Most of my work deals with radio propagation, specifically with the Department of Defense in terrestrial and celestial applications. I have had similar discussions with RR signal engineers relating to this issue due to my related work with the DoD. Without going into specifics, the concept behind a GPS based system is sound and technologically feasible. That said, however, non-military implementations are at a significant disadvantage due to their inability to access the full feature set of GPS. Additionally, military implementations, which are not at this disadvantage, still will not rely upon GPS as a primary guidance source due to unforeseeable and unmitigated errors which can be introduced, therefore, it is almost always used as a secondary guidance source.

Then, there is the military doctrine of acceptable losses when the system makes mistakes. The effects of the errors introduced into the system are somewhat predictable and measurable, therefore the military can predict, with some degree of accuracy, what losses will be incurred due to the errors.

In civilian implementations, there are no acceptable losses. This difference in ideology is where relying upon a GPS based system becomes a problem. Additionally, the errors are introduced from a myriad of sources, including errors in probabilistic logic used in filtering (somewhat like a bit-error rate), slower than needed update on ephemeris data, errors in ephemeris propagation due to non-homogeneity in the earth's density, sensitivity of L-band GPS RF transmission to ionospheric disturbances, or countless other effects.

Finally, OPB's point , as seen over many posts on the topic, has been a GPS based system is not worth the expense when compared to other existing technological upgrades which are available. GPS, whether for train control or any civilian application, is cool, "cutting-edge" technology. You tell someone something is GPS based, the automatic assumption is it is somehow better because it uses GPS, when really, it may not be.

It can be portrayed as a sexy technology when placed in front of politicians, who then become infatuated and push the technology no matter the cost. But the delta between existing technologies' capabilities and the capabilities of GPS based system is not significant and thereby does not justify the additional cost. The real point, the paradigm through which this needs to be viewed is not one of coolness, but this simple phrase I use all the time - "Just because we can do something, doesn't mean we should."