Perhaps cypher's concern is that everyone moves their BTC over to the sidechain, then the sidechain fails in such a way that it's not possible to reconvert them to BTC, and the whole Bitcoin ledger is destroyed.

To fully comprehend whether this is a concern requires some terminological cleanup. I think there's a basic misguided focus in the term "sidechain." It's not the chain but the peg that matters.

Imagine if Counterparty had delayed their launch a year. Instead of proof of burn where provably destroying 1 BTC gave you 1200 XCP, they set up a 2-way peg where locking 1 BTC gives you access to 1200 XCP (and the reverse: locking 1200 XCP gives you access to 1 BTC) in perpetuity. Why does that make XCP a "sidechain," whereas proof of burn makes it an "altcoin"? The key thing that's happening is some assets are being pegged to bitcoins. Insofar as the market was confident that you could always very easily convert between the two, forever, at 1 BTC = 1200 XCP, the price of 1200 XCP should be very close to 1 BTC. If someone sent you 1200 XCP, after all, you could easily convert them to 1 BTC and sell them for the same amount of dollars as if you had held 1 BTC from the start (and vice versa). So in theory you should be agnostic about which form you get paid in, and wallet software may not even show the end user which currency they are technically holding, i.e., whether they have 2 BTC or 2400 XCP, since it can be converted any time depending on the user's needs.

Now let's suppose everyone but you converted their BTC to XCP. You are the last BTC holder. Assuming convertibility remains, miners would still mine BTC for the block reward - since they could simply convert to XCP and sell for fiat. And you could still convert at any time. Theoretically your investment would never be at risk and you would only stand to gain if XCP were better since the value of XCP - and hence to the same degree BTC - would rise. If the peg lasts, the value of you coins can only grow if XCP is all around better.

One might even conceive that it's possible to do this without a protocol change, just using timelocks and one chain as the other's oracle. In the example, a kind of smart contract in Counterparty that issues you 1200 XCP when the oracle (trustless data directly from the Bitcoin blockchain) says that 1 BTC has been locked in a certain way, and similar smart contract in Bitcoin being set up simultaneously that unlocks the 1 BTC when the oracle (data directly from Counterparty's system) says that 1200 XCP has been locked in a certain way.

If it were done that way, it would look a lot more benign, but it may be the same in effect. Or it may not. But it's important to characterize the proponents' side correctly first in order to properly argue about any possible dangerous edge cases and have those arguments be understood by the proponents.