The anti-corruption and transparency field ten years ago was in pre-iPhone mode. Few if any of us spoke of the impact or relevance of technology to what would become known as the open government movement. When the wave of smart phone and other technology hit from the late 2000s onwards, it hit hard, and scaled fast. The ability of technology to create ‘impact at scale’ became the obvious truism of our sector, so much so that pointing out the failures of techno-utopianism became a favorite pastime for pundits and academics. The technological developments of the next ten years will be more human-centered — less ‘build it and they will come’ — and more aware of the un-intended consequences of technology (e.g. the fairness of Artifical Intelligence decision making) whilst still being deeply steeped in the technology itself.
By 2010, two major open data initiatives had launched and were already seen as successful in the US and UK, one of President Obama’s first memorandums was on openness and transparency, and an international research project had tracked 63 different instances of uses of technology for transparency around the world (from Reclamos in Chile, to I Paid a Bribe in India, via Maji Matone in Tanzania). Open data projects numbered over 200 world-wide within barely a year of data.gov.uk launching and to everyone’s surprise topped the list of Open Government Partnership commitments a few years hence.
The technology genie won’t go back into the bottle: the field will continue to grow alongside technological developments. But it would take a bold or foolish pundit to guess which of blockchain or other developments will have radically changed the field by 2025.
What is clearer is that the sector is more questioning towards technology, more human-centered both in the design of those technologies and in seeking to understand and pre-empt their impact.
First, the open data sector is starting to focus on understanding who is interested in the information in the first place and honing their data releases to those needs, rather than just blanket releases. Many will see that this as anathema, arguing that users alone are best placed to know what their needs are, and that the only way to get to those users is to let the ‘invisible hand’ of data works its magic. But the reality is that data hasn’t magically found its users. The armies of armchair auditors foretold by former Cabinet Office Minister Francis Maude did not come to be. The ‘Here Comes Everyone’ theory of citizens mounting the barricades with smartphones auditing government expenses turned out to be the ‘Here Come the NGOs’ working for, but distinct from, those very same citizens. Of those early 63 early examples, the most impactful were the professional organizations that wrangled with the data and turned it into information that could and would lead to social change.
We need to hone data releases much more to the specific needs of these NGOs, in the same way as open database company Open Corporates works closely with investigative NGO Global Witness to understand the type of information they need to follow up on grand corruption claims. Politicians, government reformers and NGO advocates that have successfully pushed for transparency reforms over the past decade have a great deal of political capital at stake. We will only capitalize on and refresh this political capital, that has been spent around the world releasing sensitive information (e.g. contract information), if those releases lead to tangible impact, both social (e.g. better services) and political (e.g. cleaner government, healthier democracy). If we succeed in doing so, the open government movement will go from strength to strength. If we fail it could be dismissed as a fad.
We are also starting to seriously wrestle with the un-intended consequences of technology on fairness, discrimination and inclusion. 2016 is the year fairness and accountability in Artificial Intelligence (AI) decision-making broke into the mainstream. The consequences of this realization will be far reaching for the open government movement, which has tended to over-rely on and be under-critical of the use of technology in decision-making.
Modern societies have developed a set of institutions to guarantee that decisions are subject to proper norms and systems of accountability to safeguard fair and beneficial outcomes. Yet today automated decision-making challenges these ethical and accountability structures. The consequences are extensive: from fairness in the allocation of public resources, to the future of labor and the on-demand economy, criminal justice, algorithmically-controlled media and citizen empowerment. Can and should you know how a computer came to a decision that may be used as evidence against you in court? ProPublica has recently shown how software used across the United States to predict future criminals is biased against African Americans.
NGOs and research institutes in this nascent sector are trialing activities to ensure those automated decisions are more ethical and accountable. Groups are questioning whether such large amounts of (mostly but not only) personal data should be collected in the first instance, advocating for the transparency and openness of algorithmic decisions, setting up systems so that these decisions can be audited, developing norms and standards to set best practices for companies and governments, as well as fairness-aware machine learning techniques.
We’ve moved from cyber-utopianism less than ten years ago to born-digital organisations taking a much more critical look at the deployment of technology. The evangelical phase of the open data movement is coming to an end. The movement no longer needs to preach the virtues of unfettered openness to get a foot in the door. It seeks to frame the debate as to whether, when and how data might legitimately be shared or closed, and what impacts those releases may have on privacy, surveillance, discrimination. An open government movement that is more human-centered and aware of the un-intended consequences of technology, has a bright and impactful future ahead.