Anyone building an online health community or a health app dreams of scale, of going viral, of making a shedload of money. No matter that 85% of apps aren’t opened after the first six weeks or that there have been at least 42 different platforms for rating residential care homes. The dreams of the digirati are fed by the existence of a very few winner-takes-all apex Unicorns and by the normative behaviour of digital entrepreneurs and investors that flows from their success.

When we began this research we too assumed that online communities could and should scale to be massive and involve millions. But what we learnt from interviewees and from surveying progress across the field was that for online health communities scale is a bug not a feature.

An expectation of winner takes all

Whether it is Google and Facebook, The ALS Ice Bucket Challenge or MoveOn.org you do not have to look far to see how strong network effects can lead to a winner takes all situation in the digital realm. This leads to the idea that “What we need is Facebook / TripAdvisor for health”, which we heard as a much criticised but common refrain across our interviewees. Due to a lack of sophistication around technology developments and usage, social network theory, sustainable user and business models, to name but a few, there is a general rush to make sense of the world and find the “right” solution in the field of digital health. More often than not this means people seeking to emulate ‘successful’ models from other areas of the digital world: the big platforms and ratings systems that people see as ‘killer apps’.

‘If this area [of online health communities] is ever going to work someone is going to have make a shedload of money’ Digital platform leader.

‘We’re just looking for the golden thread, the functionality that connects meaning for patients to money’ Digital entrepreneur.

So why is there is no Facebook for health? As we struggled with this question the usual canards (healthcare services are inherently digital laggards. Clinicians are too busy. Managers don’t like it) seemed increasingly inadequate.

We think the answer lies in the particular scaling dynamics which operate in online health forums. The factors which drive winner-takes-all scaling for Facebook-like platforms don’t apply to online health communities which are subject to multiple centripetal forces including:

  • The experience of ill health is inherently more private than for many online topic areas.
  • The trusted relationships that people value as part of these communities have a functional limit in size. Anecdotal evidence from interviews suggests a trusted group being much smaller than Dunbar’s 150 people, perhaps being as small as a dozen people.
  • Size doesn’t matter: knowing the post about your mastectomy has been read by your mum, your best friend and the person in Auckland who has been through it too means much more than the absolute number of people who have viewed it.
  • The technology means it is easy to set up a new community.
  • We assume that joining things up is a good thing when the evidence is pointing in a different direction. An influential University of Pennsylvania study shows how three decades of network theory positing that removal of group boundaries and aggregation of networks is counterproductive[1]. The study shows that this actually delivers lower levels of social integration and knowledge transfer. And when society is more grouped, up to a point, these groups are more effective at sharing knowledge, both within and across group boundaries.
  • Scaling requires homogeneity. Inherent in the notion of efficient scaling, is the repeatability of social and business model without the need to adapt. This is eloquently expressed in a quote from the recent book by Anna Lowenhaupt Tsing titled The Mushroom at the End of the World[2]

‘Scalability, in contrast, is the ability of a project to change scales smoothly without any change in project frames. A scalable business, for example, does not change its organisation as it expands. This is possible only if business relations are not transformative, changing the business as new relations are added. Similarly, a scalable research project admits only data that already fit the research frame. Scalability requires that project elements be oblivious to the indeterminacies of encounter; that’s how they allow smooth expansion. Thus, too, scalability banishes meaningful diversity, that is, diversity that might change things’. Anna Lowenhaupt Tsing in The Mushroom at the End of the World.

Seen in this light the concept of scaling is, in many ways, not of service to the emergence of online health communities, which mainly focus on specific diseases or combinations of specific diseases and demographics, all of which have their own technical and social needs.

Two key things flow from this.

First these factors tend to neutralise the advertising-based business model than underpins most social media platforms: the audience is smaller, it actively maintains its privacy, and selling stuff to, say, post-mastectomy patients is inherently tricky. So actually Facebook itself is not that interested in becoming the Facebook for health.

Second online health communities tend to ‘fork’ into new communities rather than cohere into winner-takes-all behemoths.

The result is what we have termed ‘clustered scaling’

scale 1 copy

Each bubble represents a subset of a wider community – perhaps a topic on a forum, or a site focused on a particular disease. The set of bubbles is then all the topics on a forum, or all the sites and Facebook groups focussed on the disease.

Each online community has its own internal dynamics and internal network effects. Neither network effects not internal dynamics scale from cluster to cluster. In addition, the size of each bubble is limited by Dunbar’s number factors and there are strong centripetal forces driving bubbles to fork, split or die. These include people wanting to focus on a particular aspect (say diabetes and renal failure), ideological differences (diet is the best way to cope with Crohn’s), super-user fatigue, and low barriers to entry (it’s easy to set up your own community).

This multiple splitting, forking, fading and renewal gives the overall community the flexibility to cope with a range of motivations, topics and disease trajectories. But it also prevents winner-takes-all scaling and results in low total usage even when you add all the online users interested in particular disease together.

Footnote to diagram.

The size and rate of growth for each cluster depends on:

  • The size of the internal network effects i.e. the internal virality ratio within each community.
  • Incentives internal to each individual community – for example Patients Like Me having a relationship with a Pharma firm that is interested in a particular disease; or an NHS trust that is strongly committed to local use of Patient Opinion.
  • Increasing cluster size increases internal transaction costs, due to factors such as moderation, members’ requirements for real intimacy and quality of contributions.
  • How frequently each community forks into daughter fora.

Platform and app mentality kills the future

This rush to scale deprives us of the space needed to make sense of, and find ways forward with, the fundamentally complex issues we have begun to unpack in the Barriers section of this research. It also means that much resource is expended in pursuit of the killer app or the next internet unicorn, which in reality adds to the graveyard of digital platforms that have failed or, worse, the numbers of these entities that are “walking dead”, surviving on the internet but with a paucity of users.

This combination of over exuberant venture capital focused on inappropriate winner takes all platforms might well result in the failure of much of the current wave of digital health investment. But the prospect of its success is barely better since it is unlikely that if a Facebook-like health site does emerge that it will be able to sustain the altruism and social capital which early non-commercial online communities so clearly demonstrate.

Next

 

[1] http://phys.org/news/2015-06-social-networks-group-boundaries-ideas.html

[2] http://press.princeton.edu/titles/10581.html

Leave a Reply

Your email address will not be published. Required fields are marked *