Capitulation is, in most cases, an alternative to wiping out the civilization entirely. In order for this to be useful, you need to know with sufficient certainty that your vassal is not going to randomly turn on you and go to war, cutting off all your trade routes and deals with them (particularly if you're sanctioned).
I disagree with pretty much everything in that statement: firstly, the Vassal provides many benefits, which can be enough justification for "getting" him even without sufficient certainty that he's going to turn on you; secondly, even with how it works now you don't have that certainty, since he can, in fact, turn on you once he fulfills the conditions; if we implement the impossibility on Content status then you can create your sufficient certainty as well. Thirdly, if you're sanctioned and have no Vassals then you can only trade with City States (unless that is disallowed as well), which are much further apart and therefore even more dangerous to trade with...only with Vassals can you have a trading partner with "sufficient certainty". Fourth, it's not true that the alternative to vassalization is wiping a Civ out in most cases, even after reaching a Warscore of 100, particularly if you're playing Autocracy and thus get a bonus to your Warscore accumulation; I have managed to vassalize a Civ after capturing only one of their Cities, for example, while wiping them out would have taken a lot more effort so without a vassalage system I would have simply peaced out...often times you want to end the war and sometimes you actually have to, because someone else is attacking you, so you don't even have the time and resources to finish off the first Civ, because you need to bring your forces to the new enemy; here you would certainly vassalize even if wiping them out was the intention and you don't like the vassalage system. In fact, most wars where I win with a large Warscore are wars where I don't have any interest in wiping the Civ from the map, which would even be true if I couldn't vassalize them; for me a very successful war is one where I got what I wanted (e.g. strategic location, resource, vassalage, capital, holy city etc.), significantly weakened my opponent and did it with conquering as little as possible.
If the benefits of vassalization are not consistently controllable (other civs being able to steal them away by trading with them is not controllable) then most players would choose to eliminate the vassal entirely.
Wrong, as I already laid out above.
Making content vassals unable to rebel creates a loophole for human vassals (which do not use a vassal treatment system).
Humans also don't use the approach system or opinion system, yet we make it work. Furthermore, human vassalage is awfully rare. Lastly, the human still has the ability to decide if he wants to accept another Civ's "liberation war" offer, which will include an evaluation of how he's been treated by his Master and how powerful the offering Civ is (doesn't make sense to accept if you'll just lose the war and have to be revassalized for a minimum 50 turns again), which is, after all, what the AI would do. I personally don't care too much about Content status making liberation wars impossible or not, as long as it makes it far less likely, but given the rarity of humans being vassalized to AI at all, I'd say let's not die on this hill when trying to make humans be exactly like AI in their choices...after all, you concede in your diplomacy thread that it can make sense to deviate from this (admirable) goal in some cases.
Furthermore, a system like this requires the AI to be programmed to make suboptimal decisions based on vassal treatment level, because logically most capitulated vassals would want independence regardless of treatment, unless they'd be wiped out. I have a personal distaste for this kind of thing when it isn't used to lower game difficulty, but aside from that I think it would be a frustrating mechanic as you proposed it, because people would get annoyed if other civs could swoop in and take their vassals if they weren't content every time 50 turns went by.
Apart from the repeated hyperbole of "Civs swooping in to take your Vassals every 50 turns", which I've addressed above, the current system is actually far worse in this regard, because the AI, unlike the human, can't make deals with other players to prepare for the "Great Liberation" after the conditions are met. If I was a human and had to submit for vassalage with an AI, I would absolutely make a deal with another human (or AI if I can) to engage in temporary vassalage with them (if that's what they want in exchange) on the same turn that I declare independence if he will help me kick my old master's butt. The AI doesn't know how to do that, because it can't pre-agree on future deals and can't plan ahead this well, so this deal system is likely more AI-friendly from the (anthropomorphized) perspective of the Vassal AI, because it at least gains a powerful ally to make sure the liberation attempt doesn't just end in a pointless, suicidal rebellion with the world watching, but has an actual powerful ally to help it.
If they become a voluntary vassal of the other civ they can then leave after 10 turns and gain independence, so what you're proposing is essentially a third type of vassal with a significant amount of new code.
No, what I propose, as I have stated before, is a new kind of trade deal, which would be very similar to the existing "Voluntary Vassalage" deal; the resulting Voluntary Vassalage would just be a normal Voluntary Vassalage, so there is no third type of vassalage involved here. If
@Gazebo likes this idea but doesn't have time for implementation I could even take a look at implementing it myself, since it's really similar to the Voluntary Vassalage deal anyway.
I can understand the desire to make vassal revolts more interesting and I think it could be fun to allow other civs to support a rebellion, but they can do this already by providing gold and resources or attacking the master while a revolt is in progress.
It's not just about making them more interesting but also about making them less OP for warmongers, which they currently are; I simply want to "nerf" Vassals in a more interesting way. I think the fact that I and many others haven't seen revolts in a long time is a good testament that this would be an area where things could be improved.
If a vassal revolts, they could perhaps try to become a voluntary vassal of another civ. I think a mechanic like that would work better; honestly, being able to push around your vassals and demand things from them is part of the fun, and the hostile spy/WC actions and potential diplomatic repercussions are sufficient punishment for mistreating them. If mistreatment allows other civs to steal your vassals I don't think most people would find that fun.
Except the AI doesn't do that...it theoretically can, btw., as Voluntary Vassalge is already an existing feature, but this would require the AI to be extremely lucky, as it can't "feel out the waters" beforehand and just has to hope that a closeby AI wants war with their old master anyway and has been preparing for it just in the last turns...my proposal eliminates this problem because it implicitly coordinates the two AIs, especially with regards to the bigger, deal-offering AI. Regarding the second point...I've already addressed that, but a quick summary: in my experience it's definitely possible to have your vassals take the friendly approach, which means they often won't spy on you at all (at least until the game is close to be concluded) or if they do you can successfully ask them to stop; if you push them around, on the other hand, they will eventually be able to demand freedom anyway, so all my proposal would do is introduce some more risk before that point; why only "some more risk"? Because getting 25 Warscore as a warmonger isn't usually that hard against the AI and neither is keeping your Vassal Content. Moreover, it's simply completely unrealistic that the Vassal would just sit there and take your mistreatment for centuries if there are Civs willing to aid him free himself...ending it all in a possible pointless rebellion a millennium later is just sad. I have no problem with players going the more "sadistic" route and torturing their Vassals (though I typically don't) but if they do then there should be more risk to it; the main motivation for all of this, however, is to apply a bit of a nerf warmongering in a more interesting manner.