Companies like Deloitte and IBM have made millions building state unemployment websites that didn't work, yet keep getting contracts to fix them
Colin Lecher / The Markup :
Context & Ripple Effects
The Markup's report lands mid-pandemic, when state unemployment systems are failing at the exact moment claimants need them most — and the vendors being paid to fix those systems are the same ones that built them. It extends a decade-old pattern: the coders-for-hire team that salvaged healthcare.gov after its botched launch showed how badly large federal contractor builds can go before emergency crews step in.
The failure mode isn't just broken websites. In Michigan, an earlier [[a:954454|flawed automated system built in 2014 falsely charged thousands of residents with unemployment fraud and collected millions in fines]] — evidence that when these systems fail, the harm falls on beneficiaries while the vendor relationship survives.
First-order effects
- State agencies are paying Deloitte and IBM twice — once to build unemployment portals that didn't work, again to repair them — during a period when claimants are waiting on benefits.
- Unemployed workers bear the immediate cost of the failures: delayed payments and, where automation is involved, wrongful fraud accusations like Michigan's.
Second-order effects
- The rehire dynamic pressures states toward alternatives — the healthcare.gov rescue demonstrated that small outside teams can succeed where incumbents fail, giving reformers an existence proof for different procurement models.
- Vendors are already expanding beyond website builds into higher-stakes automation: as coverage of [[a:837233|third-party fraud-detection algorithms shows, the companies running these systems are often overpaid and under-supervised]] — the same accountability gap, applied to decisions that deny people benefits.
Third-order effects
- If incumbency keeps trumping performance in government IT awards, the sector consolidates around a few large contractors — Tyler Tech's rise to dominance in local-government software despite a history of flaws and release delays shows the same winner-take-more logic below the state level.
- The structural risk is an accountability vacuum: no mechanism penalizes failure, so each breakdown becomes new revenue rather than a lost contract — pushing the eventual correction toward regulation or procurement overhaul rather than market discipline.
The trend: Government technology procurement increasingly rewards the same large contractors whether they succeed or fail, making vendor failure itself a recurring revenue stream.