choice between a tortuous modification process and having something designed and built from scratch
It often isn't the choice, but knowing how to manage it. Risk Management = Avoid the avoidable, manage the unavoidable.
Nimrod RMPA/2000/MRA4 kicked off in the late 80s with the development of a few piecemeal modifications. In 1995, CDP threw a large spanner in the works by decreeing that programmes shall avoid contracts that require MoD to provide Government Furnished Equipment. Of course, the first items on the Nimrod GFE list, to be provided to BAeS, were a fleet of serviceable and airworthy Nimrod Mk1s, for conversion. (The former ill-defined, because the latter was not policy). I wonder how many MoD departments promptly turned their back on RMPA, interpreting this policy as "new buy only"? I know the RN did on ASaC Mk7, immediately withdrawing all support and demanding new build, which would have required a 200% hike in funding, a long slip in ISD and reopening closed production lines; not to mention re-manufacturing tooling and trying to find skilled labour. It was a matter of pure luck if project offices in PE had staff who'd done these other jobs before and knew how to circumvent CDP's attacks on his own programmes. CDP was encouraged to at least say his policy applied to forthcoming endorsements, but he ruled it was retrospective. (He was by no means daft, and most thought it a poorly concealed cost cutting measure). From that day forth, your biggest enemy was your own 4 Star. Did I mention full-bored and countersunk?