So it's hardly a slam dunk, let me start by saying that. But basically there are three reasons.
1) As seen in the ridership data above, the two primary markets for service through the Transitway are South Station-Seaport and South Station-Logan. The market for Logan-Seaport is much much smaller. (The huge caveat is that I haven't seen any SL3 data anywhere, so it's possible that there is a Chelsea/Eastie-Seaport market that is now being served by SL3, which would impact this calculus.)
If the primary destination for 90% of passengers coming through the Ted Williams Tunnel is South Station, then it seems worthwhile to consider sending buses straight there, minimizing their time in the Seaport -- perhaps one intermediate stop in front of the World Trade Center station (which happens to be just after the off-ramp). Let SL2 focus on South Station-Seaport and let SL1 focus on South Station-Logan, instead of trying to combine the two.
There isn't a great way for SL1 buses to skip stations in the Transitway, and moreover SL1 buses are forced to do a layover at Silver Line Way in order to switch to the wires. So it's a lot of wasted time.
2) The Transitway has speed restrictions due to the tight corners (which is why @Wash is suggesting the guideway). Well-maintained and segregated surface bus lanes, combined with transit priority signaling, likely could offer a (slightly) faster journey through the Seaport, especially given that the streets are already quite wide. Moreover, without the need to journey into the Transitway, SL1 buses would not need to switch to electric power, streamlining the journey overall.
Somewhere on my computer, I have a map comparing the current highly circuitous route that SL1 and SL3 buses take through the Seaport to a more direct route that could be enacted via single bus lanes on Summer St, Congress St, two blocks of D St, and one block around South Station. I'll see if I can dig it out. But I just did a quick check and the current inbound route is 1.8 miles with four intermediate stops, one of which includes a power switch, vs a bus lane alignment which would be 1.5 miles (much of it straightaway) and probably one intermediate stop.
(Or even down to 1.3 miles, if we ditch the intermediate stop, which might be worthwhile.)
3) In the long run, the Transitway should have LRT running through it. Now, I'm not opposed to also running buses alongside the LRT. But the challenge I see is that the Ted Williams Tunnel is always going to be a wildcard for buses running into the Transitway -- there will always be additional unpredictability due to running in mixed traffic. The overarching theme for LRT in Boston for the last 60 years has been trying to eliminate the intermixing of services that run in dedicated ROWs vs those that interact with mixed traffic. Relocating SL1 to the surface would increase predictability of services within the Transitway.
So, in summary:
1) SL1-via-Transitway takes detours to go places its riders don't need to go
2) The Transitway requires a power change and has speed restrictions which a surface route would eliminate
3) SL1 is always going to be intermingling with mixed traffic, which is at odds with the other services that run/will run through the Transitway
