Countering the Internet of Things’ Flow Toward Centralization

PSFK by Guardian



The grand vision of the Internet of Things is currently an exercise in imagination. It is about what happens when more and more of the real, physical world comes online, as devices and sensors proliferate, connecting everything.

The promise is that the Internet of Things won’t just connect our homes, hospitals, schools and streets – it will enable whole new ranges of interactions, services and efficiencies. It’s not just about the things, in other words – it’s about the people and environments that animate them.

This wider vision of the Internet of Things is still evolving, with considerable excitement from tech firms, entrepreneurs and governments. But it’s far from fully realised yet. And like any evolutionary process, a heady brew of nature and chaos will determine its outcomes.
Order in chaos

The chaotic evolution of the Internet of Things gives fuel to every utopian and dystopian imagining. The enthusiasts argue that it will be transformative; bringing convenience, efficiency and services never before imagined. Others see hackable vulnerabilities and a dark world of consumerism, inequality and surveillance.

But within these competing visions, one evolutionary truth is clear: the Internet of Things currently favours dominance.

One of the factors contributing to the reinforcement of dominant firms and silos is the difficulty of developing and integrating Internet of Things applications. Functionality in highly distributed, ubiquitous computing environments involves many moving parts coming together in a coordinated, but on-the-fly way.
All the complex things

Compared with traditional software and systems like social media, web services and mobile apps, there is a huge leap in complexity when you move to Internet of Things applications.

Part of this is due to the focus on customisation: accommodating the infinite diversity in what people like, how they interact, and how they use things. In the same way that each of our homes, things and choices are different, we all have different preferences for connectivity, and what we want out of it.

This complexity is compounded by the variety of deployment environments. Take a motion sensor, for example. It might be used in a car, phone, house or any number of other places. As well as the individual user’s preferences about its use, each scenario has different data access concerns, resource requirements (power, connectivity), functional constraints, and other integration needs. Things can be mobile or fixed; always on, or sometimes on. And there’s data – lots of data. All of this needs management.
Programming for the unpredictable

It is difficult for those developing and designing Internet of Things components to effectively predict and accommodate all this variability. Compare it to the web, where the endless and colourful variety of the web is realised through well-established standards and methods for access. We all use much the same software and services, in the same, predefined ways. The grand vision of the Internet of Things is different – it explodes the bounded universe.

To deal with this unpredictability, today’s developers tend towards one of three approaches. The first is to build “closed” systems: X’s sensing home, or Y’s health monitoring system. This is a somewhat old-fashioned, limiting view of the Internet of Things – where components are built for very specific purposes, and any customisation or management must be designed into the system by the developers themselves. This impacts scalability, limiting the integration of other components.

The second alternative is building things to be part of a particular technical ecosystem, such as a platform like Apple’s HomeKit, Google’s Nest and Brillo, or AllJoyn by the AllSeen Alliance. These offer developers tools, services and other components – most of which are offered by those who control the ecosystem – to assist in the development and management of the components they build. The downside is that the interoperability regime itself can create network effects and potential lock-in with the big players.

A third approach is to build things to be generically accessible or “open”. Here, developers don’t deal with the specifics of deployment, usage and runtime operation, instead offloading these concerns to the system integrators, who bring together and manage ranges of components. Currently, integration tends towards globally accessible, cloud-based services; that is, those who aggregate (which is important given the volumes, rates of data and ranges of source), and make data accessible and searchable – the Googles of this world.
Centralisation and counterpower

For the moment, all these approaches tend towards centralisation – whether towards operators of closed systems, controllers of particular ecosystems, or systems integrators for “open” systems. Data flows too, tend to be centralised, even when they needn’t be. So it seems that concerns about dominance, power, and control in the Internet of Things are based on solid ground – the end-user’s controls are left to whoever controls the centralised environment. Related: The internet of things – the next big challenge to our privacy

So, is there a way out? Perhaps, given the Internet of Things is still evolving. But the path to countering the strong forces favouring dominance is far from easy.

From a technical standpoint, it is possible to leverage and build upon existing mechanisms for interoperability to achieve a more decentralised Internet of Things. To incentivise this, what is needed is robust legal, ethical and commercial recognition that the success of the Internet of Things depends on users having much greater and more meaningful control over data flows than they have had before, especially when they concern fundamental rights.

Jat Singh (Computer Laboratory, University of Cambridge) and Julia Powles (Faculty of Law, University of Cambridge) are working on collaborative projects in technology, law and policy.

Systems integrator via Shutterstock