Oh good god...
I was a LM employee a few years back. Brought in on a project that was failing. And the main issue with the failure was their process.
For instance, LM was using Common Criteria and they were trying to get the system to EAL4. And frankly, getting there is quite doable. Unfortunately, management and the customers for the project didn't bother to actually understand anything about requirements.
For instance, in Common Criteria, your need to tailor the documents. An example would be this template being tailored to the system requirement:
FPT_FLS.1.1 The TSF shall preserve a secure state when the following types of
failures occur: [assignment: list of types of failures in the TSF].
The above template is obviously intended to be tailored to include a list of possible or predictable failures upon which the system will still remain secure. But this is how LM tailored that little beauty:
FPT_FLS.1.1 The TSF shall preserve a secure state upon a partial system failure.
Notice how the tailoring totally removed anything concrete about the requirement? What kind of partial failure? How do you test it? When is it violated? etc, etc, etc, ad nasium.
And that kind of bullshit "tailoring" was done EVERYWHERE. There would be multi-hour meetings just change, tailor, and interpret specifications tailored that way. And any suggestion by anyone working in the trenches stating that the requirements were badly done and needed to be redone properly in order to actually get a functional system was met by "We can't do that, it would be too costly."
If the above paradigm was used on the Social Security project, I can definitely see why progress has been snail slow and over budget. They're most likely still attempting to get their specifications correct.