> Are they ignoring the risk of a double spend and just going ahead once a transaction has propagated to the network for mining?
After a majority of the nodes have seen a transaction it becomes relatively safe to accept it as valid before 6 confirmations. You wouldn't want to do that for high value transactions, but I think it's fine up to a couple BTC (read: $1000 or less).
You can't determine what the majority of the nodes thinks in this way because of Sybil attacks. Otherwise we wouldn't need proof of work as a consensus mechanism.
Mycelium's "transaction confidence" feature is well intentioned but it is a disaster waiting to happen.
I agree with you, but I don't think it would be worth the effort to conduct this attack for such low value transactions. I'm certainly not an expert here, and clearly the blockchain is the only source of truth in the end.
After a majority of the nodes have seen a transaction it becomes relatively safe to accept it as valid before 6 confirmations. You wouldn't want to do that for high value transactions, but I think it's fine up to a couple BTC (read: $1000 or less).
The Mycelium wallet implements transaction confidence: http://www.reddit.com/r/Bitcoin/comments/1xqh51/new_mycelium... and http://www.cryptocoinsnews.com/2014/02/13/the-mathematically...