4. and 5. (no clue how this relates to 5) - collusion on its own is irrelevant. Of course it will be ineffectual. It would always be coupled with the compromise or the DDoSing of additional "masternodes". In such an event, the number of available masternodes is massively reduced, which means the the little Google spreadsheet is incorrect. Many hosts will null-route a dedicated server (even more so a VPS) if it is experiencing a DDoS. Other attacks such as a DDoS group constantly requesting a stream of blocks from a masternode would lead to its traffic exceeding normal use and massive traffic bills. Similarly, a DDoS group could broadcast thousands of bad blocks a second to a masternode, leading to it chewing through CPU cycles doing very heavy X11 verifications, ultimately leading to it becoming inaccessible or being killed by the oomkiller. Those are just off the top of my head, of course, and the attack stands regardless of whether these specific scenarios can be mitigated against.