I definitely agree with your last point. In fact, if it looks like all the money to be had for HSR in the foreseeable future is around $10B (bonds+some reasonable cut of the Feds) I think it would be best spent doing the Tehachapi portion of the HSR line tying back to Amtrak (BNSF) in Bakersfield and Metrolink at Lancaster - without wires for now. Amtrak trains could then get over the hill on their own (steeper but straighter and faster) line, though possibly needing 2 units instead of 1, providing a competitive service to the current bus connection. Until passenger traffic builds up more, you could probably rent a little space to intermodal trains. A downside is that there isn't much room for more than the current 6 Amtraks each way north of Bakersfield - would need considerable work on the Valley (BNSF) line to increase the number of trains even if the speeds (currently 79 most of the way) are left alone.

I see your point about the Coast and lack of freight interference. And if SP/UP did indeed reduce speeds and super then recovering that would allow some speed increase even with existing equipment. Yes, 10 hours SF/LA would be much more attractive than the current 13, and would allow a reasonable schedule for overnight travel. But if you could convince BNSF/UP to let you run something through Tehachapi in the wee hours I think you could get <12 hours SF/LA now via the Valley. Didn't those Starlight detours (granted, without all the usual stops) make it from Sac to LA in less than 8 hours?

My 10 mph speedup estimate for Talgo is on the low side, because I assumed no change in current tracks. If the line were beefed up and a few curves eased, you could probably get something closer to a 20 mph speedup (as was done in the PacNW). I still don't like the active tilting of the Acela, because of known reliability problems and the fact that you really don't need it unless you want near-HSR (>90 mph). For the kind of Coast traffic you're talking about, 60-79 would be good enough.