I'm mixed about this. I know a lot of 3rd party devs have been pinning their hopes on SMTs for their apps. Okay, as some kind of share-holding-security token I can see some use case there. But then, BUILDTEAM has been running their token (with simulated stacking) on the BitShares chain for about a year now. Their system work - and you can see staked holdings by looking at a spreadsheet (cue: that Forbes article on blockchain being no smarter than a spreadsheet heh). <- This ramble basically means that if you want a security token then, where there's a will, there's a way. You don't need an SMT but it's certainly nice to have it baked into the blockchain.
My thoughts on SMTs as a utility token are mixed; I still think that there are only a few communities/dapps that have large enough communities with underlying value exchanges that warrant a special SMT. As a matter of economic design there's too many people pinning their hopes on it.
Onto the DEX. I guess you could create some cross-chain-swap-thing to BitShares to make a DEX work.
Though I thought the idea of SMTs was to do away with an order-book driven DEX in favour of an algorithmic DEX (ala Bancor). If that's the case then I think the features for this need to be baked into a very early release of SMTs at the blockchain level even if realising them into the UI takes some time. The early implementations don't even need to do SMT-SMT exchanges, they could simply do SMT-STEEM and STEEM-SMT and then let the UIs figure it out.
I'm still a bit skeptical about SMTs. Not super skeptical, but yeah. I think SteemInc needs to deliver on them to show that they are a trustworthy company for sure.
They could also do well to help promote third-party efforts at improving the general infrastructure. Steem appears to me to be a bohemoth of a thing that really could be made more reliable and more nimble if effort was put in that direction.