Difficult, yes, but not even near impossible. I know because I have implemented a federal pension calculator for a private sector company, twice. The first iteration was more for planning, however, I am nearing completion of a new calculator that would be suitable for both retirement processing and longer term planning. It's not a problem that can be solved with money or manpower, it's a problem that is solved with a very flexible design and intense knowledge of the business rules. You can also not rely on any sort of specification provided by the federal government or the laws themselves because there is too much grey area left to interpretation. We had rely on multiple sources of information including word of mouth to pull everything together into a unified design. We have implemented our solution with a staff of only 2 developers and a handful of subject matter experts. The development process was not a "go away and build this" but rather a constant back and forth of Q&A and testing from start to finish.
I am familiar with the previous failed attempts at modernization and the inability to compute accurate retirement benefits was a big factor in the most recent failure. The company I have built this for is a small company and while they have attempted to be involved in the modernization effort they are simply overlooked. They could literally offer their calculator for free to OPM and they would be turned down. If you are not a large corporation, with a large lobbying budget, you have no chance.
The other problem of getting the data in electronic format is actually a much easier problem to solve technically, however, it's a much harder issue to physically implement due to the large scale of existing data that is paper based.
