What does the Horizon scandal say about software testing, standards and accountability? Could it happen again or are we now better prepared and guarded against such human frailty?
It isn’t often a TV drama strikes at the national consciousness and changes the law. A famous example would be the BBC’s ‘Cathy Come Home’ in 1966.
Now we have ‘Mr Bates v The Post Office’. The drama starring Toby Jones, describes the campaign by a sub-postmaster to clear his name and that of hundreds of others convicted of an array of offences from fraud to theft to false accounting. The common factor was the evidence provided by an online accounting system built by Fujitsu. The scandal has gone down as the largest miscarriage of justice in British legal history.
The background is this: in 1999 the Post Office launched a new online accounting system, Horizon, that linked each of the United Kingdom’s post offices into a network. Transactions could be recorded; income and outgoings would be logged and would be more efficient than the previous paper-based method of doing business.
There was a problem. Horizon never worked. Sub-postmasters and sub-postmistresses noticed shortfalls in takings almost immediately. Many claimed they had little to no help from the helpline meant to clear up any teething troubles. Indeed, the teething troubles only ballooned until hundreds found themselves accused of fraud and other offences. People lost livelihoods, retirement plans, health, relationships, reputations, their freedom and even their lives.
Aside from issues of corporate governance and personal conduct (there is an ongoing inquiry), the core of the scandal is the failure of the IT system.
The UK has a record of failure in large scale, governmental IT projects going back some 30 years, but what makes Horizon different is the scale of damage it caused and allegations (long reported by Computer Weekly) that the software should ‘never have seen the light of day.’
It has been claimed the team that built the system lacked the appropriate skills and qualifications; that the system was built with “no design documents, no test documents, no peer reviews, no code reviews, no coding standards.”
Whilst not wanting to second guess the inquiry and its eventual findings, it begs the question, ‘how could this happen?’
Is it that standards and procedures for building such systems were rudimentary, perhaps even lax, at the time? To what extent have things changed in the industry since 1999? Remembering that we are all flawed humans, how likely is it that something on this scale could happen now?
And what of accountability in this field? After all, when a bridge collapses or a building falls down, questions are asked of the contractors, the architects. What of software engineers when their defective work causes the kind of disaster that is only now being called to account? Leave your thoughts in the comments below.
