A year on from setting out concerns about the adoption of automation in systems, particularly behind the scenes in public services like the Post Office horizon accounting system, so it is time to review progress. Although many have said this was awful, very few have put effort into addressing the issues and implementing the robust… Continue reading One Year of Progress on Lessons from the Post Office

Systems that are both autonomous and adaptive are assuming substantial functional roles in major organisations. These ‘AI’ systems introduce a new risk profile which is requiring adaptation from existing governance practices, most obviously the audit functions. Despite enthusiasm to ‘audit’ systems, these third party services reflect basic actuarial analysis around ‘bias’ rather than the corporate… Continue reading Audit of AI Systems

Positive observations evoke a similar sentiment. Sightings, successes, substantiations are all viewed as vindication with missing, contrary and disappointing results cast as aberrations or glitches. Humans prefer to be right (confirmation bias) but they can also misjudge the information value of tests (a high prior probability of positive) and actively suppress unpalatable truths. But sometimes… Continue reading Negative Outcomes

Conceptions of the role of statistics in public life vary between nations, perhaps starting from the early British conception of political arithmetic by tables of account, even as the etymology is Teutonic. The contrast between the administrative state in France (la statistique publique) and the federal system in Germany gave rise to management and descriptive… Continue reading la statistique publique

Statistical data ethics is an applied ethics, of what ought and ought not to be done, and as such eventualities as ought not to be done potentially transpire. While some cases have suitable oversight and therefore only needs support in terms of insisting ethical requirements are acted on appropriately, others lack such oversight for several… Continue reading Raising Concerns

Data scientists, like many others across the UK, will have been shocked by revelations of what happened at the Post Office, as a flawed 1990s IT project led to prosecutions, convictions, and fines for more than 900 subpostmasters on the basis of faulty accounting data. The system that caused this “widespread miscarriage of justice” itself,… Continue reading Learning from a Post Office Disaster

In anticipation of the Lievesley review recommendations, the post on independent statistics set out some of the issues with current structures in the UK. So it was encouraging to see many of the points there, like the proposal for a deputy national statistician for methodology specifically, very much like a chief scientist. But with all… Continue reading Governance of UK Official Statistics

Research is an organised and situated human activity. But that organisation is not directly entailed on past success, not an aim. In the 1940s, Robert Merton set out four principles which encapsulate the scientific endeavour. Yet he clearly did not include any public goal-orientation beyond universalism, and the others emphasise working together without an agenda… Continue reading People, Culture and Environment for Research

In the widely adopted ‘five safes’ model for responsible data governance approvals for projects, accreditation for researchers, security standards and even output reviews are straightforward. This has meant complaints about bureaucracy, and limitations on what is permitted or specifically how research is done, but the public recognise the assurance and express confidence in it. But… Continue reading Safe Data

Public officials are explicitly intended to be ethical role models. The Nolan Principles of integrity, leadership, selflessness, honesty, openness and accountability also include objectivity, noting: Holders of public office must act and take decisions impartially, fairly and on merit,using the best evidence and without discrimination or bias Nolan Principle 3 – Objectivity (as revised in… Continue reading Objectivity in Public Life