2) The blocks from both the attacker and honest miners build on top of each other in a single chain. In this this situation the honest miners include all transactions skipped by the attacker, and the attack fails.
Obviously the attacker will not do that. It will ignore all blocks that the orthodox miners put out, and continue mining empty blocks using
his last mined block as the parent. Since the attacker has more power than the orthodox miners, his empty chain will eventually outpace every orthodox side branch, and all the orthodox blocks in the latter will then be orphaned and discarded
by all the orthodox clients.
Thus, every transaction request that the orthodox clients issue will either stay unconfirmed forever, or become unconfirmed even after having been confirmed several times. Whereas the clients who upgrade to the cartel's version will not notice any difference.
Obviously, and so we are back to what I already stated, which apparently you didn't bother to read.
1) There are two competing chains where one chain has only the attacker's blocks and the other chain has the honest miner's blocks. The goal of the attacker is to make their chain with their inclusion rules the longest. The network here can optionally choose the honest chain (which is what Gavin's proposal does). It also require 51% to pull off, which I don't think is possible anyway.
Which is the only possible path. This path both requires at least 51% of total hashing power, and it requires that the honest network (users, P2P nodes & miners) to ignore the 2nd chain with all transactions included. Neither of these are likely IMHO, and why the attack fails.
Bitcoin is an ecosystem of human participants, not an autonomous machine that we have no influence over. Example during the March 2013 fork the autonomous machine decided to take the 0.8 branch, but a majority of human participants decided the 0.7 branch was a better path and shifted the machine back to that (causing a 30+ block re-org btw).
If a 51% attack ever happened it would be very visible and resisted by the majority of human participants, who all want bitcoin to succeed, and who would manually choose the shorter chain made by honest miners as the valid path. Gavin's proposal would automate this selection. A 51% attack, which is what you are relying all your arguments on, is both extremely unlikely, but also ineffective. Bitcoin would easily survive a +51% attack, but it will never have to because it is not going to happen.
But thank you for reminding my why I bailed on the academic path years ago and decided to live in the real world. You, like most academics I've encountered, seem to believe that writing long-winded obtuse arguments in a vacuum consisting of your own reality, somehow makes you intelligent or valuable.
You are not, you are like a colossi striding over a very small ant-hill.