Why failing isn't final: A new taxonomy of failure in government

Failure is hard. Whether as an individual, a team, or an entire organisation, few people relish the idea of failing. In government, where the spotlight of political and public scrutiny can be particularly bright, and the consequences particularly severe, failure is often considered to be "not an option".

Of course, this doesn't stop failure from happening. As we are exploring through the hundreds of case studies in our Observatory, failure to achieve the intended impact is a common feature of many government initiatives around the world. Entire books have been written collating the many spectacular and disastrous failures that governments have precipitated over the years.

The policymakers' dilemma

How should policymakers respond? The literature on innovation and personal growth encourages us to embrace failure, pointing out that many of the world's most successful organisations and individuals repeatedly "fail". The logic goes that it is not the act of failing that is the problem, it is failing to learn from failure.

Even if this is true (and I think it is bounded by some important caveats discussed below) it seems particularly naive to suggest that politicians and civil servants will happily embrace every failure as a new learning opportunity.

Aside from the fact that behavioural psychology argues that loss aversion bias is an innate human characteristic, the consequences of public policy failure can be disastrous. If a pizza delivery platform fails, that will no doubt be annoying for the hungry customers. If a welfare payments system fails, people could starve. The political implications of policy failures can be career-ending or even government-felling.

Productive versus unproductive failures

Our research is now helping to distinguish between the different types of failure that policymakers may experience - a taxonomy of failure. We argue that governments should welcome productive failures that help to increase the chances of public impact, whilst avoiding unproductive failures that deplete public resources and risk causing harm without improving outcomes.

Productive failures are those resulting from genuine experimentation, often in complex systems, where it is impossible to know a priori what the outcome is likely to be. Any result, successful or otherwise, therefore provides valuable new insights about the system in question which can be incorporated into future thinking. The cost of the failure is more than offset by the value of the learning.

A good example of a productive failure is the Scared Straight initiative first deployed in the 1970s in US state of New Jersey. The programme aimed to deter juvenile delinquents by bringing them into contact with existing offenders through organised prison visits. The results showed that not only did the visits fail to deter crime, they in fact led to an increase in offending behaviour. A surprising result, but one that was repeated and found to be robust, resulting in the termination of the programme.

In contrast, unproductive failures are those which could have been avoided because they resulted from error, oversight or poor judgement. Such failures are unhelpful because they don't reveal any significant new information about the world, simply confirming what is already known. The cost of the failure is not compensated for by the learning.

An example of an unproductive failure would be any attempt to reproduce the Scared Straight initiative today that ignores the strong evidence that such programmes don't work. Such a programme would cost money and other resources while risking poorer outcomes for the participants. They also undermine public trust in government's ability to act wisely and in the best interests of the citizens they are supposed to serve.

Maximising the productivity of failure

So the task facing policymakers is not to avoid failure per se but rather to maximise the productivity of failure.

We've observed a set of four behaviours that governments with greater failure productivity display. They fail early, fail often, fail small and fail smart. This helps them to maximise the benefits of failure while minimising the costs, therefore boosting failure productivity and increasing the chances of positive public impact.

Failing early means front-loading the opportunity to fail, and therefore learn, as early as possible into any initiative. Policymakers that that include an explicit learning phase in their programmes help to encourage this mentality. Keeping initial assumptions and specifications to a minimum also helps maximise the learning opportunities and degrees of freedom to respond to evidence and new information as it emerges.

Failing often means adopting a continuous learning approach that tests, learns and adapts on an ongoing basis.  Adopting a fully agile approach is a tried-and-tested methodology for managing entire programmes along these lines but simply including regular points at which assumptions are revisited and adapted as necessary can make an enormous difference.

Failing small means avoiding high-profile, high-stakes, "big bang" roll-outs. As the cost of failure is usually correlated with the scale of the endeavour, the bigger the programme, policy or initiative, the more costly any failure is likely to be whether financially, organisationally or politically. Adopting a scaled approach which uses small pilots to test critical assumptions before ramping up reduces the risk of catastrophic failure and helps to improve the chances of success.

Failing smart means ensuring that the learning potential of any failure is maximised. This has two implications. Firstly, it means being clear what is known and what is not to avoid repeating mistakes and direct policymakers' attention to the genuine questions. What is "known" should ideally incorporate the best available evidence from around the world, wherever that may be, which is a non-trivial requirement. Fortunately evidence-sharing schemes such as the UK's What Works Network or the Alliance for Useful Evidence in the US help to make this task a little easier. Secondly, when failures do occur it means ensuring that the lessons from those failures are rapidly understood and incorporated into the wider programme of work (as well as ideally shared further afield).

Underpinning these behaviours is a further essential requirement: that of clarity of outcome. When the impact being sought by policymakers is not clearly articulated or is poorly tracked it is difficult to learn and unproductive failure becomes inevitable. To use the Scared Straight example again, without clarity that reducing offending behaviour was the goal, and having a way of measuring that outcome, it would be very difficult to make any claims about whether the programme was working or not. Too many government initiatives are far too vague about the objectives they seek risking built-in unproductive failure from the outset.

Encouraging a productive failure culture in government

While these may be the behaviours that are likely to increase the failure productivity of government it is another matter to encourage the cultural and behavioural shift required to follow them through. Here, we believe that the Public Impact Fundamentals can help in two ways.

Firstly, they force policymakers to be clear what success looks like in terms of the public impact being sought. Focusing first on ends rather than means helps to avoid the trap of settling on a specific solution, or set of assumptions too early thus encouraging greater experimentation and openness in the policymaking process.

Secondly, they expand the suite of considerations beyond pure "policy" to encompass legitimacy and action as well. This helps to highlight areas of greater uncertainty where a more humble approach from policymakers is likely to make more sense. This also shows how apparently simple policies can sit in complex systems where unintended consequences and feedback loops can lead to surprising results.

Our research into failure is still at its early stages, but we hope that by shining a light on the nature of failure in government through case studies as well as theory we can encourage the adoption of a more productive approach that helps to ensure failure acts to improve public impact rather than undermine public trust.

FURTHER READING

• When impact falls short: learning the lessons of policy failure. Jorge Hargrave and Claudia Irogyen report on an array of new case studies on policy failures now available in the Public Impact Observatory

• Helping governments bridge the gap between intentions and performance. The ideal of good government is one shared by billions of people around the world but more needs to be done for it to become a reality, says Adrian Brown

• Focusing on the Fundamentals. The RSA's Matthew Taylor tells Adrian Brown about how the Public Impact Fundamentals has influenced his thinking and approach to a high-profile government review on modern employment.

• Recognising and renewing governments' legitimacy. Preserving their legitimacy in such a fast- changing world should be a priority for governments the world over, says Maryantonett Flumian

• Action stations: advancing to impact. Beth Blauer explains why management, measurement and alignment are critical to the successful transition of a policy into real-world effect

• Picking the policy that will have the greatest impact. Overcoming the barriers to policy implementation involves not only the setting of priorities and ambition but also defining what outcome is actually intended, says Sir Michael Barber