If we diet I-90 to three lanes in each direction (or maybe even keep four lanes in each direction), possibly reduce the width of Montfort St, probably shift the tracks south, and rebuild the St Mary's St bridge across I-90, couldn't we make space for an I-90 eastbound off ramp and an I-90 eastbound on ramp at St Mary's St?
I think eliminating Storrow Dr / Soldiers Field Road between Charlesgate and the Allston Interchange should be the priority, because that would simplify the construction through the throat between the BU Bridge and the Allston Interchange, and probably simplify a potential future pedestrian bridge to connect BU to the waterfront across the throat.
Eastbound Pike ramps are going to be brutal. And this particular area around Mountfort is reserved for the BRT-option Urban Ring tunnel as means of getting between the BU Bridge hillside and Lansdowne Station (the LRT option being an under-reservation burial of the B). So until the UR picks its mode of choice amongst the two primary Alternatives don't expect the rump of hillside around Mountfort to be fair game for anything else.
Westbound you do have plenty of ramp options. There's the downtown augmentations previously studied, and then Allston at Birmingham Pkwy/N. Beacon where if you cut wholly-redundant Birmingham from the street grid and repurpose it as a WB offramp into the rotary with reciprocal onramp just west of the rotary off Nonantum Rd. you'd be doing everything in Allston a favor on load-spreading by taking loads away from the overloaded eastern river roads and shifting it to the underutilized western river roads...as well as taking huge Watertown-bound volumes off dense Galen St. in favor of way under-capacity Nonantum. It's outside of the locus as where we see the problem primarily piling up in the form of traffic jams, but the half-assed connectivity between the Allston exit and first Newton exit contributes to all the overloading from afar.
So while Storrow/SFR pretty much has to remain (Bowker-less, mind you) to Charlesgate, it'll be asynchronously loaded more eastbound less westbound traffic if you work the Pike WB extra exits to pry induced demands off the river roads. Which is a huge and decided improvement over that section of Storrow being overloaded in both directions. Means you don't have to go searching all that hard for operational improvements to the WB side because it'll be less critically overloaded...whereas EB you definitely need to find a lot more breakdown lane space everywhere but where the most-pinched riverbank allows, and more accel/decel room at the University Rd. and Charlesgate offramps. So the residual parkway keep in light of the limited Pike EB options really isn't all that terribly impactful in the New World Order of rebalanced loading, and much less a festering traffic sore than it is now even despite the fact that there are no currently viable alternatives for outright elimination.
Ultra-ultra long term I guess we can keep squinting at the environment to see if that segment of riverbank can ever be liberated. I'm not all that optimistic anything game-changing will be found, but keep overturning every stone in study as due diligence. In the meantime you have to chunk out the Kenmore-Charles Circle midsection as a whole different diet/elimination project and not worry about Allston-Kenmore coattails. Separate universes for now. The midsection is the one that's completely Pike-redundant to both directions, that has the largest share of purely induced demand traffic, that has the most overbuilt 6-lane capacity to slash back, that has the EB underpass tunnel nearing end of maintenance life, that has the most parkland acreage to restore, that has the potential "equal or better" transit trade-in that can go immediately on the board as an offset, and that has some Back Bay flood protection bona fides to weigh if the vacated EB carriageway + Back St. retaining wall were framed with a semi-surface transit box tunnel that doubled as an enhanced passive flood wall. Target the corridor in independent projects: midsection diet hashing out the politics of transit trade-in + WB bust-down to lazy park access lane first, because that's got the strongest bona fides and cleanest/least-impactful trade-in allowances. Study Allston-Kenmore as the wholly separate next step...keeping expectations appropriately low because the Pike EB options are all difficult to stage and pretty ugly-flowing even if they can be staged. And treat Charles/Public Gardens to O'Brien as a unique forever-keep catchment because it's the MA 28 mainline, but explore lane-dieting from its 6-lane + half-frontage sprawl if area traffic loads get reshaped by things like McGrath teardown removing the induced demand on O'Brien, Rutherford Ave./93/1 improvements strengthening Gilmore Bridge as the head-and-shoulders get outta town route for Kendall, and some of the Somerville/Medford interchanges on 93 that are missing key legs getting infilled so 28 can be a lot less load-bearing.
