With an agenda to reduce government spending in areas it considers individuals should take greater responsibility for (e.g. their lives / livelihoods, personal income, unemployment benefits), there would be a tendency for confirmation bias to impact decisions on actions to reduce said spending. Any program that is projected to save money in terms of immediate reduction to the budget / forward estimates would be appealing, irrespective of a lack of evidence to confirm that true savings will delivered across the entirety of the program i.e. there may be costs to government to address adverse outcomes of the scheme which are not considered, assessed, or budgeted for.

Another issue may have been overconfidence bias, based on a belief that all technology always increases productivity, and that any such increase will deliver savings that translate to benefits greater than any adverse outcomes that translate to costs. An overconfidence in technological solutions may be based on past successful experiences where technology did deliver efficiencies.

The major flaw with the scheme seemed to be that debts were incorrectly calculated, but what appears to have been lacking was a considered process for addressing issues with incorrectly calculated debts. It should have been apparent that an adverse outcome of any debt collection scheme is the risk that an incorrect assertion of debt is made against a relevant individual. What recourse would the individual have to address the assertion? What compensation and associated processes would be considered as a mitigation strategy etc?

The use of automated technology meant individuals lacked reasonable access to a system for refuting allegations of debt i.e. there was no 'human at the other end' to listen, reason, negotiate and resolve conflicts.

A thorough pre-mortem / risk analysis should have addressed these concerns. Whilst it may not have stopped the major flaw from occurring, it would have at least allowed for a more immediate and effective mitigation response and probably avoided further harm to relevant individuals. For a program of this magnitude it would have been a justifiable investment to secure external risk assessment expertise to produce these assessments. I don't know if this was in fact done but if so, it suggests another layer of failure on the part of the 'independent verifier'.