You're right, I haven't showed all the work. Let's look:
Where workers from Watertown and Allston are going:
Where workers are coming from to jobs in Watertown and Allston:
Now let's look at two alternatives:
The GLX from Union Sq to Watertown via Porter is just about 6 miles. The line would run along the old ROW until halfway between Arlington St and School St since the rest of the ROW has been mostly built up. From there it runs on the street, preferably in dedicated lanes.
I'm estimating costs based on the cost/mile of the current GLX ($530 million) and the high end of street running light rail ($100 million). This brings the GLX cost to about $2.6 billion.
The black line is a new tunnel connecting to the abandoned Brattle Yard lead tracks which, theoretically, could be converted into a light rail terminal. The tunnel only needs to run to Western Ave, along which and out to Arsenal St it would run along the street (again, preferably in a dedicated lane). The tunnel is just shy of a mile so we can peg it at $1 billion. The rest of the street running costs $300 million. Total: $1.3 billion.
If we wanted to simply run light rail on North Harvard and JFK Streets (into the bus tunnel) we are looking at a total of $390 million.
Not shown, but for sake of the argument, would be a street running light rail line from Watertown Sq to Central Sq. At 4.3 miles that comes to $430 million.
There is obvious demand west of Watertown Sq but any service there relies on figuring this part out first, so I've excluded it from the analysis.
There are a lot of ridership factors we aren't including here, mostly trips to school and non work related trips. But I think the jobs numbers paint a pretty clear picture as to where people are going. There looks like possibly a couple thousand workers coming from north of Harvard Sq into Watertown and Allston. There is virtually no one commuting from Cambridgeport or Kendall. But most people work close to home so a solid east-west transit line through Watertown and Allston, not up to Porter, is the clear winner.
The question is where do we send it after? There is demand to Harvard but not north of Harvard. Saying riders can go to Central Sq and backtrack isn't going to be popular, especially when it would be cheaper to just go to Harvard. Kendall, via the Grand Junction, could work too, but then commuters coming from north of Harvard are stuck since they'd have to go all the way to Kendall and double back (which they just wouldn't do).
Harvard looks like the winner. Building anything past Harvard, north or south, just adds cost and travel time. We can build a transit corridor in stages, first BRT, then light rail, then add a tunnel if need be. Extending the GL past Union Sq might be very useful for relieving the RL but it seems to me to be a stretch to suggest that it would help Watertown (not to mention completely miss Allston). If so much of the Watertown/Allston ridership is headed to central/southern Cambridge and downtown Boston then why make them transfer up at Porter? It just adds time to their trip and most of the ROW runs around every place these riders want to go, not to where they want to go.