A few simple observations make completing the MFE an obviously poor idea:
(1) We should prioritize providing transportation infrastructure where there is already proven, unmet need, rather than where you would need the project to stimulate large amounts of new development outside the current development footprint in order to supply sufficient demand to justify the expenditure, and the MFE in particular has always depended on the expectation of such development to create its promised benefits (indeed, all that promise of new development is still threaded through the article linked above);
(2) Highway lanes are a very low peak capacity transportation technology, so even if new development did eventually fill up the MFE, you'd end up capped at a relatively low benefit for the cost;
(3) And finally, all the projections of highway demand using the old models have overshot dramatically and increasingly in recent years, largely because they are based on very old data that fails to account for recent experience, which means it is almost certain the MFE would never actually come close to meeting its original expected usage.
The bottomline is that you shouldn't be spending billions of dollars on transportation infrastructure without a very thoughtful, data-driven, reasonably up-to-date cost-benefit analysis, and if you subject completing the MFE to such a cost-benefit analysis, it will fail miserably. It is simply broken politics and a lot of special interests looking to line their pockets that is keeping it alive, not rational transportation policy.
Edit: This is a good chart to keep in mind in these discussions:
What it shows is how the DOT's projections of total VMT in the U.S. have been way off track in recent years, but all they keep doing is resetting the line, and as a result they just keep getting it wrong again.
The reason this is useful to keep in mind is that things like the MFE were planned based on those old expectations of how total VMT were supposed to grow, basically on the assumption that the Pittsburgh region would be participating in those trends. So conversely, once you realize the old VMT projection model is broken, then OF COURSE you must reassess old projects like the MFE. And again, under any sort of reasonable, up-to-date analysis, there is no way its massive costs could be justified.