In contrast, converting 1M LOC of code per month is a much more solid measure, as long as you measure LOC of the source, not the new code. Sure, in the short term you can pick the easy/verbose things to port, but it's hard to do sustainably. A 5M LOC code base would still be expected to be ported in 5 engineer months.
Granted, you can still rush the work, not test properly, neglect good planning and engineering. Ported lines of code should not be the only measure (just like with any other measure). But it's a much less problematic measure than coding 1M LOC
Which is the core point of my reply and not something to just be casually handwaved, thank you very much.