It's only possible to assess the current situation. It surely may improve in the future, but it's important for users to know what they are getting into right now.
And of course development on top of an existing protocol is faster than making changes inside the protocol.
Even if they address all the issues and create a real decentralized product, exchanges still will have the overhead of replaying not only the steem blockchain, but also filtering out the steem engine related transactions, which will take considerably more time on itself.
It would have been way more efficient for everyone to develop the functionality in the core. But of course that would have required qualifications few developers have, and wouldn't have allowed for easy monetization of a half finished product.
They can do this, and it currently takes about three days if they already run
steemd. There are improvements to speed that up. Or, they can bypass this option and just replay the sidechain from a block log which takes less than an hour.This is a viable, safe option because checkpoints can be verified, in theory, although a little work is still needed right now to ensure this option is cryptography verifiable. But we’re talking about the hypothetical scenario where they’ve solved all of the decentralization issues, so checkpoints will be addressed too.
Thanks for those insights! Definitely interesting to see where everything is heading.