The one hope I had for a seismic early warning system (actually, it wasn't much hope) would be that a real-time GPS network would catch the pre-slip of a large fault system.
For a large fault, you never know at the beginning if the rupture will continue, or how far. Thus, you have to wait for it to finish. That's a critical 60 seconds. These GPS things are slow creatures, sampling at 5 times a second, so it takes another while for them to report.
All faults need to pre-slip to the 'critical displacement'. Had this signal been caught, you would know that was going to be large. Then you get more time, for whatever good that would do. You also want to get the fact that the rupture was travelling towards the city which ups the PGV by a factor of two at least (directivity).
All in all, another nail in the coffin of early warning. :)