I agree with most of what you said, but a few points:
Sent from the planet it has a short distance to accelerate, from the fleet they could place engines on the comet and have it accelerate steadily for months or years. They could definitely target a fleet, but once they get to a certain range it's probably next to impossible to hit them whereas the planetary position is predictable.
I can see your point. I think it takes a couple of assumptions, 1. that it is possible to calculate a long shot (years/months of acceleration) and 2. that the sort of acceleration necessary to intercept it from the planet is difficult/impossible once they notice it is incoming. Not to say I think those are necessarily wrong, I think they are valid assumptions, but they are assumptions

. So, taking these constraints into account, I would say the efficacy of the planet based defence systems would depend on how quickly they could notice and then respond to the asteroid. My argument for this scenario though (the sneaky long-shot) is that it is not the sort of scenario planetary defences would be good for - this is where a defence in depth tactic makes more sense (eg. batteries, sensors and ships on the edge of the system). From a game perspective, it is more interesting if you require a defence in depth approach (eg. orbital, system AND planetary based defences) to really achieve an effective defence.
Definitely. Problem is if you want to keep your people alive as well.
You're right, to an extent. The point of a defence system is to raise the potential cost of conflict - even if you don't anticipate that you'll use all of your options. There is a great Yes Prime Minister episode about the use of Nuclear Weapons - the idea being that even though you possess nuclear weapons as a deterrent, you may not ever use them. So the idea behind strengthened points buried deep within the planet is to suggest to an enemy that it is very likely that in order to succeed in their invasion they have to cause significant damage to the planet, raising the cost of the invasion and thus acting as a deterrent. This is true, even if you plan to surrender the planet before things get to that point.
edit: "Respectfully Disagree"? Oh yeah? Which part?
Well that wasn't me - but here's where I'll go through my respectful disagreement
I think you're conflating too many things here. For one thing, having a space elevator means infrastructure for putting cargo in orbit. It doesn't mean you have an effective anti-space missile battery or anti-space cannons of some kind.
It's a fair point, but I still stand by my original position. My point about getting cargo into orbit cheaply and effectively was a counter-point to those who put the gravity well of a planet up as an absolute barrier to a planet based defence system. Even if the system is slow (say a space elevator), it will be able to get weapons into orbit where they can be used. To be fair, that's moving away from a planet based defence system to a... planet supplied? weapons system.
That said, I don't think it's beyond the realm of possibility to suggest that we will have a system much better able to get things into space at high speeds in the not too distant future. In fact, it's hard to imagine that missiles in the future aren't able to escape orbit easily. If an object can accelerate quickly enough, then the force exerted by Earth's gravity is trivial compared to that exerted by the engine (please don't jump on me physics people

).
Second, it doesn't matter if you have an "effective" ground-based weapon. The point is that it would be a better weapon in orbit. A laser cannon battery, of any strength, is simply a better laser cannon battery if it's on a platform in orbit. Same thing with a missile battery. Why? Mainly because of the atmosphere and gravity. Shooting a laser through an atmosphere is a huge waste of destructive potential compared to firing it in a vacuum. Same thing with firing a missile out of a gravity well. Given the same engine and the same fuel, you can put a bigger warhead on a missile fired from orbit.
Almost certainly true of lasers, but I'm not convinced about other weapons. I would certainly agree that there are advantages (as you have mentioned) but there are advantages to planet based defences as well. On a side note, I am definitely not arguing for planet based vs orbital, I think there is a place for both and it would be sad if Stellaris didn't reflect that.
As for planet based defences - they are easier to hide (I will get to that

) easier to reinforce (bury them in the ground), which makes them less susceptible to disruption from enemy ships. In addition, they will have access to the planetary energy grid, so while they may be less efficient, they may well be orders of magnitude more powerful. Particularly as they can be hidden, they don't make quite the same target-size vs power (stored ammunition, size of the weapon etc.) trade-off that orbital weapons do.
I've seen someone argue that it's "harder to destroy" if it's on the surface. But how? Because you buried most of it under some rocks? And why do you think you can't do that in space? Rocks are really, really cheap and easy to find in space. If you want to surround your weapon with tons of rock, that's trivial to do in space, and you get exactly the same defensive advantages. Plus, your orbital platform has a major defensive advantage: it's moving! You can put thrusters on it and move it around! Even if it's also covered in tons of rocks!
Yes you can collect lots of rock for your weapon platform, but the bigger it is, the more obvious a target, losing it's mobility and small profile advantages. Earth weapons can be buried underneath kilometres of rock and become less obvious. The moving part helps I suppose, but if it's an orbital platform then its movement is pretty predictable anyway.
Yep. At least for organisms with human-like bodies. AI or some other kind of non-animal sentience, who knows.
This point I can't argue with - but I suppose in the context of Stellaris that's less of an issue - we can assume our enemies have bodies and want to physically occupy planets
It depends on your goals and planning scale. If you're only even going to build artificial habitats on/around the planet, who cares if you completely ruin its surface and atmosphere? I mean, sure, radiation from nukes could be a problem, but you can probably just throw lots of fast dark rocks at the planet to fry its surface. Or if you're not planning on colonizing it for decades anyway, nukes work just fine.
Sure, if you're already going to build artificial habitats. Sure, if you're not interesting in harvesting biological, or even surface non-biological resources, then there is no real drawback to glassing/nuking a planet. Assuming those are something you're interested in (and again, in the case of Stellaris I think we are) then having to do those things raise the cost of invasion - which is the point of a defence system.
There's "fast" and then there's relativistic speeds. Even the fastest-moving comets and asteroids are very, very slow. Even a comet on a sungrazing course, whipping around Sol, doesn't even reach 500,000 mp/h, or less than 1/1000 the speed of light. We could get into a whole debate about how easy it would be to create a military projectile that was invisible and capable of decimating or even destroying a planet, and what level of solar system infrastructure you'd need to ensure nobody did that, but it doesn't matter. The real point, again, is that any anti-asteroid defense platform you want to have is better if it's in orbit. It can detect things sooner and more accurately, it can maneuver, etc etc. There's just no reason to put it on the surface.
I agree, orbital platforms have a purpose as well, and detecting a long range projectile that has been given a lot of space to reach some incredible speed would be one of those things. In fact, I would argue a defence in depth approach, with a system of batteries and sensors would be optimal. I believe I have already gone into detail about a similar point above
Nope, just the opposite. The planet is a gigantic gravity well. All things being equal the fleet can launch bigger, faster missiles, because they don't have to overcome the planet's gravity.
You're forgetting that ships are much, much smaller than planets and have to use their limited space/resources for a whole host of other things. All things being equal, the planet can launch bigger, faster and more missiles as it doesn't have to worry about transporting them. This becomes untrue with a sufficiently large fleet - so I suppose the point of contention here would be how large a fleet do you need to match the potential arsenal of a planet?
No way. A fast enough object would have no problem cracking a planet open or even completely destroying it. Destroying anything on the surface is trivial. You wouldn't even need a relatively large asteroid if you could get it going fast enough. There are plenty of asteroids in the 200 km^3 range (asteroids get much, much bigger) with masses on the order of 30x10^18 kilograms. If you have space travel technology and you can accelerate it up to even 1% of c, you're going 1000 times faster than any asteroid we've ever seen, and the impact damage would easily knock a chunk off the planet. Not to mention that you could easily aim it for a direct impact, not an oblique entry from orbit.
If you go deep enough... but I do see your point. In that case, the bunker has achieved its objective in preventing occupation/use of the planet - raising the costs and decreasing the benefits of invasion.
Depends on the projectiles (rockets are going to be detectable as soon as they leave a silo because of their heat signature), but either way it's equally true of hiding projectiles in a dark satellite. Except your dark satellite can also move.
On this point I absolutely disagree - though dark satellites, or even inactive rockets in orbit would be effective. But, one shot weapons aside, Earth is much better at hiding weapons. Firstly, a planet, especially an advanced one, will have a whole heap of heat, and in fact, with the storage capacity of a planet it wouldn't be hard to generate false ones, or so many that it would overwhelm a defence system. So, you have to have, on your fleet a system of sorting through all of the heat signatures to find ones that aren't false positives.
Now, I'm going to go out on a bit of a limb here, so bear with me, or feel free to (respectfully) tell me where I have made erroneous assumptions

I don't really know how you detect heat signatures - but let's assume that you have to be able to scan the surface of the Earth to detect a signature - that's about 510 million square kilometres (so wikipedia informs me) you have to scan. You then have to track matches, discarding false matches. Working out the source of the weapon would require yet more computation. This would require a huge amount of computing power, which would have to be dedicated solely to the task of detecting incoming projectiles.
Of course, you can cut down this power by only focusing on a smaller area, or by looking at objects in space where there are presumably fewer sources of noise that have to be dealt with.
But more importantly, the fleet has similar advantages. You can easily throw rocks at Earth from behind the Moon, and nobody is going to see you.
Sure, if a fleet can get behind a blocking object then it might be fine, but given how cold and empty space is, it's much harder to hide a ship than an object on a planet which is hiding a warm object in a warm environment. But that point about sneaky rock lobbing I think goes back to the point that there is a place for system, orbital and planetary based weapons and sensors.
No way. Planets are death traps.
I am guessing we won't see any of the considerations discussed above. It'll all be magic laser cannons shooting from the planet into space. But that doesn't mean it won't be a fun game.

Like many other Paradox games, reality is considerably less exciting.
I suppose my main point is that planetary defences have a place, which is mainly system denial against a fleet, whereas orbital and system based defences are better against long range bombardments (assuming low levels of acceleration (relativistically speaking

).
I hope Paradox surprises us all and gives us something awesome here instead of magic laser cannons

.