It is impossible that USDT is not corrupt. One can surmise that the likely way they got exchanges to use it, is by sharing the fractional reserve pie with the exchanges.

We really need decentralized exchanges (DEX), but problem is that liquidity is created by those playing with margin and traders _need_ to be where the most liquidity is. So we'd need to build loans, interest, and margin into the DEX the way Poloniex did. The problem with Bitshares' DEX is the user had to convert their native tokens to pegged BitUSD, BitBTC, etc.. so its not the same and traders want the "real tokens" (although exchanges create fractional reserves thus they really aren't real tokens but traders _need_ liquidity). But there is no way to do real-time trading between different blockchains. I studied TierNolan's cross-chain trading, and it is too slow, can be jammed with spam, etc.. Side-chains and merged mining (e.g. Blockstream and Cosmos) are fundamentally flawed. However, trading within the same chain in real-time (on DPoS' Graphene or BitNet's blockchain design) is possible.

However, there are two fundamental barriers to emulating Poloniex's margin trading on a DEX.

1. Must have another token to borrow to go long, and to sell to (or at least a price feed for settlement) for long and short margin trading respectively. A blockchain could have two tokens, but that would really be the same as different assets. Bitshares' tried to solve this problem by pegging external assets but all pegs are unstable and eventually fail. BitUSD and SBD have only survived thus far because of being small closed ecosystems (so more likely the entire projects fail before the pegged assets do).

2. Settlement ordering of trades (and also w.r.t. to who is providing price feeds if using those as documented as problemmatic in Steem's whitepaper) is possible to game by the witness, miners, or whom ever is ordering the transactions in blocks. Because settlement order moves prices, etc..

3. Too much leverage in the system can cause margins to be exceeded and those who loaned tokens can lose some tokens. This is analogous to debt defaults due to overextension of leverage in the fiat world.

So this is why DEX is a waste of time. Liquidity is king in the fungible money world. Fungible money is a very corrupting paradigm. We really need to end it and move forward to the next stage of the development of economics with the Inverse Commons.

Without short-selling, the OTC markets can go no bid:

http://www.investopedia.com/terms/o/over-the-countermarket.asp

While OTC markets function well during normal times, their lack of transparency can cause a vicious circle to develop during times of financial stress, as was the case during the 2007-08 global credit crisis. Mortgage-backed securities and other derivatives such as CDOs and CMOs, which were traded solely in the OTC markets, could not be priced reliably as liquidity totally dried up in the absence of buyers. This resulted in an increasing number of dealers withdrawing from their market-making functions, exacerbating the liquidity problem and causing a worldwide credit crunch.

OTC markets will have huge spreads at times.

People only use OTC markets when they have no other choice, such as to avoid regulations or the only place where certain assets are traded. Because OTC markets have inferior liquidity, spreads, and pricing information (transparency). Thus OTC markets will only be popular when normal exchanges are too painful. If ever exchanges become too regulated to the point where a large number of people want OTC markets, then a DEX will be viable. But not until then.

Leveraged trading is impossible in DEX (will always fail eventually). Thus it is interesting to note that OTC is plausible only in a scenario where centralized exchanges are no longer viable, i.e. seems to correlate well with the slow death of fungible money over the coming decades. So perhaps DEX will become important in coming years.

With the CLTV opcode added by SegWit, I think it is possible to make a truly trustless and decentralized DEX wherein a Bitnet token holder can post an ask price in any coin that supports CLTV, then anyone can make a bid entirely decentralized. The first bid taken by the Bitnet blockchain (as it detects the bids make on the blockchain with CLTV) gets locked in and the rest of the bids get refunded. This is the only way to do it that won't be jammed attacked (which was my original complaint against TierNolan's flawed protocol). So looks like Litecoin will get CLTV now.

The CLTV will pay refund to bidder after a sufficient delay, else it will pay asker's address with a HTLC. Bitnet confirms a transaction paying bidder's address with same HTLC. Bidder releases HTLC solution and both parties can spend the traded tokens. This was not possible with nLockTime. It requires OP_CLTV. Note that this proposed protocol can still be jammed if bidder doesn't release the HTLC solution, takes refund, and repeats the attack over and over. Yet my protocol has the advantage over TierNolan's that it can try all the bids in order of their appearance on the blockchain, so that eventually a good bid is reached; whereas, TierNolan's protocol was indefinitely jammable (a race condition). Also my proposal requires on very little delay to wait on the bidder to release the HTLC solution, thus can cycle through the bids much faster (given Bitnet can confirm in milliseconds instead of minutes). Also the bidder's funds are locked up for some time before they are refunded so this isn't cost free (time opportunity cost) for the attacker. Also bids in the same block could be prioritized by the ones with the longest refund time, so that time opportunity cost can be ramped up as needed to squelch attacks. This is clever solution to TierNolan's flawed concept!

Note have bidders that want to enter a DEX trade sign with a stable public key (not necessarily the same as their spending key) so as to establish a reputation of those who don't jam the trades, is not a solution because it prevents anonymity and it means that newbies are on par with attackers. This was what I complained to @TierNolan and @jl777 about at BCT. My proposed protocol above is the only solution I've found.

https://github.com/bitcoin/bips/blob/master/bip-0065.mediawiki#escrow
https://en.bitcoin.it/wiki/Atomic_cross-chain_trading#Algorithm
https://en.bitcoin.it/wiki/Hashed_Timelock_Contracts

Bitsquare is not trustless and requires reputation, security deposits (all of which can fail and are not perfectly reliable):

https://bitsquare.io/faq/#8