How would such a proof protocol work? I spent most of yesterday trying to answer that question and came up with nothing. Imagine that you and I are mining and we've agreed to use the ES "selfish" strategy. When I want to getWork, I have to choose which branch I'm targeting as the parent of whatever block I'm trying to mine (I have to choose which hash to put into my block). But how do I prove this to you? I could send you a commitment to the hash and you could promise to come beat me up after the fact if I lied, but that's outside the stated protocol. In fact, if your model is that you can coerce me to follow your protocol when it's better for me to do something else, then you may as well coerce me to give you all my bitcoins.
Our argument is that the protocol, as stated, is not incentive compatible. That's ironic, because the protocol, as stated, is offered as an argument that Bitcoin is not incentive compatible.