Sorry for the delay in replying to this sincere post. I got inundated with noise as you can surely see above.
I haven't gone through the entire thread but here's my provisional view on it anyway:
The kind of attack described seems plausible, but only if there's no mechanism for checking if the blocks being propagated over the network actually confirm transaction requests that were previously propagated over the distributed memory pool. Anonymint, your theory seems to assume that there are unsafe levels of trust between nodes. This is my impression of how it goes:
-a node has transaction requests X, Y, and Z waiting in its copy of the queue.
-It then asynchronously receives a block with X and Y on it, but not Z (maybe the fee was a bit low for that one). Everything seems OK according to the node, so it accepts the block and helps propagate it by sending it to any connected peers that might not already have a copy.
-the node then asynchronously receives more transaction requests: U, V and W, which it adds to its local copy of the memory queue.
-the node then receives an "Amazon block" which claims to confirm some strange transactions: Q, R, S, T, in addition to the known ones: U, V, and W. Depending on the exact set of rules that the node is following, maybe it will provisionally accept the block if it forms part of the longest known chain, but it's deemed suspicious so the node doesn't forward it onwards to its peers. This lack of forwarding should mean that strange blocks tend to get orphaned, and this protects against man-in-the-middle attacks. On average, blocks that play by the majority's rules are able to propagate further and faster throughout the network, enabling a longer chain to build on each.
Nice try, but this is technically incorrect.
The orphan rate only affects a very tiny % of the blocks found by necessity otherwise the network would diverge into forks.
If you are arguing that Amazon doesn't have the connectivity to propagate its blocks to pools where the hash rate is concentrated, I think not many would agree with you.
Thus most of the time Amazon's (the cartel's) blocks will propagate.
If you said Bitcoin would alter the protocol to refuse to process Amazon's customers payments by refusing Amazon's blocks (any block where they hadn't seen already many or most of the transactions in the block), I think it would harm Bitcoin immensely. The crypto-currency needs to be all inclusive otherwise the potential for strong competitors and forking arise. The protection against this attack is simple! Zero transaction fees and perpetual coin rewards. So simple.
Who wouldn't love a zero transaction fee coin! The main use of Bitcoin other than speculation is money transfers.
The rest of what you wrote thus doesn't apply.
See this is yet another example of why people think I don't understand, when in fact their technical knowledge is too limited to realize that I understand and they have missed something that causes them to be overconfident in their analysis.
The proof in the real world will come soon. Enough of this talk!