The folly of 'fail fast'

Public sector IT agility is a good thing, but not when it undermines confidence.

Modern applications development is all about agility: build fast, fail fast, fix the problem and move on. However, the underlying politics of government service delivery haven’t changed. Sometimes, public failure is just a bad idea. In the government sector, it is not just about ‘fail fast’ but ‘fail smart’.

The government sector differs from the private sector in one very important way. The private sector business model is based on competitiveness. It is all about winning market share and delivering profitability. The government sector exists in an area referred to by economists as ‘market failure’, where traditional market forces do not operate as expected under the private sector model. For example:

• There is no competitive advantage for a government in growing the number of welfare recipients.
• There are no prizes for a government collecting welfare debt if it is seen to abandon the principle of acting as a model litigant.
• The public will only continue to provide good Census data if the public maintain confidence in the soundness of the process.

A public service organisation exists to serve the public. It can only continue to perform that role if it retains the confidence of the public it serves. When projects fail badly, it strikes at the very heart of the public sector’s raison d'être. This is why the ‘fail fast’ doctrine should be approached with great care.

The public sector has moved a long way from earlier days of absolute risk aversion. Risk management is now accepted as a legitimate part of good management practice. Failing and recovering fast is the latest development of sensible risk management practice. However, there is a fine line that must be walked. Any whiff of recklessness or naivety will be dealt with harshly by the community.

Consider the Census — an excellent example of the significant downside in failing in a public way, even if the recovery was relatively fast. The move to an online Census was a reasonable decision and a good example of digital government in action. On Census night, the inability to deal with a series of denial-of-service attacks was a significant issue for the project. From a purely technical perspective, it could be argued that the Census did produce good data, and it did save money, so what is the problem?

The government’s own review into the project outlines the problem very clearly. The executive summary commented “the public’s confidence in the ability of government to deliver took a serious blow, more so than any previous IT failure … crucially important is the need to understand how the Census got to the point where the cybersecurity arrangements brought into question the trust and confidence in a fundamental government service. The public’s lack of confidence will linger.”

The lesson is clear. In contemporary government, it is good practice to be agile and, if necessary, to fail and recover fast. But it is extremely foolish to take any action that places community confidence in government at risk. The challenge is not to fail fast, but to fail smart.

Pictured: Kevin Noonan.

Follow us and share on Twitter and Facebook