Public services in 2030: barriers to change

20 April 2016

Last month, the Home Secretary called for more predictive policing: the use of data to prevent crime from occurring, rather than simply reacting to it. In doing so, she exemplified a trend forecast by speakers at Reform’s recent strategy day: that the public sector will make increasing use of ‘big data’ to deliver services with the help of predictive analytics. Using data in this way allows governments to move from ‘one-size-fits-all’ services to bespoke solutions. In his presentation to the Reform team, Nesta’s Tom Symons spoke of a far more relational dynamic developing between citizen and state by 2030. The potential gains are significant. The potential barriers, however – both practical and political – are also considerable.

On a practical level, the manageability of ‘big data’ cannot be taken for granted. Kate Glazebrook of the Behavioural Insights Team forecasts a forty-fold increase in the volume of data shared online by 2040; in healthcare the increase will be exponential. The total quantity of digitised healthcare data is currently around 500 petabytes – by 2020 this is set to skyrocket to more than 25,000 petabytes. This is enough information to fill 25 million pickup trucks with paper; making this much data usable will be a major undertaking.

Reform has consistently drawn attention to this challenge in recent papers. One initiative in policing was found to spend half the time allocated to data analysis on preparing it for use. In healthcare, much data is siloed and requires a “concerted effort” from providers and commissioners to use – including the skills needed to analyse and synthesise data into comprehensible reports.

Even with clean, useable data sets barriers remain. Politicians, the public and service providers have historically resisted diversion of resource from acute services to preventative measures. The difficulty in showing the benefits of prevention also leads policymakers towards shorter-term fixes – such as funding increased acute services – over longer-term prevention.

Overlaying all of these challenges is the sensitive issue of personal data sharing. The NHS recently attempted to integrate data between general practice and hospitals through its programme; however, roll-out has been hindered by concerns over privacy, and lack of awareness over patients’ ability to opt out. One journal castigated the scheme for “its flawed protection of patient anonymity, an unsuitable opt-out system, unclear criteria for accessing the collected health data, and the risk it poses to the trust between patients and general practitioners”.

Tom Symons also outlined a number of potential ethical pitfalls. For example, road repair services responding to data from smartphones have resulted in resources being directed towards the affluent neighbourhoods from which more data came. Concerns too are being raised over predictive policing, with critics suggesting a tendency to entrench existing demographic biases. In one highly-publicised incident, a man from a deprived Chicago district – with no history of violent crime – was visited by a police officer who warned him that he was on a “heat list” of individuals considered prone to violence.

Apprehension over how personal data is being handled, alongside serious practical challenges, presents significant barriers to realising data-driven public services. Government and providers must be alert to these barriers; if poor implementation turns the tide against predictive analytics early on, an opportunity to revolutionise public services may be lost.

Alasdair Riggs, Research Assistant, Reform.



No comments yet.