Yeah I've ridden the Zephyr from Chicago to Colorado (and back) a few times, and it was off-schedule about half the time, one of them was at least 4 hours of just sitting there in the train in the middle of nowhere.

One hour I can understand, four is ridiculous. The trip is pleasant and the views are great but for me to take it again they're going to have to do better than 50% on-time.

I used to live on a small town on the Empire Builder line a few hours west of Chicago. It was once a day each way, and we'd get one direction in the morning, and the opposite direction in the afternoon. Or that was the theory. Often the morning train was so late that when you heard them sound the horn you wouldn't know if it was the afternoon train or the "morning" train.

The "Late Shore Limited" I've taken has on-time performance of ~50% and train interference as 50% of delays. That sounds about right.

Apparently, according to your 2nd cite, that law was determined to be unconstitutional.

Sounds like something that could be enforced by giving the passenger trains the ability to lock out switches further ahead of them. Like, say, an hour ahead of them. (Though, without signalling the particular switches that lower at-grade crossing guards. It’d be a rather fraught thing to engineer, all-told.)

Sure, you couldn’t run densely-packed passenger trains very well on such a system... but they’re not densely-packed. If they were, the system would have been a success and could then be turned off. ;)

This is starting to sound like distributed systems ... how do you prevent deadlock? https://wiki.openttd.org/Deadlock :-)