What do you mean we, Kripke?

I didn't see Mike's post on the identity stuff until now, so this isn't at all a response to it, it's just my take on the Kripke paper. Here's what I think is going on:
According to identity theorists (IT), pain is identical with a neural state of type x. Those who oppose IT argue that we can imagine pain existing without that neural state, and therefore it can't be the case that pain is identical with a neural state. The standard IT reply is that this possibility of pain existing without that neural state doesn't justify a conclusion that pain is not identical with a neural state. Rather, this possibility shows that the identity statement is contingent, rather than necessary, and this shouldn't worry us because this is just another contingent scientific identification, similar to the identification of heat with molecular motion.
According to Kripke, there is an important difference between the identification of pain with a neural state, and the identification of heat with molecular motion. True identity statements of the form 'x is y' have the following setup: We observe something by way of some sort of description. We then give a name to the object that matches that description, and that name 'x' rigidly designates the object. Now, here's the important thing: If the description that we use to pick out the object rigidly designated by 'x' is an essential property of the object, then, if it is true that 'x is y', 'y' must have that essential property as well.
Often what happens in the case of (what Kripke thinks are mistakenly labeled) contingent identity statements is that we pick out 'x' with a contingent property, pick out 'y' with a different contingent property, and then discover that 'x' and 'y' rigidly designate the same object. This is what happens in the case of the identification of heat with molecular motion: we pick out 'heat' with the contingent property of producing a certain sensation (Kripke argues that this is contingent because we can imagine the existence of heat without it being felt, so therefore the sensation of feeling heat must be a contingent, rather than essential, property of heat), and pick out 'molecular motion' with a different contingent property, and then discover that 'heat' and 'molecular motion' designate the same object.
However, according to Kripke, pain is different: "Although we can say that we pick out heat contingently by the contingent property that it affects us in such and such a way, we cannot similarly say that we pick out pain contingently by the fact that it affects us in such and such a way." That is, Kripke has the following intuition (to be fair, derived from a thought experiment): There is a crucial distinction between pain and heat. We are able to imagine the existence of heat without it being felt, so therefore the sensation of feeling heat must be a contingent, rather than essential, property of heat. We are unable to imagine the existence of pain without it being felt, so therefore the sensation of feeling pain must be an essential, rather than contingent, property of pain.
If Kripke's argument works, it seems to undermine the IT reply to the objection that points out that we can imagine pain existing without that neural state. Remember, the IT reply is that this is just another contingent scientific identification. But Kripke has argued that contingent scientific identifications are true identity statements insofar as it isn't the case that [we pick out the object designated by 'x' with an essential property of that object, yet we can imagine the object designated by 'y' lacking that property]. And in the case of pain, Kripke has argued, we pick out 'pain' with the essential property of feeling pain, yet we can imagine the neural state (say, C-fiber stimulation, although that's not what neuroscientists think anymore, right?) lacking that property.
When I first read this article, I was extremely frustrated by it. To some extent, I still am. But I think I'm understanding the argument a little better, and it's a pretty cool way to block the IT reply. My problem with the argument is that it relies on notions such as "essential" that I can't make sense of and weird modal intuitions. How do we determine whether a certain property is an essential property of a particular object, or whether it is a contingent property? Is it dependent on what we would call that object if it lacked that property? Who is the "we"?