I would hope that any sidechains to bitcoin that use proof of work would have a shared mining model, so that miners could mine both simultaneously.
That's the problem: merge-mining means there's zero cost to attacking a sidechain. Of course, people like to invoke economic incentives here and say there's opportunity cost with the idea that the sidechain will pay ongoing fees or something. But this is a very untested idea to put it charitably - keeping Bitcoin itself secure with only fees is widely considered to be an unsolved problem, and that problem doesn't have a reward in the form of a bunch of 2-way-pegged Bitcoins waiting to be stolen.
Ultimately I think media coverage of this is really misunderstanding how sidechains are very far from a proven idea, except in the centralized model where you trust someone to maintain the sidechain. Equally it's pretty clear that they have a high risk of making the incentives behind Bitcoin mining even further tilted towards centralization of hashing power.
I feel that the other side: zero cost to protect a sidechain, is a much bigger benefit. Presumably, the sidechain has mining rewards as well, which devalue the amount of BTC you get when going back to the main chain.
Re:A Bitcoin battle is brewing
by davidgerard on Thursday December 25, 2014 @01:51PM
(#8796870)
Altcoins also have one important advantage over sidechains: they exist.
It's far from clear that will in fact happen - mined sidechains that aren't widely adopted by miners are insecure, just like small altcoins are vulnerable to 51% attacks. Even worse the 2-way-peg mechanism Blockstream proposes trusts hashing power to determine where coins go so an attacker not only destroys the sidechain, but has a direct financial incentive to do so because destroying the sidechain lets them steal all the money that has been moved to it. Equally even if a majority of miners support a given sidechain, they can always change their minds later and take the funds. (quite possibly due to a temporary problem, like a hack or a a court order)
To solve this problem Blockstream explicitly proposes that smaller or experimental sidechains without majority support use a "federated" model where one or more trusted authorities controls the sidechain. Of course, needless to say needing a trusted third party to keep the sidechain secure isn't a good solution in many cases - they're essentially acting as a bank and it's extremely difficult to design systems where that bank can't profitably steal all the funds.