Just two unsophisticated observations.
1. Granted that government by referenda is a bad idea, if you are going to have them at all, then major infrastructure projects are sensible topics. Especially this one; once you have spent the first $5bn on a stretch of high-speed rail, economies of networking and me-too politics will make the complete $45bn inevitable. And spending $1bn buys you nothing useful, unlike with roads. So be aware that Proposition 1A is really a referendum on the whole thing and that for three years there will be little to show for the money.
2. Kevin Drum is simply wrong that 2½ hours is infeasible for a train from San Francisco to Los Angeles. SF to LA is 347 miles direct, say 400 on the actual route. The Spanish AVE from Madrid to Barcelona does the 386 miles in 2 hr 38 minutes nonstop, on a 100% new track designed for 300km/hr (186 mph). The newest French line from Paris to Metz runs trains at 320 kph (200 mph); but it’s designed for 350 kph (217 mph), just awaiting the next generation of trainsets. China already runs trains at 350 kph on the Beijing-Tianjin line. The Californian plan for 220 mph running is conservatively based on tested technology, without any futurology.
I suspect KD’s disbelief over speeds comes from a misunderstanding of the technology. Trains on dedicated high-speed lines operate more like aircraft than cars; they reach cruising speed quickly then stay at it until they slow down for the next station. The high power-to-weight ratio of the trainsets mean that the speed is maintained on slopes that are too steep for conventional trains. They don’t do bends well so the lines are straight, vertically polarized as it were like Roman roads.
Where construction estimates go wrong is in long tunnels, since the detailed terrain is unknowable in advance. The Madrid-Malaga line goes through, and partly under, the Sierra Morena (1000m) and the Sierra de Chimenea (1200m); the Barcelona-Perpignan and Florence-Bologna lines under construction go under more considerable ranges. The tunnels near San Francisco will come in late and over budget.
Kevin may be right that ridership estimates are optimistic, but the story that ridership on the TGV Paris-Lyons is only 15m a year is out by 10m. (Here, page 53. The number of 25.5m is for TGV Sud-Est passengers to and from Paris in 2003, but there are essentially no intermediate stops before Lyon.)