In my last blog post, I proposed one reason why young researchers don’t use social media as much as many people expected they might. In a nutshell, postdocs and other early career academics are extremely focused on boosting the performance metrics that they believe will get them to the next stage in their careers, and their only source of guidance on what those outputs are, are often their supervisors, who are sometimes unfamiliar with the potential benefits of social media.

Several publishers and companies in the scholarly communication space have experimented with the social web and it’s fair to say that results have been mixed. There have been a number of unsuccessful attempts to create networks, like Nature’s ill-fated Connotea site that discontinued service in March of last year. Other publisher led projects are still around, but generally as part of larger efforts. Frontiers, for example, recently won the ALPSP innovation award with a combination of its journal for kids, post publication peer-review and author level metrics. While Frontiers is committed to being a social network, their success so far has come from their publishing work. Then there are the disruptive companies like Academia.edu, ResearchGate, and even Mendeley, who all boast high numbers of users, but whose popularity likely stems from their use as a file-sharing platform, rather than as a social network. In all of these cases, a visit to the site turns up plenty stub or half-filled profiles, with few user pages kept up to date. None of the social networks for academics have yet to achieve the kind of adoption that Facebook has managed in people’s personal lives or LinkedIn has done for business.

Facebook for Scientists?

When publishers talk about building a social network for scientists, they often think about something that looks like Facebook when it was first launched to the public; a series of profile pages filled out by individuals, a bit like an online CV with links to published articles. If we look at social media trends outside of science, however, a profile driven experience is not where the growth and excitement are (even Facebook know that the newsfeed is the feature that users find compelling). Twitter does have profile pages, but they’re not the focus of the application, instead, the experience is driven by tweets and hashtags. Reddit, Pinterest, Instagram, and even YouTube are all content driven with little attention paid to user profiles. Contributors upload content as they generate it, and online personas are the sum of those contributions.

Some people might argue that taking the focus away from the individual in this way will remove the incentive to contribute, but if we look at content driven networks, we see that this isn’t the case. Consumers browse based on subject, conversation or content type, but over time begin to filter by user as they learn who’s putting the best stuff on the network. As Chris Anderson explained in his TED talk on crowd driven innovation, this creates a dynamic, sometimes competitive environment that keeps people contributing and consuming.

Extensive user profiles are time consuming to create and not dynamic, making them high investment, low return exercises for contributors and consumers alike. Instead, content should form a series of nodes, with the contributor being one of several connectors. A successful scientific social network will need to embrace this aspect of how people want to interact online and put the data, protocols, articles, filesets and other content at the center of the network.

So the Answer is Blogs?

When people talk about successful social media use in science, they often talk about blogs. The trouble with blogging though, is that it’s hard work and time-consuming. I know this from personal experience as a contributor. Given the fact that social media experts say that to have an impact, you have to publish on a regular basis, it’s easy to see why many people shy away from the idea due to lack of time.

To get wholesale buy-in from busy researchers, a scientific social network will have to rely less on contributor-created narratives and enable users to quickly and easily upload content to an interconnected social web that automatically provides the context by cross-linking it to other related content via meta-data. To give an example of how that can work, a single 140 character tweet is quick to type but doesn’t mean much on it’s own, but a thread of tweets associated with a conference hashtag gives a remote follower a multi-viewpoint picture of an entire plenary.

The Scientific Social Network of the Future

It’s easy to point out the difficulties in making science social on the web and tempting to just throw up our hands and say that we’ve experimented with social and it doesn’t work for science. A minority of researchers, however, are using a combination of blogs and twitter to boost impact, find collaborators, and even win grants. The challenge for publishers is to find ways to scale those uses of social media to make them accessible and valuable to all.

Here are my predictions for what the breakthrough scientific social network will look like:

1. It will be easy to get started. There will be no long, complicated profiles and certainly no nagging pie chart or reminder of how complete your profile isn’t. This is off-putting for contributors because they’re being asked to put in a lot of time and effort before they know if a platform is going to work for them.
2. It will be content oriented. Just like Reddit, Instagram or Pinterest, a contributor’s persona will be made up of the aggregated content that they have submitted. Good content will be read, downloaded and used, allowing contributors to build a reputation based on their research outputs.
3. Meta-data will provide context. There’s no need for contributors to re-write the introduction of their article while trying to avoid plagiarizing themselves. For example, In order to put a data set in context, linking to the abstract will do. Other connectors might include subject area, object type, institution, funder, or any number of helpful parameters.