James Dewar's latest book he proposes the development of a new solid core highly enriched uranium rocket engine based on the B-4 core developed in the Rover/NERVA program, but unlike that program he recommends starting small, testing in a dedicated exhaust processing facility and building successive generations of engine to prove safety and gain operational experience.
The first engine to be put into operation would have 40,000lbf (800MW), an ISP of 1,000s and weigh 6,000lb. It would have a maximum burn time of 15 minutes. The gross mass for the stage would be 91,000lb with 45,000lb of LH2 fuel, and a 3,000lb cocoon to recover the engine, to deliver a 17,000lb payload to LEO*. The stage would be dropped from 50,000ft by a cargo plane (such as the C-5A), and solid rocket boosters would carry it to 100,000ft before the solid core engine engages. The deorbited engine in its cocoon would be recovered from a splashdown for processing, as the U-235 would only be ~1% spent in the short burn required for orbital insertion.
This is just the first operational vehicle, bigger payloads to justify the cost would follow - Dewar makes recommendations for other rocket engines in the book, but this is his pièce de résistance.
Obviously, there are a lot of political issues to overcome before this could ever be funded at a government level and, excluding Bond villains, it almost certainly would have to be a government program. But what are the technical arguments against this? What are the (non-nuclear-hysteria) safety arguments?
And, to counterbalance the risk, are there good arguments for such a program?
These are not rhetorical questions, I want your opinion.
* in metric: 177,929N (800MW), 2,722kg. Gross mass for the stage would be 41,277kg with 20,412kg of LH2 fuel, and a 1,361kg cocoon, to deliver a 7,711kg payload to LEO.