As far as being a superior choice, I don't think it's quite apples to apples. Project Orion (nuclear pulse propulsion) was designed to get very large payloads directly into orbit. Nuclear thermal rockets like NERVA were apparently designed to be used as drop-in replacements for upper stages on conventional chemical rockets, or as space "tugs" in orbit. That might imply that it was difficult to design a nuclear thermal rocket which would deliver the high initial thrust(?) necessary to lift a payload directly out of our gravity well (but IANA rocket scientist, so I may be mistaken on this). That said, a small blurb on Wikipedia does mention that "Robert Bussard proposed the Single-Stage-To-Orbit "Aspen" vehicle using a nuclear thermal rocket for propulsion and liquid hydrogen propellant[...]" -- so maybe it would indeed have been a superior design choice in the end.
There is a safer and cheaper way to harness nuclear bombs for heavy lifting, though, as long as you don't mind a very bumpy ride for the cargo: the Verne Gun (http://nextbigfuture.com/2010/03/150-kiloton-nuclear-verne-g...).
Let's say we have a unit power output of P_therm and P_bomb (in case of the bomb its time plot should look close to dirac pulses while P_thermal should look like a flat line), we have efficiencies eps_therm, eps_bomb and we have the mass of the propellant systems m_ohterm, m_ohbomb.
I'm assuming P_thermal to be 1e+7W (1/100 of a powerplant), P_bomb = 0.15 kiloton per second ~= 6e+11W => lets make that 1e+12. Efficiency: eps_therm = 1'000 * eps_bomb Mass: m_therm = 10 * m_bomb
so if these numbers work out the bombs give you about 10^5 times the power output of the furnace, deliver 10x the power to the ship (efficiency differences) and yield about 100x the effect per unit of mass.
hmmmm... I think I'm starting to get it ;). However here's the thing: I'm imagining that the furnaces can be scaled up much easier than the orion design. You can just make them bigger up to a certain point and when that doesn't help anymore you can strap them together. You can't really strap together the orion propellers because it would become incredibly susceptible to timing (one bomb goes off 1ms earlier and you end up spiraling everyone to death with 20G acceleration or so :D ).
The timing isn't so critical. It could handle a 10ms timing variation, and a misfire.
George Dyson's book, Project Orion, is well written. That being the son of Freeman Dyson, who worked on the project.
Orion was intended to be safe for humans, as long as the G-forces of the pulsed detonations were kept in check.
Nuclear pulse propulsion is something different entirely. It has vastly higher exhaust velocity and vastly more thrust but has very different design requirements than NTRs. With it you could put enormous payloads into orbit (think battleships and aircraft carriers) if you were ok with the fallout, or you could intercept dangerous asteroids or comets with very short warning times, or you could launch payloads to neighboring stars at truly enormous speeds.