This is a two-part series taking a closer look at some of the opportunities and challenges of digitizing manual processes. The article has been framed around DOGE’s recent recommendation to streamline the US federal government’s retirement process. In Part 1, we consider difficulties previous administrations have faced trying to digitize the retirement process and why this time might be different. In Part 2, we will explore some of the issues and challenges that may be encountered when digitizing this type of process.
Part 1: An Old Problem & New Opportunities
DOGE and the Government Retirement Process
If you have been following the work of Elon Musk and his team at DOGE, then you have probably heard about the old limestone mine in Pennsylvania where federal government employee retirement documents are processed and pensions calculated. Back in February, DOGE flagged this as a prominent example of government inefficiency, posting on X:
“Federal employee retirements are processed using paper, by hand, in an old limestone mine in Pennsylvania. 700+ mine workers operate 230 feet underground to process ~10,000 applications per month, which are stored in manila envelopes and cardboard boxes. The retirement process takes multiple months.”
It is fair to say the work of DOGE is controversial. Elon Musk has established a high-profile team that includes founders and executives from Silicon Valley and investment banking, tasked with the mammoth enterprise of identifying $1 trillion in cuts to US government spending. Democrats tend to view the work of DOGE with suspicion, an excuse to cut services and cull staff. In contrast, Republicans tend to believe DOGE is unearthing all manner of bureaucratic waste, abuse and inefficiency.
For my part, I’m fascinated with the work of Musk and the DOGE team. In general, I am supportive of efforts to improve government efficiency – I believe the more effectively government money is spent, the better outcomes it delivers. In my experience, however, reining in expenditure and driving efficiency can be complex. Identifying the inefficiency is often the easy part, actually delivering the savings can prove much more complicated.
Revisiting an intractable problem
The US government has been trying to automate their retirement process for over 40 years. All told, across multiple US administrations, more than $100 million has been spent with limited success.
As explained in a recent Wired article:
- Automation was first recommended back in 1981 and the first automated system was launched in 1987. This system was eventually shut down after failing an independent review.
- Another attempt to automate the process began in 2001 (with contracts awarded in 2006). This system, RetireEZ, launched in 2008, before being shut for quality issues.
- In 2013, the system was partially digitized, enabling government retirees to receive 80% of their pension quickly, but with the residual 20% payment requiring manual processing. In 2019, the Government Accountability Office reported that the Office of Personnel Management, who runs the system, had ran out of money to finish digitizing the system.
What this history makes clear is that, it’s not for lack of trying that the system has not yet been automated. Significant time, effort and, not least of all, money has been spent to streamline the process. While important progress was made in partially digitizing the system in 2013, the digitization process has proven more challenging than anticipated on prior attempts.
Cause for optimism
Despite the issues faced in previous automation attempts, I am optimistic that further digitization the process could be successful this time. The key 3 reasons for my optimism are:
- Benefit of previous Partial Digitization – Based on the Washington Post’s 2014 feature story, “Sinkhole of bureaucracy”, significant headway appears to have been made in 2013, with approximately 85% of documents digitized into the digital archive.
- Evolution of cloud computing – The cloud computing landscape has evolved significantly since 2013. The next round of digitization will be able to leverage substantial improvements in cloud infrastructure, applications, integrations and workflows.
- Ability to leverage AI – As part of the digitization process, there may also to be an opportunity to leverage Artificial Intelligence to streamline some of the more manual data entry tasks currently in the system.
The failure rate of digital transformations can be high. It has been estimated that between 70-95% of digital transformations fail to meet their original objectives. In order to realize the efficiency benefits the DOGE team (and prior administrations) have anticipated from automation, the digitization program will need to be successfully implemented. In Part 2, we will explore some of the issues and challenges that may be faced in digitizing this type of process.
Originally published by Aidan McShane on LinkedIn, April 2025