In Part 1, I looked at one of the more high-profile examples of inefficiency flagged by DOGE – the federal government employee retirement process – and examined the historical difficulties in streamlining it. In Part 2, I want to explore why those difficulties exist. With the benefit of hindsight, DOGE’s record is mixed. This example explains why.

How constraints shape efficiency

There’s a trap that regularly catches digitization and other efficiency-driving programs. An organisation spots what looks like obvious inefficiencies – in this case, paper forms, manual processes, filing cabinets full of documents – and launches a program to fix it. The business case looks clean. The timeline looks reasonable.

The solution itself is perfectly deliverable – the requirements are well within the system specifications. The business supports the change—they’ve often spent years frustrated by the limitations of the existing process. Yet, when it comes time to execute, all manner of issues, challenges and roadblocks arise.

What’s going on? One common scenario I’ve found is that those old “inefficient” processes often aren’t as inefficient as they first appear. In fact, existing processes have usually been optimised, over time, around the existing organisational constraints. Those constraints aren’t always well understood, in fact they are often hidden and hard to map.

Here’s the important thing – often those efficiency gains in your business case can’t be truly realised without fixing, or at least appropriately addressing the underlying system constraints.

No doubt, given the historical difficulties the US federal government’s employee retirement process has a range of system constraints.  From the outside, four (and a bit) constraints jump out at me. Let’s take a closer look.

Challenge 1: The document storage trap

The first constraint – that really needs to be worked through at the beginning of the program – is the existing document storage contract. These contracts are notoriously messy: long terms, hefty break fees, and per-document handling charges for moving or transferring records. The break costs alone can negate the cost efficiencies you’re seeking to realise. I’ve seen contracts where exiting early still requires paying out the full remaining term.

Getting this right requires careful planning of how, and when, you exit existing arrangements, and what storage looks like on the other side. It may not be as simple as making everything digital. For example, regulatory requirements may mean some hard copy records must be retained regardless or you may determine not to digitize the existing archive. Either scenario would mean, you’re not replacing one storage process with another – you’re managing two.

It’s no coincidence that the document storage industry is actively positioning to profit from exactly this kind of transition.

Challenge 2: The form is more than a form

Digitising a form easy. Sometimes it is – a basic intake form feeding into a CRM can be a genuine quick win. The federal retirement form is not that sort of a form.

What appears to be a simple, antiquated paper form is actually a complex, multi-agency data and evidentiary collection process. Information is drawn from four separate sources:

• The retiring employee – personal details and supporting documentation
• The employee’s agency – employment and service history
• Third-party and historical sources – submitted as hard copy supporting documents
• OPM itself – cross-referencing submissions against its own records

The complexity isn’t in the interface. It’s in the structure of the information behind it:

• Data comes from multiple systems that don’t naturally integrate
• Agencies have inconsistent internal processes and record-keeping practices
• Supporting documentation often exists outside any structured database
• Identity, employment history, and entitlements must be verified across sources

A digital form can collect inputs. It cannot, by itself, resolve inconsistencies across organisations.

Any digitisation effort will quickly runs into a set of unresolved choices:

• Do you integrate directly with agency systems?
• Do you require employees to upload supporting documentation?
• Do you accept continued use of paper evidence and digitise it downstream?
• Or do you redesign what evidence is required in the first place?

Each choice has cost, complexity, and user experience implications. The decisions made here ripple through the entire program – get them wrong and you’re rebuilding later at significant cost. The antidote is proper user research – not a requirements document, but genuine engagement with employees, agency staff, and OPM workers to understand what the process actually looks like on the ground.

A closely related issue (the “bit”): incomplete data collection

One of the biggest bottlenecks in the existing process is chasing outstanding information. This makes sense given we are dealing with a complex data collection process. Workers can reportedly spend months pursuing a single missing agency signature. As part of efforts to streamline and improve the data collection process, the program will also need to consider how to address scenarios dealing with incomplete information. Done well, the knowledge the program builds about the process can streamline collection and non-compliance processes. Handled in isolation, there’s a risk the process becomes more inefficient.

Challenge 3: Bureaucratic complexity

It’s tempting, in any transformation program, to treat existing bureaucratic rules as noise – legacy overhead to be swept away –  to “start from scratch”. This is a mistake.

Those rules have been put in place for a reason, before you can dismiss them, you need to understand what is driving them. Failure to do so will create all sorts of headaches (and costs) later down the track.

In this case, the key rules appear to govern two things: what information and evidence is required to establish entitlement to benefits (covered above), and how benefits are calculated.

The benefits calculation is genuinely complex. Federal retirement rules are the product of decades of legislation, amendments, and agency-specific provisions. Under the current system, calculations can take between one hour and two days. Getting this right digitally requires rigorous analysis, extensive edge-case testing, and – critically – a design that can absorb future rule changes without needing to be rebuilt from scratch.

Challenge 4: The archive is a problem of its own

The current facilities – that limestone mine isn’t just where the applications are processed, it’s the archive. Decades of records sit in physical storage, and any new system has to account for them.

As of 2014, roughly 85% of applications could be processed using already-digitized records. The remaining 15% required physical retrieval from the mine. In 2019, OPM ran out of funding to digitize that last 15%. The questions now are blunt:

• Can those records be digitized? Some may be old or fragile. The economics of digitizing rarely-used files may not stack up.
• If not, what’s the plan for the 15% that need physical retrieval? A hybrid process isn’t necessarily wrong – but it needs to be designed deliberately.
• What does data migration look like across decades of records held across multiple systems and organisations?

The real lesson

Back in April 2025 DOGE raised the prospect of addressing inefficiencies with the federal government employee retirement process. Four decades of failed attempts and more than $100 million spent suggest the retirement process isn’t fundamentally a technology problem. It’s a systems problem.

Successful digitisation isn’t about replacing paper with software. It’s about understanding the constraints that shaped the existing process—and deciding which of those constraints need to be redesigned, which need to be managed, and which simply have to be accepted.

The organisations that succeed don’t ignore complexity. They make it visible, work through it deliberately, and build solutions around the realities of how the business actually operates.

Published by Aidan McShane on 12th July 2026