors listed in about 18 million articles accessible through the PubMed database (BiomedExperts.com, 2011), the most important database on medicine. Only articles published during the previous ten years were selected to create profiles of active authors at that time. The authors of these profiles were then invited to participate in the network, edit their profiles more deeply and create new ties between their partners. However, BiomedExperts put up a number of restrictions when it came to creating a profile. Only users with a paper indexed in Pubmed within the previous ten years could create a full personal profile. Other users could simply browse the network of experts, but without any type of participation (BiomedExperts.com, 2010). This meant that the range of possible users was limited and most of them could only glance at the database.

The automatic creation of profiles of the authors of papers on the one hand produced duplicated profiles (i.e. distinctly different profiles corresponding to the same person) and on the other resulted in the merging of different authors with similar names. This is because researchers can author a research paper using different variations of their name. This problem is usual in academic bibliographic databases (e.g. Scopus, Microsoft Academic Search) and provokes inconsistency and noise (Ortega, 2014). In the case of social networks, it can cause mistrust and low performance. In BiomedExperts, each profile was disambiguated by identifying the ‘fingerprints’ of each author in keywords, places of work and age (Oswald, 2009) when it could not help but find duplicate and erroneous profiles.

However, in October 2014 there were only 473,000 profiles ‘validated’ by their respective authors (26.1 per cent), which informs us that interest from the research community was not very enthusiastic because only a quarter of the profiles were validated during the six years it had been running. To build the network, profiles were connected through bibliographical information, co-authors and research interests (Regazzi, 2013). In this way, any user could browse the network, jumping from one profile to another through any element the profiles had in common.

2.2.2 Interaction Tools

BiomedExperts, as a specialized directory of scientists, did not display any collaboration tools. Instead, it offered only a reduced range of communication tools. These could be divided into contacting devices at the profile level and instruments to assess articles. Thus each profile could only send internal messages to other members, bookmark researchers and add contacts to follow their updates and new publications. This reduced range of contacting tools limited the possibilities of user interaction, reinforcing the assumption that BiomedExperts was a scientific directory in which it was only possible to locate experts in specific fields and identify potential collaborators. In addition to this interaction function, there were other external tools that connected the network with other social sites. Thus it was possible to recommend a paper, profile or any other element through Twitter, Facebook, Connotea, etc.

At the publication level, it was possible to assess a paper (Congratulate), add it to a reading list or recommend it to other users. These mechanisms enabled the quality of a research paper to be evaluated, introducing impact elements into the publication list of a user. However, these instruments were limited to the publication list of a profile which could only gather articles from Pubmed.

2.2.3 Structure

As seen elsewhere, a user could only search partners and manage his or her own profile. The structure of the site was rather simple – just a display of a profile page from where other colleagues may be found and information on the user may be added. There were eight sections from the viewpoint of each user:

2.2.4 Pre-elaborated Profiles

As said above, at the moment of closure it was claimed there were 473,000 profiles, far from the 1.8 million automatically pre-elaborated profiles. Using the WayBack Machine of Archive.org, it was possible to track the evolution of the number of validated profiles since 2010. Unfortunately, previous years did not offer information on this element.

Figure 2.2 presents the evolution in number of validated profiles for BiomedExperts since 2010, the first year in which that information is displayed in the home page, counting 258,000 users already registered. The distribution clearly describes a pattern of descent from 2011, the moment at which most users claimed their profiles. The annual growth rate is the lowest of the social sites analysed with an increase of 10.6 per cent each year. These results make clear the lack of success enjoyed by this platform with a constant reduction in new active users from 2011, three years following its creation.

Each profile (Researcher Overview) included a name and research topics (Figure 2.3). The only quantitative data were the number of publications and the members included in their personal network, detailing contacts at the first (Co-authors), second and third levels. Next, it presented the only networking functionalities on this site: send a message, add a contact and bookmark a profile. Below, there were six elements that gathered together the academic information on the profile:

2.2.5 UniPHY: The Physics Sequel

Following in footsteps of BiomedExperts, in 2009 Collexis launched a similar product specializing in Physics for the American Physics Institute (AIP). UniPHY reproduced the same architecture as its big brother, creating a pre-elaborated population of 180,000 (300,000 in 2011) profiles from articles and papers published in 100 leading physics journals from Searchable Physics Information Notices (SPIN) in the previous 30 years as well as approximately 100,000 papers from scientific conferences (Seybold, 2009). However, its life was more ephemeral than its predecessor and it was closed in 2012.

2.2.6 A Static Network of Automatic Profiles

BiomedExperts and its younger brother UniPHY presented a particular strategy for creating an online community based on automatic profiles that were later claimed by their authors. In this way, Collexis could quickly generate specialized networks of researchers extracting authors from a large repository of scientific articles. These products were improved by means of a good semantic classification system that categorized each profile through relevant keywords. But, above all, the success of these social sites was due to their impressive visualization which became the key instrument for exploring profiles in the system.

However, while the collaborative relationships expressed in scholarly papers were the ties to build those networks, they were not the links that profiles established with other colleagues using contacting tools. BiomedExperts and UniPHY could not be considered online social networks because their architecture was not supported by the interaction of their users, but in the co-authorship network shaped outside the website. Although it is true that these platforms included contacting tools (adding a contact, bookmarking a researcher, etc.), these instruments were oriented to establishing further contacts through other media, not in the platform itself. In this form, Collexis’ products were merely expert databases focused only on browsing and locating scholars, instead of platforms where the knowledge and expertise could be shared by the online community in the same place. This model produced passive profiles that were only used to show the research production of their users, more so if we take into account that two-thirds of the profiles were not edited by their authors. This model ended up as a static yearbook where the users were considered passive subjects that only showed their papers so they could be selected as partners.

A further factor that provided evidence that these products were not oriented to social networking was that the population of profiles was not only limited to disciplines, but to specific bibliographic sources such Pubmed in BiomedExperts and SPIN in UniPHY. This restrictive policy blocked the growth with new profiles and limited the publications that users could offer. To this may be added that these platforms were designed as isolated and specialized islands that did not represent the interdisciplinary nature of science. Perhaps this would explain why only one-third of their profiles were claimed in BiomedExperts.

In summary, the cause of the collapse of the Collexis model could be explained by a wrong understanding of the academic social networks, where users need a medium to share things, going beyond contact among them. This ignorance is also reflected in the restrictive conditions surrounding opening a profile, which meant a large proportion of researchers were not able to participate in the site.