According to Wired, Musk has tapped SpaceX's brain trust, Steve Davis, to head a scheme to migrate the SSA’s extremely outdated but extremely functional COBOL systems — comprising approximately 60 million lines of code — to Java in just a few months.
The same codebase that’s been stitched together to track every American worker and deliver benefits to over 65 million citizens.
Experts say such a migration is already a logistical bear of a project, even with years of planning, and trying to complete it in months could risk upending Social Security payments altogether.
The SSA itself previously mapped out a glorous five-year plan to modernise its systems in 2017. That got quietly shelved as COVID reshuffled priorities, but even that timeline looked ambitious for a system held together with legacy glue and bureaucratic voodoo. Besides, the system worked, and there was little reason to rush to touch it.
While these things can take months in Musk's brain, in the real world, replacing it would take five to 10 years with a large, experienced team (hundreds of engineers), decades of institutional knowledge, full test coverage, thorough documentation, and robust QA processes.
There would need to be parallel system validation to avoid breaking payouts. This isn’t just about converting code syntax. COBOL is procedural and deeply intertwined with old mainframe architecture. Java’s object-oriented nature means that the logic and architecture need rethinking.
And that assumes the code can be converted. Much of it probably includes undocumented business rules, dead code, or logic designed around 40-year-old hardware quirks. DOGE will spend years just trying to understand it.
Firms like IBM and Accenture specialise in this kind of conversion, often relying on automated tooling and phased migrations — and still take many years with huge cost overruns. A 2022 study on government COBOL system replacements revealed that projects with just 10–20 million lines of code often took more than seven years to complete.
Musk’s team appears to believe it can bypass the usual testing, validation, and edge cases that often arise from decades of regulatory complexity. The advantage is that if Musk’s team cocks it up, it is only poor people who will suffer.
