[A majority cartel] can only force soft-forks, hard-forks are ignored by full-nodes and clients. An attempted forced-hard fork results in hostile miners forming an alt-coin with no users. The limiting factor is soft-forks are quite flexible and can do a lot, some of which could be undesirable.
Perhaps I did not explain myself clearly. Here is a more detailed scenario:
http://www.reddit.com/r/Bitcoin/comments/2qdfat/without_downvoting_me_to_hell_can_someone_explain/cn5s41zIs there anything wrong with that, technically or economically?
(Summary being from the reddit attack described at the above link that a miner with > 50% DoS's transactions on bitcoin chain (ie does not accept any transactions just mines empty blocks on the main chain people are trying to use) until people either abandon bitcoin or capitulate and adopt the rules the DoS miner is threatening.)
Yeah that is kind of logical, there are some caveats. The miner has a lot of capital invested maybe > $300m to do that now. (You can be sure most pool miners would abandon a pool that did that.) This is going to be disruptive to bitcoin confidence and price and that affects the value of bitcoins they are mining, or the equipment itself which is worthless if bitcoin fails or crashes badly as a result. Dangerous game to play with a $300m good behaviour bond. Secondly while they are DoSing bitcoin, they are not mining coins nor on their proposed alternative chain at any kind of usable speed as they have almost no hashpower left (if they have 55% and they're using 53% to DoS bitcoin that gives them 500min blocks if the new chain has the same difficulty. If the new chain has reset difficulty the honest miners might DoS it in retaliation (block transactions there). Now if the attacker had maybe 70% they could dominate both chains reliably.
This is beyond mining a new hard-fork protocol version (which honest users and full nodes would ignore) and more DoS warfare to kill the main chain to give users a choice of no transactions or to fold and use the new chain. I would imagine users would be annoyed enough about that so as to be a scenario where the bit red button might get pushed - destroying the $300m capital of the attacker (and the $245m of the rest of the miners who probably are going to sue the attacker, and its not easy to hide the delivery and location of $300m worth of mining equipment drawing perhaps a GW of power.)
the nuclear big-red-button [...] MAD argument that keeps miners somewhat sensible as if that button is pushed they are sitting on $500m of scrap electronics
[...] Putting it more simply: an entity that can jam some process for sufficiently long time can force the people who depend on that process to accept changes to it, as long as the changes are not as bad as the jamming itself.
Yep its a unfortunate. This is why decentralisation of mining is important.
Outside of spam limits which could be protocol enforced, its caveat emptor, you shouldnt put money into a chain unless there is some assurance that security & bitcoin protocol knowledgeable people have audited it.
That is my understanding, and that is I why I cannot see any technical content in the sidechains proposal. Unless it specifies some things that every sidechain must do (or must not do), with a mechanism to enforce those constraints, it will not bring any new tools or ideas to cryptocurrency technology. So far it is only a nomenclature proposal: "let's use the word 'sidechain' for any project or entity that could 'own' some bitcoins for some time".
I dont think you cant really technically enforce freedom from security issues or freedom from bad economic parameter choices kind of stuff, thats probably AI / halting-problem type difficulty which we dont know how to do. You could certainly make some best practices statement about how security should be achieved, and acceptable parameters and configurations for user safety, and have someone competent audit that and certify their audit.
Why wouldn't Bitstamp be already a sidechain, for example?
Well bitstamp isnt trying to algorithmically peg - they're saying trust our host security, cold wallet physical security, audit, governance / separation of duty etc. Ie you are trusting humans to manage an IOU. (Not saying bitstamp is a bad exchange).
Adam