Looking through the Lower (Near) Northeast Master Plan, I noticed that the early-2000s study for a Boulevard subway cost $3.2 billion, and was tabled due to cost.
And then I looked at the render, and it hit me that it doesn't need to be that much, at least on the public-transit end of things.
A subway is merely one of the (rather long) list of improvements the Roosevelt Boulevard desperately needs. For somebody who is normally very pro-highway-privatization and anti-highway-investment-other-than-maintenance, it may come as a surprise that I agree that the middle (express) lanes of the Boulevard very, very desperately need to be sunken out of the way; because it acts effectively as something between a malformed boulevard and a surface interstate (though, I hasten to add, not as much a true stroad), the Boulevard has a deplorable safety record. Indeed, multiple Boulevard intersections are regularly among the country's 10 least safe.
And because the Boulevard is a 4x3--a 12-lane surface road with three medians (four cartways of three lanes each), it is hell for pedestrians to cross, except at the few intersections where the express lanes have already been depressed (e.g. Cottman). Which leads to abler pedestrians pushing the light. Which leads to accidents between them and drivers (particularly on the express section)...
The Boulevard is one of the few places where "improvements" will actually improve things. Safety on it is that bad, due mainly to it retaining a 1930s-era design.
The only effective way to make it safer is to depress the express lanes, so they act as a true road, and the local lanes restructured to act as an, er, boulevard.
In the act of depressing the Boulevard's express lanes, you'll also create a rather large, rather wide median. (About 40 feet.) A rather lard, rather wide, rather transit-ready median. In doing so, we export the costs of prepping the land (i.e. the median) to the highway department.
And this, in turn, sinks--by quite a lot--the necessary cost of developing a Boulevard subway.
Here's what I believe is the optimal phasing:
Phase 0: sinking the express lanes, incidental prepping of the transit median, highway expense
Phase 1: extending along this median to Cottman
Phase 2: extending from Cottman to Bustleton or Byberry
...And another thing: By sinking the express lanes and providing good transit in the median, we can also free up the space above for an alameda with a bikeway in the middle. It wouldn't be cheap, sure, but the potential's certainly there.
And with the middle express lanes sunk out of sight, out of mind, the heart of the Northeast would start feeling much more pedestrian-friendly.