For one thing, you could weaponize asteroids as a second-strike response, if there's an imbalance where one superpower has asteroid deflection and one does not. Even if it's delayed by months, the certainty that the asteroid will eventually hit establishes a deterrent against attacking the asteroid-controlling country. If there's defenses against ballistic missiles, this circumvents those.
For another, the asteroid-deflecting nukes are dual-use as weapons themselves. A nuclear warhead in deep space defeats conventional ballistic missile defenses. If that warhead is turned towards Earth, that (first-strike) attack's not visible to the adversary until a few seconds before impact—unlike an ICBM launch, which has tens of minutes of warning, enough to launch (conventional) interceptor missiles.
I don't think a second strike from asteroids would work. You can't just arbitrarily redirect an asteroid towards Earth whenever you want. You need one that's already going to pass nearby. You might have to wait decades for it to hit, and during that time your enemy can send their own nuke and nudge it away.
But if you could do second strikes with asteroids, that would be a good thing. Second strike capability helps prevent nuclear war. Effective first strikes are what's destabilizing.
Which brings us to your second point. I agree that stationing nuclear warheads in space is a bad idea. But we don't have to do that. With years or decades of advance warning, we can just launch them from Earth.
- "Second strike capability helps prevent nuclear war. Effective first strikes are what's destabilizing."
I don't disagree!
- "You need one that's already going to pass nearby."
There's a very large number of these already, within a small delta-v of Earth. We don't know where they are yet (it was a show-stopping issue with NASA's Constellation (?) program, when they wanted to demonstrate capturing an asteroid, but couldn't find one), but that's going to rapidly change.
- "and during that time your enemy can send their own nuke and nudge it away"
That's pretty useful: that brings it to a point where the defender has to invest an amount of resources comparable to the attacker. That's a win for the economically stronger country. If the difference grows large enough, they can simply overwhelm them with numbers.
That's even before opening the technological possibilities of stealth asteroids.
Eh, we already have steath aircraft that could slip in past air defenses and drop a nuke with next to no warning. We probably have a stealthy hypersonic that can also do so. That doesn't negate a nations second strike ability.
Part of the reason the US, Russia and China have vast land based missile fields is to visibly show the enemy that they're gonna have to wipe out huge swaths of land with hundreds of nukes (and THEN sink our boomers) to have a successful first strike.
Right; but if the enemy thinks that's likely to happen, it'd only take them 15 minutes to nuke that one airbase that stores all the B-2's. Strategic bomber aircraft aren't very useful weapons for superpowers fighting each other (or else we'd have a lot more B-2's).
Space-based weapons (which are banned by treaty) are more like submarine-basing, in that the enemy isn't necessarily sure where they are; but, in addition to that, they have potentially very short warning times before a strike. No one actually wants that unstable dynamic, with the short/no-warning; which is why all the nuclear powers have (so far) agreed by treaty not to build any.
edit: But if someone were to build a civilian spaced-based asteroid defense, which involved sending nuclear weapons into space... that could get destabilizing fast. Whether or not Sagan can predict specifics, the big-picture idea of humans not trusting each other with apocalyptic weapons is not wrong. (Worth recalling the USSR genuinely mistook the civilian US Space Shuttle for a nuclear weapons platform).
Forgive my naivete, but do we really have the precision measurements and kT payload calibration needed to alter an asteroid's trajectory and speed to the point that we can target specific (even very large) country landmasses as the Earth's rotation exposes them to the oncoming projectile?
Nuclear second strike has always been from the submarines, which guarantee a response within minutes.
Versus your proposal here of launching a highly visible rocket, to manipulate a highly visible object, with a flight time measured in days to weeks that's mostly ballistic.
The physics of tomorrow though will be the physics of today however: a many months flight time out to an orbital object to affect trajectory change is a response time so far away as to be meaningless: any opponent with an ICBM strike capability on Earth could happily blast away your entire civilization, and reasonably expect to use the same technology that enables that ICBM capability to blast away (or deflect) your asteroid.
Put it another way: for considerably less expenditure of effort, you could just build a hydrogen-bomb large enough to obliterate Earth's biosphere completely[1]
Consider that ICBM's might not work forever. If e.g. new types of anti-ballistic missile defenses were to become cheap and reliable, then ICBM/SLBM's could even become obsolete (say, within the century). If that happened, there'd be a slew of new arms races, which are unimaginable to us, being beyond our technology horizon.
Physics isn't changing over time, but engineering is.
For one thing, you could weaponize asteroids as a second-strike response, if there's an imbalance where one superpower has asteroid deflection and one does not. Even if it's delayed by months, the certainty that the asteroid will eventually hit establishes a deterrent against attacking the asteroid-controlling country. If there's defenses against ballistic missiles, this circumvents those.
For another, the asteroid-deflecting nukes are dual-use as weapons themselves. A nuclear warhead in deep space defeats conventional ballistic missile defenses. If that warhead is turned towards Earth, that (first-strike) attack's not visible to the adversary until a few seconds before impact—unlike an ICBM launch, which has tens of minutes of warning, enough to launch (conventional) interceptor missiles.