Here is a back-of-the-envelope calculation for just one of the societal benefits of a high-speed rail system:
When a person drives a car, it costs about 30 cents per passenger-mile in external costs (i.e. costs that are incurred by society, rather than by the driver). These externalities include pollution, road maintenance, highway police, emergency services for accidents, congestion, land use, etc.
So let's use that Tampa-to-Orlando route that keeps being cited as an example of an absurdly expensive boondoggle, with that $3 billion price tag for an 86 mile track. And let's assume a payback period of 20 years (which seems reasonable, given that a high-speed rail system is intended to be a long-term investment). How many people would need to make the round trip between Tampa and Orlando each day via the train in order to justify the cost? Here's how we could calculate it:
N = The number of customers per day
$51.60 = The total cost to society for each round trip in a car ($0.30 per mile * 172 miles)
N * $51.60 * 365 days * 20 years = $3 billion
N = 7,964
So roughly 8,000 people would need to make the round trip each day in order to recover merely the EXTERNAL costs to society...and that's not even taking into account the money that would actually be saved by the individuals making the switch. Nor am I including the non-driving related externalities, such as airport congestion.
How many people travel this route on the highways every day? I bet a hell of a lot more than 8,000. If you could convince a fraction of them to switch to rail, even this route being cited as grossly overpriced would be well worth the cost over the course of 20 years.
And I bet the numbers are quite a bit better still for places like California and the Atlantic Corridor.