Even if not "cash-flow neutral", if any significant portion of the cost of a derailment is externalized, then carrier's appetite for derailment risk will be higher and there will be larger trains.

The solution here would seem to be penalizing derailments using a large fine that is proportional or progressive to weight or length.

I'm not sure what portion of derailment costs are externalized. I've seen a few in town, none that involved hazmat spills thankfully, mostly coal trains. But the repair of the rail and track bed, hiring of services to lift the derailed cars back onto the track, repair of damaged cars, etc. is all going to be on the railroad. Loss or damage to the cargo would be on the owner of the cargo as in any shipping scenario, but they should be insured for that.

If there was damage to private property, a hazmat spill contaminating private land, or injury or loss of life, the railroad is going to get sued.

The major specific externality called out in the letter is engineer burnout. I would be surprised if other costs are not significantly externalized, but I don't have data to back this up.

Not all the externalities mentioned are specific to derailments, but also apply to the general difficulty of our infrastructure in supporting such long trains.

Your reasons are unconvincing. A rail company that burns out its engineers now has to either deal with a low performing engineer or come up with the costs and delays of retraining another engineer.

And a derailment sounds like a massive cost to a rail company. You need to clear the tracks (which can't be used until you do), probably pay higher insurance costs for the material you are transporting that is lost, etc, etc