Cool at least we're not grudging, you're welcome, let me explain how its going to work.
The situation you describe is true in the case where the validator is a single entity hosted on SafeNet, even if in order to get the private keys, one must basically break safenet.
The difference now (and this is a disclosed information) will be run simultaneously from different wallets, each wallet will have its own validator Instance, its job is to verify the orders and the transaction informations. During a trade, if someone manages to get access to its private key, he will be the unique victime of his cheating attempt, as the other instances won't accept his order.
It doesn't means that NVO will allow 0 confirmation bitcoin transactions, at this stage it would be hypotethical, a minimum of 1 block confirmation is required, however, LN integration is a must, as it solves a lot of issues experienced by the DEXs, Segwit will be helpfull too.
Regarding the possibility to send a transaction with low fees, the different validators will check this information too as the minimum will be determined by the validator depending on the network state.
I know about Bitbay and BitHalo, and i understand that you defend your ideas about atomic trading
and i appreciate the exchange of idea.
Right, well I'm glad you re-read my response I wasn't attacking you. The problem is, it is not decentralized exchange. Linking 2 of 2 multisig is almost the same as 2 of 3 but slightly more secure.
The validator can SPOOF the buyer or seller and default on their obligation after getting a valid signature.
So the example:
Bob wants to send LTC to Alice for BTC
Bob forms 2 of 2 with unique validator, Alice forms 2 of 2 with unique validator (potentially the same one or not doesn't matter the attack is the same either way).
Bob signs to send LTC under the assumption that Alice will be beholden to being honest by the validator.
The problem? The validator IS ALICE (and potentially the 2nd signature in Bobs tx too).
Bob sends the LTC, Alice doesn't bother to send her BTC in fact she withdraws it and takes Bobs LTC.
This is a pretty serious attack and thus this cannot possibly be decentralized exchange.
Only a lightning network and AT are decentralized exchange (and potentially microtrading in a 2 of 2 double deposit escrow).
Again, I'm not trying to rattle your cage or "insult you". I definitely want more people doing projects that are more secure than modern exchanges. But what you propose in safe network is a series of green addresses 2 of 2 accounts. It doesn't eliminate the risk of theft at all. Thats because the validator could collude with buyer or seller. Whomever runs those servers has a lot of power to spoof deals.
The validator works as instances, is any instance is broken, the user won'tbe able to realise operations on the exchange.
Bob wants to send LTC to Alice for BTC
Bob forms 2 of 2 with unique validator, Alice forms 2 of 2 with unique validator
(potentially the same one or not doesn't matter the attack is the same either way).
Bob signs to send LTC under the assumption that Alice will be beholden to being honest by the validator.
The problem? The validator IS
NOT ALICE
(and potentially the 2nd signature in Bobs tx too).
Bob sends the LTC, Alice doesn't bother to send her BTC in fact she withdraws it and takes Bobs LTC.The role of the validators is to check for the validity of informations, the instances won't trust each other and will cancel any order at the first sign of manipulation untill both parties release the funds.
