Why?

One of my first thoughts on the derailment in Glendale last week was, wouldn't the train normally smash the vehicle, not the other way around?  According to Metrolink chief David Solow:

"Normally, in grade crossings, the car is smashed into smithereens, and the train is still on the track," Solow said. "Normally, we'd have little damage. The reason he couldn't get off the tracks was because he was wedged on the tracks."

Lessons learned have been a common topic of discussion. The lessons appearing most frequently appear to be about the safety of allowing locomotives to push from behind rather than pull from the front and the need for grade separations.

The debate on the push-pull issue is whether a cab car placed in front is heavy enough to protect itself and the rest of the train.  Experts disagree on the answer--therefore, no lesson learned yet in that area.  However, outlawing push operation would be terrible for a service like Caltrain, which to my knowledge has no capability of turning around locomotives for every one of its SF-SJ runs.  How could instituting a pull-operation-always policy require anything less than huge handicapping of rail service?

Alternatively, if they are not already, couldn't the leading cab cars just be engineered to withstand collisions as well as locomotives can?  And, as for precedent, BATN has this comment:

BATN notes that these "lighter-weight" cars weigh more than many locomotives used to out in the civilized world. Of course, the very idea of attaching a hugely over-massive, dead-weight, non-revenue-earning, non-passenger-carrying "locomotive" to a passenger train is a quaint historical anachronism on systems which carry non-trivial numbers of passengers. How many locomotives does BART operate?

Then again, BART is completely grade-separated, which eliminates the risk of hitting crossing vehicles.  But applying that approach to traditional railroads requires funding and physical space for the separation.  Often, we don't have either.  Where other than local, self-help taxes do we find money to pay for these separations in these days of budget deficits in both California and the US?  And elevating or depressing roadways sometimes requires removal of buildings, as is being done in Fremont to make way for Irvington BART and the Warm Springs Extension.

So, what solutions can be employed realistically?  Tracks and grade-level crossings are almost everywhere, and modifications are expensive and not always compatible with the community.  Is this a reasonably solvable problem?  Is it appropriate just to say that rail transportation comes with a certain level of risk, as do all modes?  Although the Glendale accident was tragic, are our fatality-prevention efforts best spent on improving rail safety or on improving safety of roads, which have a higher fatality rate than trains* and the users of which are killed on a much larger scale?