^ Well, like I said, I still think the "parallel + reverse branch" design is all around a better approach than this "criss-cross" model. But I do think there are some merits.
First, overall, it's a
simpler network. Seaport trains form a single trunk, with two or three branches. Park St trains form a single core trunk, with broadly matched secondary trunk pairs to north and south (Union and Medford, Huntington and Nubian). No reverse-branching, no interlining, and full coverage can be provided with a smaller number of named services:
Second: it more rigidly
divides up homogenous services. In this model, the "subway-streetcar" services of the A/B/C are totally isolated from the "heavy metro" Huntington services. And removing the Kenmore trains from Park St then opens up more flexibility to accommodate short-turns on the Huntington and Nubian semi-street-running services (e.g. E to Hyde Square, F to Nubian with surface stops), using the loops at Park and/or Government Center. (Which also means that you can theoretically squeeze in a little more capacity into Park Street from Huntington and Nubian than you can under the "parallel + reverse branch" model -- though probably not too much more.)
(EDIT: one drawback of the “parallel” model is that — theoretically — all Kenmore trains will need to funnel through the Park St loop linearly in single file. That makes it harder to prevent cascading delays. You’d have more flexibility with a pair of stub-end tracks with an island platform, but that would be a challenge at Park St. This is where “theory”gives way to real life: probably what would actually happen is that some Kenmore trains would get directed ad hoc to run through to the GC Loop — which means, yes, there would probably be at least some entanglement between the Kenmore and Huntington divisions under the “parallel” model.)
Third: even though they are routed away from Park St, there's an argument to make that a
Seaport routing provides better access to Downtown (beyond obviously the much improved access to the Seaport).
As I outlined earlier this year, South Station is a little bit closer to a slightly higher number of jobs than Park St is. And then of course the access to the Seaport blows the whole thing out of the water. In general, it's always better to aim for transit lines that cross
through the core, instead of terminating in it; routing Kenmore trains to the Seaport is a little bit of a "cheat" in this respect, because the Seaport is acting like a "mini-downtown" in its own right, but still overall you end up with a more balanced system.
Fourth: the "criss-cross" model provides a significant improvement in
transfer connectivity for Kenmore trains. This includes a much shorter transfer distance to the Orange Line, direct access to South Station with the Indigo Line and Regional Rail, and direct access to whatever South Station <> Logan transit service we want to imagine.
(Those last two points are related to the first. A simple Kenmore <> Seaport trunk extends the "lattice" formed by Park-DTX-State-GC, and turns this "Magenta Line" into an "east-west route" like the Red and Blue Lines, crossing the "north-south" Green and Orange Lines at roughly right angles. That enables the through-routing as well as the improved transfers.)
So, I do think there are merits to the model.
What are some reasons the "criss-cross" model
doesn't work well?
First, as mentioned above, it is somewhat more
tunneling-intense than the "parallel + reverse-branch" model (which is honestly kinda amazing for how much it achieves with only modest tunneling).
Second, the "criss-cross" model probably
under-utilizes potential capacity a bit: the "parallel" model sees something like 60 trains per hour between Boylston and Park, split across the quad tracks, while realistically at the absolute most the "criss-cross" model might see 50, more likely 40 or 45 (assuming 10 tph to Nubian via surface, another 10 tph to Nubian via subway, and something like 20ish tph on Huntington, divided amongst S Huntington, Needham, and Riverside.) Likewise, the "parallel" model robustly uses both sets of portals at Pleasant Street, with 10-20 tph going into each portal; the criss-cross model trims that down significantly -- in the crayon shown above, I have dual branches to Nubian, one for each portal, but it's very possible that there would only be one Nubian branch, only requiring a single portal.
Third, it's not
ideal to serve the Seaport exclusively with "subway-streetcar" services -- the
reliability will likely be lower than it would be being served by Huntington trains originating from "sealed" ROWs. Probably you would need some sort of layer-on service that pings back-and-forth between the Seaport and either South Station or Back Bay to ensure some minimum headway standard. The advantage of the "parallel" model is that the Kenmore Division becomes minimally load-bearing in terms of the rest of the system: it's mostly focused on getting its local riders into Back Bay and Downtown; it's
not providing a transfer service getting Red, Orange, Huntington, Nubian, and Regional Rail commuters into the Seaport. Under the "parallel" model, the Kenmore Division can have its ups and downs with less impact on the rest of the network -- no such luck in the "criss-cross" model.
Fourth: I do think it's a shame to
lose the one-seat ride between South Station/Back Bay and Longwood. Back Bay gets a partial consolation, in that passengers can do the somewhat-longer transfer to "New Copley", but South Station really loses out -- the closest you get is cumulative high freq Regional Rail services to Ruggles and Lansdowne, which are both a little under a mile away from the hospitals. The "parallel" model's Seaport <> South Station <> Back Bay <> Longwood sealed-ROW heavy-metro line is
very enticing.
So, there are pros and cons to both models. I maintain that the "parallel" model is better on the balance, but I think it's worth keeping the "criss-cross" model in mind.